International Journal of Molecular Sciences Article On the Computational Design of Azobenzene-Based Multi-State Photoswitches Miquel Moreno 1, * , José M. Lluch 1,2 and Ricard Gelabert 1 1 Departament de Química, Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain 2 Institut de Biotecnologia i de Biomedicina (IBB), Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain * Correspondence: miquel.moreno@uab.cat; Tel.: +34-935-812-174 Abstract: In order to theoretically design multi-state photoswitches with specific properties, an exhaustive computational study is first carried out for an azobenzene dimer that has been recently synthesized and experimentally studied. This study allows for a full comprehension of the factors that govern the photoactivated isomerization processes of these molecules so to provide a concep- tual/computational protocol that can be applied to generic multi-state photoswitches. From this knowledge a new dimer with a similar chemical design is designed and also fully characterized. Our theoretical calculations predict that the new dimer proposed is one step further in the quest for a double photoswitch, where the four metastable isomers could be selectively interconverted through the use of different irradiation sequences. Keywords: molecular photoswitches; azobenzene dimers; theoretical photochemistry; absorption spectra 1. Introduction Molecular photoswitches are nowadays at the heart of a large number of very hot Citation: Moreno, M.; Lluch, J.M.; topics in research within a broad range of sciences, ranging from nanotechnology to phar- Gelabert, R. On the Computational macology all the way through catalysis and biochemistry [1–9]. The most frequently used Design of Azobenzene-Based photoswitches are azobenzenes (ABs) as they are easy to synthesize, are highly reversible Multi-State Photoswitches. Int. J. Mol. Sci. 2022, 23, 8690. https://doi.org/ and photostable and have quite different molecular structures between the photoconvertible 10.3390/ijms23158690 cis and trans isomers [2,5,10]. Moreover, the photochemical properties of ABs can be largely modulated by the introduction of different substituents on the benzene rings [11–13]. Academic Editor: Mihai V. Putz The traditionally used molecular photoswitches involve a binary (on/off) pair corre- Received: 11 July 2022 sponding to the two photoconvertible isomers (usually cis/trans or Z/E). Recently, interest Accepted: 2 August 2022 has turned to the design of multi-photochromic compounds that contain two or more Published: 4 August 2022 switching centers usually linked by covalent bonds [14–16]. In the simpler case of two such centers, four states can be interchanged by means of light (or another external stimulus). Publisher’s Note: MDPI stays neutral This allows for more complex photochemical processes including the possibility to design with regard to jurisdictional claims in molecular motors that perform unidirectional cycles [17–19]. Not surprisingly, ABs are the published maps and institutional affil- most commonly used units in the design of such multi-photochromic systems [20–30]. iations. The first and simplest double photoswitch analyzed was a bis(azo) compound where two azobenzene units shared a central phenyl ring. Such a conjugated system does not allow for selectively addressing each AB unit, but the introduction of different substituents Copyright: © 2022 by the authors. at each phenyl group leads to four potentially stable isomeric forms [20–23]. In order Licensee MDPI, Basel, Switzerland. to have two (or more) AB units that could be “orthogonally” photoexcited (i.e., where This article is an open access article each AB unit could be independently and selectively activated), a more complex design is distributed under the terms and necessary. One strategy towards this goal involves incorporating AB units in macrocyclic conditions of the Creative Commons arrangements, resulting in azobenzophanes [24,25]. More recently, interest has focused Attribution (CC BY) license (https:// on the design of star-like triple azobenzene units linked to a central core, as the resulting creativecommons.org/licenses/by/ tripodal design affords a wide range of applications [26–29]. Another strategy involves 4.0/). Int. J. Mol. Sci. 2022, 23, 8690. https://doi.org/10.3390/ijms23158690 https://www.mdpi.com/journal/ijms is necessary. One strategy towards this goal involves incorporating AB units in macrocy- clic arrangements, resulting in azobenzophanes [24,25]. More recently, interest has fo- Int. J. Mol. Sci. 2022, 23, 8690 cused on the design of star-like triple azobenzene units linked to a central core, as2the of 20 resulting tripodal design affords a wide range of applications [26–29]. Another strategy involves the synthesis of azobenzene dimers with bulk substituent groups so that both moieties are almost orthogonal due to steric repulsions [30–33]. the synthesis of azobenzene dimers with bulk substituent groups so that both moieties are In some of the cited references on multi-azobenzenes, theoretical calculations are also almost orthogonal due to steric repulsions [30–33]. performedIn some [20–24,26–29,33], though on of the cited references they are usually restricted multi-azobenzenes, to thecalculations theoretical ground electronic are also state. Slavov et al. [23] reproduce the absorption spectra performed [20–24,26–29,33], though they are usually restricted to the ground of unsubstituted bisazobenzenes electronic with state.a Slavov common phenyl et al. ring, obtaining [23] reproduce good accordance the absorption spectra ofwith the experimental unsubstituted results. bisazobenzenes The photoinduced E  Z isomerization of these azobenzenes with a common phenyl ring, obtaining good accordance with the experimental results. is also studied by Floss and Saalfrank through a Esurface-hopping The photoinduced → Z isomerization molecular of thesedynamics azobenzenes study is [22] also demonstrating studied by Floss that, and depending on the connectivity pattern of the two azo groups, Saalfrank through a surface-hopping molecular dynamics study [22] demonstrating that, they may act as uncoupled units or else present depending a cooperative on the connectivity behavior. pattern of the Bléger two azoet al. [33] carry groups, theyoutmayextensive ab initio act as uncoupled calculations devoted to obtaining the absorption spectrum and perform units or else present a cooperative behavior. Bléger et al. [33] carry out extensive ab initio a quite compre- hensive theoretical calculations devoted analysis of the potential to obtaining energyspectrum the absorption surface ofand the perform ground electronic a quite compre-state ofhensive the unsubstituted bisazobenzene theoretical analysis with theenergy of the potential two moieties surface uncoupled. of the ground Onelectronic the theoretical state of side, the most studied the unsubstituted multiazobenzene bisazobenzene with the systems are theuncoupled. two moieties systems with three On the azobenzene theoretical side, units linked to a central core [26–28]. The pure theoretical work the most studied multiazobenzene systems are the systems with three azobenzene units by Yang et al. [27] points tolinked a decoupling to a central of the three core azo groups [26–28]. The pure so that introducing theoretical workdifferent by Yang substituents at each et al. [27] points to a AB branch should lead to an effective triple independent photoswitch. decoupling of the three azo groups so that introducing different substituents at each AB Given branch that previous should lead to antheoretical calculations effective triple independenthave been restricted to investigating spe- photoswitch. cific systems Given that already synthesized previous and experimentally theoretical calculations have studied, beenin this worktowe restricted propose a investigating comprehensive specific systems analysis alreadyofsynthesized the factors that and govern the photochemistry experimentally studied, in this of a work generic weazo- pro- benzene-based pose a comprehensive multi-state photoswitch. analysis As a first of the factors thatbut relevant govern step towards this the photochemistry ofgoal, we a generic will consider azobenzene azobenzene-based dimers multi-state with different photoswitch. Assubstituents so thatstep a first but relevant fourtowards E/Z isomers are this goal, potentially stable.azobenzene we will consider The ideal goal would dimers with bedifferent to find systems substituentswhere all the so that interconversions four E/Z isomers are potentially (up to six instable. a dimer) Thecould ideal goal would bebytodifferent be triggered find systems where of sequences allirradiation, the interconversions so each (up tocould isomer six inbe a dimer) selectivelycould be triggered by different sequences of irradiation, so each addressed. isomer could be selectively addressed. 2. Results and Discussion 2. Results and Discussion The first step will be to assess the reliability of our methodology, comparing with knownThe first step will experimental be For data. to assess this we thetake reliability of our methodology, the bisazobenzene synthesized comparing and experi-with known experimental data. For this we take the bisazobenzene mentally analyzed by Zhao et al. [30] as a model to analyze the factors that govern the synthesized and experi- mentally analyzed photochemistry of suchby Zhao et al.Scheme a dimers. [30] as1ashowsmodelthe to molecular analyze the factors that structure of thisgovern system, the photochemistry of such a dimers. Scheme 1 shows the molecular labelled dimer 1 from now on. It consists of an unsubstituted azobenzene covalently structure of this system, labelled linked to dimer a full 1ortho-fluorated from now on. Itone. consists The twoof anmethyl unsubstituted groups azobenzene at one end covalently of the benzeniclinked to a full ortho-fluorated one. The two methyl ring are introduced to keep the two units in quasi-orthogonal planes. groups at one end of the benzenic ring are introduced to keep the two units in quasi-orthogonal planes. Molecularstructure Scheme1.1.Molecular Scheme structureofofdimer dimer1 1ininitsitsmost moststable stable(EE) (EE)form. form. The four E/Z isomers expected to be stable for this system will be labeled EE, EZF , ZEF and ZZ. A subindex F is used when the E/Z state of the azobenzene unit (first letter) and the fluoroazobenzene part (second letter) are different. Table 1 presents the energies of the lowest excited electronic states for the four stable E/Z isomers of dimer 1. The excited states are calculated at the geometry of the minima in Int. J. Mol. Sci. 2022, 23, 8690 3 of 20 the ground state (vertical excitations) so that the calculated excitation wavelengths should approximately match the peaks of the bands in the experimental absorption spectra. Only excited states with an absorption wavelength above 250 nm are considered in Table 1 as this is the approximate lower limit of the experimentally obtained spectra. f values report the oscillator strength of the transition from the ground electronic state. A value below 0.05 usually means that the S0 → Sn transition is forbidden, so it should be barely visible in the actual spectra. The last column in Table 1 indicates the main contributions to the excitation. Capital H and L letters stand for HOMO and LUMO, respectively. Table 1. Relevant lowest electronic excited states of the four E/Z stable isomers of dimer 1. Isomer State E (eV) λ (nm) f Excitation S1 2.67 464 0.10 H−1 → L; H−7 → L (F) S2 2.75 451 0.00 H−2 → L+1 (H) EE S3 3.74 331 1.61 H → L+1 (H) S4 4.10 302 0.78 H−1 → L; H−7 → L (F) S1 2.75 451 0.00 H−2 → L+1 (H) S2 2.89 430 0.06 H−1 → L (F) EZF S3 3.75 330 1.29 H → L+1 (H) S4 4.32 287 0.08 H−3 → L+1 (F → H) S1 2.67 465 0.09 H−1 → L; H−7 → L (F) S2 2.78 445 0.05 H → L+1 (F) S3 4.06 305 1.33 H−1 → L; H−7 → L (F) ZEF S4 4.47 277 0.07 H−4 → L; H−5 → L (H → F) S5 4.47 277 0.03 H−4 → L; H−5 → L (H → F) S6 4.48 276 0.16 H−2 → L+1 (H) S1 2.78 445 0.05 H → L+1 (H) S2 2.88 430 0.05 H−1 → L (F) ZZ S3 4.46 278 0.36 H−2 → L+1; H → L+1 (H) S4 4.50 275 0.02 H−8 → L (F) S5 4.57 271 0.12 H−4 → L (H → F) The first use of the numerical results presented in Table 1 is that they allow us to compute the absorption spectra of the four stable isomeric forms of dimer 1, as shown in Figure 1. The spectra, obtained using the GaussView program [34], are convoluted with a Lorentzian function with full-width-at-half-maximum of 0.1 eV. At first sight, comparison of Figure 1 with the experimental absorption spectra as presented in Figure 2 of [30] reveals that our calculation is accurate enough. It is to be considered that bands depicted in Figure 1 below 250 nm are beyond the experimental range. Using the data reported in Table 1, a more detailed comparison of the theoretical and experimental spectra can be made to actually assess the feasibility of our computational strategy. • EE isomer: Experimentally this is the only stable isomer in the absence of irradiation. The main feature is a huge broad band that extends between 280 and 390 nm, with a maximum at approximately 340 nm. An additional, much tidier, but also broad band is seen at 400–500 nm. Our theoretical calculation reveals that the larger band includes both the transitions to S3 and S4 that have large f values 1.61 and 0.78 in our calculation, whereas the less intense band at lower energy will correspond to the transition to S1 calculated at 464 nm that has a small, but not negligible, f value of 0.10. The close transition to S2 calculated at 464 nm has an oscillator strength of 0.10, a small though not negligible value. Int. J. Mol. Sci. 2022, 23, 8690 4 of 20 • EZF isomer: Experimentally this isomer is the prevalent one (79%) when the original dimer is irradiated at >500 nm, so we can compare the results in Table 1 with the photostationary state (PSS) spectra after irradiation at this wavelength. In comparison with the EE case just discussed, there are very few differences in the position of the absorption bands. Intensities are different but this may be partially due to the fact that we are now analyzing a mix of different isomers. At any rate the calculations of the excited states of EZF isomer disclose again the presence of three allowed transitions with f > 0.1: The more intense one to S3 at 330 nm (f = 1.29) would make for the higher intensity experimental band with the little help of transition to S4 calculated at 287 nm with f = 0.08. The lower energy will correspond to the S2 transition predicted at 430 nm with a very low but non-zero f value of 0.06. • ZEF isomer: This structure is not obtained from direct irradiation so that now there is no experimental data to compare with. In any case the calculations predict again rather small differences with respect to the previous discussed cases: The most likely transition is to S3 , calculated at 305 nm with f = 1.33, while the lower energy transitions to S1 and S2 are predicted to peak at 465 and 445 nm, respectively, but have, as in the previous cases, quite small intensities (f < 0.1). • ZZ isomer: This species is dominant (74%) when the original dimer is photoexcited at 350 nm. Now, the obtained spectrum is clearly different from the previous cases just discussed: The larger band is now seen at noticeably shorter wavelengths that almost reach the end of the experimental available range (λ < 300 nm). The less intense band at lower energies is also slightly blue-shifted with a maximum slightly above the 400 nm mark. Our results are in quite nice agreement with these facts as the more intense band is the one to S3 calculated at 278 nm while the two lower transitions to S1 and S2 have very low f values (0.05) and are predicted to appear at 430 and 445 nm. Another noticeable fact of the ZZ absorption spectrum presented in [30] is the lower intensity of the high energy band as compared with the corresponding one of the other two obtained isomers. This is also in nice agreement with our theoretical results as the higher oscillator strength of the ZZ isomer just reaches the 0.36 value for the transition to S3 , well below the corresponding maxima of the EE (1.61) and EZF (1.29) structures, as seen in Table 1. A different question not directly related to the electronic absorption spectra but crucial in the design of dimer 1 is the orthogonality of the electronic excitations. That is: dimer 1 was synthesized in order to be able to selectively excite each one of the two azobenzene units. This is the reason for the presence of the two ortho-methyl groups, as in one of the benzenic rings: so that the π systems of both moieties remain in almost perpendicular planes. Our calculation reveals that, concerning the geometries, this goal is accomplished (See Figure 2, top left). To address this question more systematically, in the last column of Table 1, a letter H or F is used to indicate whether the electronic excitation is located in the parent azobenzene unit (H) or in the ortho-fluorinated one (F). As the two AB moieties are almost orthogonal in the most stable EE form, we note that all the electronic excitations of the EE isomer are fully located in one AB unit and so can be safely labeled as H or F. This separation is also clearly seen in Figure 2 where the orbitals that are involved in the different excited electronic states described in Table 1 are drawn. The shape of the molecular orbitals depicted in Figure 2 confirms that the two lowest excited electronic states, S1 and S2 , come from excitations from orbitals that might be labeled as n (as they are not π) to a clear π*-type orbital (the asterisk appears as the arriving orbital is formally empty in the ground electronic state), so that these transitions are formally forbidden (low value of f), whereas the next two states, S3 and S4 , come from excitations between π orbitals (and are formally allowed). The shape of the orbitals seen in Figure 2 also confirms the orthogonality of the different electronic excitations so that all the orbitals are located in one side of the dimer. For the other three isomers, the analysis of the molecular orbitals is more problematic as the planarity of the benzenic rings is partially lost in the Z isomers due to steric repulsions. At any rate the orthogonality of the excitations mostly remains, though Int. J. Mol. Sci. 2022, 23, 8690 5 of 20 some exceptions appear, and a few excited states show excitations delocalized between the two sides of the dimer. This comes from the well-known fact that the geometries of the Z isomers are not perfectly planar so that the spatial orthogonality of both moieties is strictly no longer present beyond the EE isomer. As these states imply charge transfer between both moieties (H → F or F → H), they have very small transition probabilities, so their impact in the photochemical behavior of these molecules will be negligible. An additional effect of the loss of planarity of the Z isomers is that the classification of the electronic excitations as nπ* or ππ* is not possible beyond the EE isomer. However, it is clear that for all the isomers two bands are clearly observed: A very intense one at high energies Int. J. Mol. Sci. 2022, 23, 8690 that loosely corresponds to the expected ππ* allowed transition in azobenzene and a4 muchof 21 less intense band that could be classified as nπ*. The shapes of all the molecular orbitals implied in the electronic excitations of the other isomers are depicted in Figures S1–S3. Figure Computed absorption Figure1.1. Computed absorptionspectra spectraofofdimer dimer 1 in 1 in acetonitrile acetonitrile at the at the optimized optimized ground-state ground-state ge- ometries of of geometries thethe four stable four isomers stable of dimer isomers 1. Vertical of dimer red lines 1. Vertical redindicate the position lines indicate of the vertical the position of the transitions vertical while the transitions black while thecontour corresponds black contour to the to corresponds convolution with awith the convolution Lorentzian function a Lorentzian de- function picted inin depicted order ordertoto theoretically theoreticallyreproduce reproduce thethe experimental experimental absorption absorptionspectrum. spectrum. • EE isomer: Experimentally this is the only stable isomer in the absence of irradiation. The main feature is a huge broad band that extends between 280 and 390 nm, with a maximum at approximately 340 nm. An additional, much tidier, but also broad band is seen at 400–500 nm. Our theoretical calculation reveals that the larger band in- cludes both the transitions to S3 and S4 that have large f values 1.61 and 0.78 in our calculation, whereas the less intense band at lower energy will correspond to the transition to S1 calculated at 464 nm that has a small, but not negligible, f value of 0.10. The close transition to S2 calculated at 464 nm has an oscillator strength of 0.10, a small though not negligible value. • EZF isomer: Experimentally this isomer is the prevalent one (79%) when the original clear that for all the isomers two bands are clearly observed: A very intense one at high energies that loosely corresponds to the expected ππ* allowed transition in azobenzene and a much less intense band that could be classified as nπ*. The shapes of all the molec- ular orbitals implied in the electronic excitations of the other isomers are depicted in Fig- Int. J. Mol. Sci. 2022, 23, 8690 ures S1–S3. 6 of 20 HOMO-7 HOMO HOMO-1 LUMO LUMO + 1 Figure 2. Shape of the molecular orbitals involved in the relevant excited electronic states of the EE isomer isomer of of dimer dimer 1. 1. The The drawings drawings correspond correspond to to an an iso-level iso-level of of 0.02 0.02 a.u. a.u. of of electronic electronic density. density. The The left left top top corner corner of of the the figure depicts the figure depicts the molecular molecular structure structure of of the the EE EE isomer. isomer. The remarkably remarkably good goodagreement agreementofofthe theexperimental experimentalabsorbance absorbance electronic electronic spectra of spectra thethe of E/Z E/Zisomers isomersofofdimer dimer11with withour ourcalculations calculationsgives givesus usconfidence confidence to to tackle tackle the much more complex complexstudy studyof the actual of the photochemistry actual photochemistryof azobenzene multi-state of azobenzene photoswitches multi-state pho- when irradiated toswitches by light ofbydifferent when irradiated light of wavelengths. To approach different wavelengths. this goal, To approach thisadditional goal, ad- calculations ditional have been calculations haveperformed for dimer been performed for1.dimer First 1. of First all, the of transitions states for all, the transitions the states different isomerization processes in the ground state have been exactly for the different isomerization processes in the ground state have been exactly located. A located. A second type of calculation, second relevantrelevant type of calculation, to studyto thestudy photochemistry of dimer of the photochemistry 1, is the relaxation dimer of the 1, is the relaxa- geometries in the excited electronic states. tion of the geometries in the excited electronic states. Figure 33 contains Figure contains all all relevant relevant data data computed computed so so far far so so that that the the discussion discussion of of what what happens upon happens upon excitation excitation of of the the more more stable stable EE EE isomer isomer (the (the only only one one present present previous previous to to any light irradiation) with light of different wavelengths can be undertaken. any light irradiation) with light of different wavelengths can be undertaken. Again, the Again, the goal is goal is to to understand understandthe thereason reasonbehind behindthe theobserved observed experimental experimental facts so that facts we can so that use we can our theoretical results to design brand new azobenzene-based multiple use our theoretical results to design brand new azobenzene-based multiple photoswitches photoswitches with improved with improvedproperties. properties. Figure 3 is quite complex as it includes a large number of relevant data, so prior to any discussion, it is important to explain what is to be seen (and what is not to be seen!) in the figure. Horizontal lines indicate different states of energy (not necessarily energy minima as it is explained below). The energy of each structure is indicated in kcal/mol. Energies are relative to the most stable structure: The EE isomer in the ground electronic state. Energies of the four optimized isomers in the ground state are depicted in black. The transition state for the interconversion between two of these isomers is also depicted in black and connected through sloped straight lines to the corresponding minima. As the ZZ isomer can be obtained either from the EZF or the ZEF forms, it appears twice at both the right and left edges of the figure, though it is obviously the same structure. Above each E/Z isomer the energies of the vertical transitions with low energy (basically the ones already analyzed in Table 1) and non-zero oscillator strengths are indicated by blue horizontal lines. For the EZF isomer, the transition to the lowest excited electronic state S1 is forbidden (f = 0.00), but this state is nevertheless shown as its presence may be relevant in the deactivation Int. J. Mol. Sci. 2022, 23, 8690 7 of 20 processes following irradiation. This fact is graphically displayed in Figure 3 by the use of a dashed blue line and a dashed green arrow indicating the impossibility of a direct excitation to this state. Each excited state has an energy value and a parenthesized letter (H or F) to indicate in which side of the dimer the electronic excitation is localized. The Int. J. Mol. Sci. 2022, 23, 8690 wavelengths of the vertical transitions are also indicated by green arrows. Finally, the lines 7 of 21 and numbers in red indicate the approximate location of the conical intersections between the ground and the lowest singlet excited electronic states. Figure3. Figure Fullenergy 3.Full energyscheme schemeof ofthe theground groundandandlowest lowestexcited excitedelectronic electronic states states of of dimer dimer 1. 1. See Seetext text for details. Relative for Relative energies energiesofofthe different the structures different (horizontal structures lines)lines) (horizontal are given in kcal/mol. are given Green in kcal/mol. Green arrowsarrows indicateindicate the wavelengths the wavelengths of the vertical of the vertical excitations excitations from eachfrom each conformer. conformer. Black Black lines lines indicate indicate the ground the ground electronicelectronic state whereas state whereas excited electronic excited electronic states states are are in in blue. blue. Red Red lines lines corre- correspond to spond to crossing crossing sections between sections between S and S S. 0 and S1. 0 1 Figure To gain3insight is quiteinto complex as it includes the mechanism of thea different large number of relevantrelaxed isomerizations, data, so prior scans to were any discussion, obtained by taking it is the important CNNCto explainangle dihedral what ofis each to beazobenzene seen (and what is not moiety as to thebereaction seen!) in the figure. coordinate and Horizontal optimizing lines the indicate different rest of the statesparameters geometrical of energy (not in thenecessarily energy ground electronic minima state. Foraseach it is geometry explained obtained below). The this energy of each structure way, a single-point is indicated calculation of the in kcal/mol. excited state Energies are relative was performed. Figureto 4the most stable displays these structure: scans for theThefour EE available isomer inisomerization the ground electronic reactions. The electronic state. Energies states of the upfourtooptimized S6 are followed isomersadiabatically in the ground so the lowest state energy corresponds are depicted in black. Theto the groundstate transition electronic for thestate S0 , the secondbetween interconversion to S1 and so of two on.these The states isomersare is specifically labeled also depicted in in Figure black and4.connected All the graphics through show a clear sloped decrease straight linesoftothe theenergy gap between corresponding S0 andAs minima. S1 that the becomes ZZ isomeralmost can bedegenerate obtained either whenfrom the two phenyl the EZ rings F or the ZEFare orthogonal forms, it appears(CNNC twicedihedral at both angle of 90 ◦ ). Conversely, S , that is quite close to S at both ends of the scan (close to the the right and left edges of the 2 figure, though it is obviously 1 the same structure. Above corresponding isomer minimum), maintains the energy each E/Z isomer the energies of the vertical transitions with low energy gap with S0 all (basically along the the scanones and this is also true for the higher excited states depicted in Figure 4. already analyzed in Table 1) and non-zero oscillator strengths are indicated by blue hori- zontal lines. For the EZF isomer, the transition to the lowest excited electronic state S1 is forbidden (f = 0.00), but this state is nevertheless shown as its presence may be relevant in the deactivation processes following irradiation. This fact is graphically displayed in Fig- ure 3 by the use of a dashed blue line and a dashed green arrow indicating the impossi- bility of a direct excitation to this state. Each excited state has an energy value and a pa- renthesized letter (H or F) to indicate in which side of the dimer the electronic excitation is localized. The wavelengths of the vertical transitions are also indicated by green arrows. Finally, the lines and numbers in red indicate the approximate location of the conical in- tersections between the ground and the lowest singlet excited electronic states. To gain insight into the mechanism of the different isomerizations, relaxed scans lowest energy corresponds to the ground electronic state S0, the second to S1 and so on. The states are specifically labeled in Figure 4. All the graphics show a clear decrease of the energy gap between S0 and S1 that becomes almost degenerate when the two phenyl rings are orthogonal (CNNC dihedral angle of 90°). Conversely, S2, that is quite close to S1 at both ends of the scan (close to the corresponding isomer minimum), maintains the energy Int. J. Mol. Sci. 2022, 23, 8690 gap with S0 all along the scan and this is also true for the higher excited states depicted in 8 of 20 Figure 4. Figure 4. Relaxed scans of the potential energy surfaces of the ground electronic states along all the Figure 4. Relaxed isomerization scans of pathways theCNNC (the potential energy surfaces dihedral angle of ofeach the ground electronic azobenzene states unit). Atalong eachall the of the point isomerization pathways (the CNNC dihedral angle of each azobenzene unit). At each point scan the energies of the lowest six excited electronic states are computed. The lines are drawn of the with scan the energies of the lowest six excited electronic states are computed. The lines are drawn with different colors and adiabatically follow the energies of each electronic state so that they never cross. different colors and adiabatically follow the energies of each electronic state so that they never cross. The quasi degeneracy of S and S1 right The quasi degeneracy of S00and S1 right in the middle of the reaction coordinate is a in the middle of the reaction coordinate is a clear indication of the presence of conical intersections clear indication of the presence of conical intersections (CIs).(CIs). CIseffective CIs are are effective funnels funnels that that promote promote thethe radiationless radiationlessdeactivation deactivation of the excited of the excitedstate statetotoeither either side side of of thethe isomerization isomeriza- path and tion so and path theirsopresence is relevant their presence in order is relevant to grasp in order the photochemistry to grasp the photochemistryof any molecule. of any Approximate geometriesgeometries molecule. Approximate of the CIs offorthe dimer 1 were CIs for dimerobtained through through 1 were obtained full optimization full of the first excited electronic state near the crossing point. This usually led to points with almost null excitation energies where the optimization fails or enters a loop. Given the multiconfigurational character of these crossing regions, our theoretical procedure is not able to fully characterize these points, so the CI energies, depicted in red in Figure 4, and the geometries, shown on Figure S4, have to be taken with caution. For the sake of our discussion on the global mechanism of the multi-state photoswitches, the presence of such low-lying CIs is relevant as an indication of the presence of internal conversion, but a more precise and detailed analysis is not necessary. In the experimental work of dimer 1, three different wavelengths (350 nm, 410 nm and >500 nm) are used to irradiate the EE isomer. This triggers some interconversions, but some remain absent [30]. In the following, we analyze the photochemical paths open at each irradiation in light of our theoretical computations. In the following discussion it is assumed that, as virtually all accessible excited states show an excitation localized in just one side of the dimer, it is that part of the dimer the one that is likely to isomerize upon Int. J. Mol. Sci. 2022, 23, 8690 9 of 20 photoexcitation to this state either through rotation/inversion in the excited state either by internal conversion through a conical intersection when present). • Irradiation with λ > 500 nm: This low-energy radiation is slightly below the lowest calculated energy excitation of the EE form from S0 to S1 (464 nm), but taking into account the error bars of the calculations and the broadness of the electronic transition bands, we can assume that this radiation allows the excitation of the EE initial form to S1 . The oscillator strength of this transition is quite small (0.1), but such a value is usually considered enough to permit the electronic transition. Once in the S1 excited state, the molecule will lose energy and eventually return to the ground electronic state, emitting radiation to return to the original EE structure or else by internal conversion passing through the CI that links EE to EZF . This would explain the predominance of the EZF structure as a result of this irradiation. • Irradiation with λ = 410 nm: This radiation does not produce any noticeable isomeriza- tion reaction as the EE isomer is the dominant species after irradiation. This is easily explained from our results as there are no allowed excitations close to this value for the EE isomer. • Irradiation with λ = 350 nm: In this case, it is experimentally found that the final majority product is the ZZ isomer. This seems anomalous as in order to obtain the ZZ isomer from the initial EE one, two consecutive isomerizations are needed: The first one from EE to either EZF or ZEF and a second one from each one of these “crossed” isomers to ZZ. According to the results shown in Figure 3, this wavelength should clearly excite the largely allowed transition of EE to S3 and probably also the transition to S4 , again with a large f value even if it is slightly above in energy. From our reasoning, this would mean that both EZF and ZEF isomers could be obtained. To explain the formation of the ZZ isomer, a second radiative process should sequentially occur. Our results allow for this to occur given that the 350 nm irradiation can photo-excite both the EZF and ZEF isomers as the two crossed isomers have mainly allowed transition at values slightly above 300 nm. In both cases this second electronic absorption would allow for the isomerization leading to the ZZ isomer. Previous literature extensively reports the feasibility of using such a stepwise, two-photon absorption process to fuel specific photochemical processes of interest [35–37]. Parenthetically, we note that at λ > 500 nm, the ZZ adduct is also experimentally obtained as a secondary but noticeable adduct (18%). Its presence should also be explained as the product of two consecutive irradiations of the same energy. In addition to the photoexcitations allowed for the EE isomer, there are additional transitions between pairs of isomers that are also possible according to the experimental results: • EZF → EE and ZEF → EE at λ = 410 nm: At this wavelength EZF could access the S2 state. Given that this state implies excitation of the F part it could isomerize to EE. Almost the same pattern can be invoked to explain the isomerization of ZEF back to EE. • EZF → ZZ and ZEF → ZZ at λ = 350 nm: These two transitions have been already discussed as the second step of the sequential two-photon absorption from EE to ZZ. The high energy irradiation would allow the excitation of the most probable transition (highest value of f) for both crossed structures where isomerization to the ZZ isomer is feasible. • ZZ → EE at λ = 410 nm: We encounter again a double isomerization process so that, according to our scheme, we will have to claim for a sequential double photo- absorption. For the ZZ isomer the transition to S2 is calculated at 430 nm and at this state the ZEF forms could be obtained. A quite close wavelength of 445 nm would excite the ZEF also to S2 , where the isomer EE could be finally obtained. The same process could be envisaged to take place also in the inverted order: ZZ → EZF → EE, implying successive light excitations of 445 and 430 nm (that is: identical values of the former process but also inverted!). • ZEF → EZF at λ > 500 nm: This seems the most difficult transition to explain as it has to involve a sequential double-electronic excitation by means of two low-energy Int. J. Mol. Sci. 2022, 23, 8690 10 of 20 photons. However, our calculations nicely explain this process assuming that tran- sitions calculated at wavelengths up to 440 nm can be accessed by the considered irradiation. In fact, as in the previous case, we have two possible paths and both are feasible: ZEF → ZZ → EZF with absorptions predicted at 465 and 445 nm, respectively, and ZEF → EE → EZF with two consecutive absorptions now calculated at 445 and 464 nm (curiously enough, again almost the same values but inverted!). In addition to the photochemistry of dimer 1 just discussed, we may also further analyze the isomerization mechanism in the ground electronic state. As seen in Figure 3, the energy barriers for the isomerizations in the ground electronic state are high enough (larger than 38 kcal/mol) to guarantee the relatively large stability of the final isomer that, eventually, will thermally revert to the most stable EE structure. In fact, it is found experi- mentally that half-lives of the unstable isomers obtained after light irradiation range from days to years [30]. This conclusion is not modified if we take into account the thermody- namic corrections. Table S1 of the Supplementary Material gives the enthalpies and Gibbs free energies of the four isomers and the corresponding transition states (using the quite drastic rigid rotor and the harmonic oscillator approximations). The nature of the transition states can also be disclosed by analyzing the direction of the transition vector (the motion that corresponds to the only one imaginary frequency in the localized transition states). In all the cases the motion of this imaginary frequency corresponds to an in-plane inversion of the N=N-C angle with the three atoms almost aligned in the transition state structure. Then, the inversion mechanism prevails over the torsion (rotation of the full phenyl ring) as already observed in previous systematic theoretical studies of different azobenzenes [38]. From the comprehensive analysis of dimer 1 just performed, we have obtained a profound insight of the physical reasons behind the quite complex photochemical behavior of this molecule. This analysis can now be used to our main goal: The design of more efficient azobenzene-based multi-state photoswitches whose properties could be modified “à-la-carte” using a computationally guided protocol. It is not easy to manage all factors affecting the photochemistry of a given multi-state photoswitch. A factor that appears as primordial, in order to be able to selectively excite the two AB moieties, is to pick up two monomeric units with absorption bands as separated as possible. Also, foreseeing the use of the designed photoswitch to be used in life sciences, the excitation sources should be kept at the lowest possible energy range. To selectively interconvert these structures by means of light irradiation is not an easy task as the absorption bands are usually quite broad, so a large energy separation between each pair of absorption bands is compulsory. Besides, to simultaneously isomerize both units, a double photon-absorption must take place. If the corresponding absorption bands are close enough, a single irradiation could accomplish such a goal (such a mechanism has been invoked previously to explain the behavior of dimer 1), but this contradicts the otherwise necessary large separation of bands to selectively isomerize each side of the dimer. Here, we open the door to a possibility, not considered from the experimental side: The irradiation of the sample with two different wavelengths that would trigger the two consecutive E/Z isomerizations. Let us now use these generic ideas to conceptually design an azobenzene multi-state photoswitch where all the isomerization paths could be selectively photoactivated. The ex- perimental and theoretical analysis of dimer 1 has clearly shown that the absorption spectra and isomerization ratios of the dimer are very similar to the sum of the photochemical prop- erties of the two isolated azobenzenic monomers [30], so in order to design new molecules, we can rely on the well-known behavior of substituted single azobenzene systems and the expected effect of different substituents. It is well known that azobenzene functionalized with four ortho-fluorine atoms results in a larger separation of the absorption bands of the E and Z isomers, so we keep that AB unit in our newly designed four-state photoswitch. As for the other moiety, we introduce a strong electronic-acceptor NO2 group in para and two electron-donor hydroxyl groups in ortho position in the other phenyl ring. This design is typical of a push-pull molecule. It is well known that the presence of such antagonistic substituents on opposite sides of the molecule dramatically shortens the HOMO–LUMO new molecules, we can rely on the well-known behavior of substituted single azobenzene systems and the expected effect of different substituents. It is well known that azobenzene functionalized with four ortho-fluorine atoms results in a larger separation of the absorp- tion bands of the E and Z isomers, so we keep that AB unit in our newly designed four- state photoswitch. As for the other moiety, we introduce a strong electronic-acceptor NO2 Int. J. Mol. Sci. 2022, 23, 8690 group in para and two electron-donor hydroxyl groups in ortho position in the other phe- 11 of 20 nyl ring. This design is typical of a push-pull molecule. It is well known that the presence of such antagonistic substituents on opposite sides of the molecule dramatically shortens the HOMO–LUMO gap, so that the bands in the absorption spectra could be considerably gap, so[39]. red-shifted thatPrevious the bands in the studies absorption of single spectra azobenzenes withcould be considerably push-pull red-shifted [39]. substituents have Previous confirmed thisstudies of single effect well-established azobenzenes with push-pull [40,41]. Finally, substituents all the ortho have confirmed this positions separating the well-established two ABs have been effect [40,41].with substituted Finally, methylall groups the ortho positionsthe to reinforce separating the two orthogonality of ABs have the been two moieties. The with substituted dimermethyl so designed, groupscalled dimer 2 from to reinforce now on, is depicted the orthogonality of the in two moieties. Scheme 2. The dimer so designed, called dimer 2 from now on, is depicted in Scheme 2. Scheme Scheme 2. Molecular 2. Molecular structure structure of2dimer of dimer 2 in stable in its most its most (EE)stable form.(EE) form. FirstFirst ofwe of all, all,proceeded we proceeded to perform to perform a studyasimilar study tosimilar toundertaken the one the one undertaken for dimer for dimer 1. Asdimer 1. As dimer 2 has 2 has notnot been been yet yet synthesized, synthesized, therethere is no is no experimental experimental data todata to compare compare with with and andTable Table 22 gives allthe gives all therelevant relevant information information of theofcalculated the calculated energiesenergies and characteristics of and characteristics of the theexcited excitedstate soso state that direct that comparison direct comparisonwithwith the same calculations the same for dimer calculations 1 (pre-1 (presented for dimer sented in Table 1) can be done. The two moieties of the dimer are now identifiedwhere in Table 1) can be done. The two moieties of the dimer are now identified where necessary necessary with subindex with subindex F (ortho-fluorinated F (ortho-fluorinated part) part) or or P (push-pullpart). P (push-pull part). Table 2. Relevant lowest electronic excited states of the four E/Z stable isomers of dimer 2. Isomer State E (eV) λ (nm) f Excitation S1 2.54 489 0.12 H−1 → L; H−4 → L (P) S2 2.69 462 0.10 H−2 → L+1; H−6 → L+1 (F) EE S4 3.59 345 1.36 H−1 → L; H−4 → L (P) S6 4.10 301 0.9 H−2 → L+1; H−6 → L+1 (F) S1 2.53 490 0.11 H−1 → L; H−4 → L (P) S2 2.89 429 0.06 H−9 → L+1; H−2 → L+1 (F) EP ZF S4 3.60 344 1.21 H−1 → L; H−4 → L (P) S8 4.44 280 0.11 H−5 → L+1; H−6 → L+1 (F) S1 2.58 480 0.13 H−1 → L; H−1 → L+2 (P) S2 2.67 464 0.08 H−6 → L+1; H−2 → L+1 (F) ZP EF S4 3.89 319 0.55 H−1 → L; H−5 → L+2 (P) S6 4.07 305 1.03 H−6 → L+1; H−2 → L+1 (F) S1 2.58 481 0.12 H−1 → L; H−1 → L+2 (P) S2 2.89 429 0.05 H−9 → L+1; H−2 → L+1 (F) ZZ S3 3.39 366 0.06 H → L; H → L+2 (F → P) S4 3.89 319 0.36 H−1 → L; H−5 → L+2 (P) S12 4.64 267 0.41 H−9 → L+1 (F) As done for dimer 1, we can use the calculated values posted in Table 2 to draw theoretical absorption spectra of dimer 2. The results are presented in Figure 5 for the four stable isomers of dimer 2 S4 3.89 319 0.55 H−1  L; H−5  L + 2 (P) S6 4.07 305 1.03 H−6  L + 1; H−2  L + 1 (F) S1 2.58 481 0.12 H−1  L; H−1  L + 2 (P) S2 2.89 429 0.05 H−9  L + 1; H−2  L + 1 (F) Int. J. Mol. Sci. 2022, 23, 8690 ZZ S3 3.39 366 0.06 H  L; H  L + 2 (F 12P) of 20 S4 3.89 319 0.36 H−1  L; H−5  L + 2 (P) S12 4.64 267 0.41 H−9  L + 1 (F) To proceed to the more relevant point here (the expected photochemical behavior of As done the newly for dimer designed 1, we can multi-state use the calculated photoswitch), we havevalues posted to turn in Table 2totoFigure the attention draw 6. the- It oretical absorption spectra of dimer 2. The results are presented in Figure 5 for the is identical to Figure 3 but with the data of dimer 2 (as presented in Table 2). This figure four stable isomers contains all theofrelevant dimer 2.information for the analysis of the photochemistry of dimer 2, which will be used from now on. Int. J. Mol. Sci. 2022, 23, 8690 13 of 21 Figure5.5. Computed Figure absorption spectra Computed absorption spectraof ofdimer dimer2 2ininacetonitrile acetonitrile at at thethe optimized optimized ground-state ground-state ge- geometries ometries ofofthethe four four stable stable isomers isomers of dimer of dimer 2. Vertical 2. Vertical red indicate red lines lines indicate the position the position of the of the vertical transitions vertical while the transitions whileblack the contour corresponds black contour to thetoconvolution corresponds withwith the convolution a Lorentzian function a Lorentzian de- function picted ininorder depicted ordertototheoretically theoreticallyreproduce reproducethetheexperimental experimentalabsorption absorptionspectrum. spectrum. ATosimple proceedandtoquick inspection the more relevantofpoint Figure 6 and here (the its comparison expected with the behavior photochemical analogous of Figure 3 fordesigned the newly dimer 1 multi-state reveals thatphotoswitch), the first goal we of our havedesign to turnis the fulfilled, as now attention for each to Figure 6. It isomeric form is identical there are to Figure four 3 but transitions with the data ofwith more dimer or presented 2 (as less relevant (f ≥ 0.1) in Table oscillator 2). This figure strengths that correspond to the selective excitations of just one unit of contains all the relevant information for the analysis of the photochemistry of dimerthe dimer, either2, through which willwhat can be be used loosely from now classified on. as an nπ* excitation (low energies but also low probabilities) or through ππ* excitations (higher energy but also much higher probability). Additionally, we note that the isomerization barriers in the ground electronic state remain high enough (though they are slightly lower compared to the ones for dimer 1 shown in Figure 3) to ensure long enough stability of the metastable isomeric structures which, in turn, are slightly more stable as compared with dimer 1. As for dimer 1, the thermodynamics of the ground electronic state does not qualitatively modify the energetic scheme of dimer 2 seen in Figure 6 (see Table S2). The analysis of the transition vectors in the four located Figure 5. Computed absorption spectra of dimer 2 in acetonitrile at the optimized ground-state ge- ometries of the four stable isomers of dimer 2. Vertical red lines indicate the position of the vertical transitions while the black contour corresponds to the convolution with a Lorentzian function de- picted in order to theoretically reproduce the experimental absorption spectrum. Int. J. Mol. Sci. 2022, 23, 8690 To proceed to the more relevant point here (the expected photochemical behavior 13 ofof 20 the newly designed multi-state photoswitch), we have to turn the attention to Figure 6. It is identical to Figure 3 but with the data of dimer 2 (as presented in Table 2). This figure contains transitionallstates the relevant information also confirms for themechanism the inversion analysis of for thethe photochemistry isomerization of dimer 2, reactions in which will be used from the ground electronic state.now on. Figure Figure6.6.Full Fullenergy energyscheme schemeofofthe theground groundandandlowest lowestexcited excitedelectronic electronicstates statesof ofdimer dimer2.2.See Seetext text for details. Relative energies of the different structures (horizontal lines) are given in kcal/mol. for details. Relative energies of the different structures (horizontal lines) are given in kcal/mol. Green Green arrows indicate the wavelengths of the vertical excitations from each conformer. Black lines arrows indicate the wavelengths of the vertical excitations from each conformer. Black lines indicate the ground electronic state whereas excited electronic states are in blue. Red lines correspond to crossing sections between S0 and S1 . Depictions of the molecular orbitals involved in the lowest-energy relevant excited electronic states of the EE isomer are shown in Figure 7. As for dimer 1, the shape of the orbitals allows for an easy assignation to either being in the corresponding benzenic plane (n) or else located symmetrically above and below this plane (π). They are also clearly localized in just one moiety of the dimer so that orthogonality of the electronic excitations (as seen in dimer 1) is preserved in the proposed new dimer. The corresponding orbitals for the other three isomers of dimer 2 are to be seen in Figures S5–S7. As done for dimer 1, the approximate geometries of the localized conical intersections (red horizontal lines in Figure 6) are presented in Figure S8. As we do not have any experimental data to compare with, it is not necessary to specifically analyze each feasible isomerization reaction. A more efficient way to grasp which processes can be photoactivated and which irradiation will activate them is the scheme depicted in Figure 8, which shows all the possible single photon reactions and the wavelength that is predicted to trigger each one of them. Again, we will assume that the azobenzene part that is photoexcited is the one that is able to eventually isomerize. As now there are always two such photoexcitations, involving either the nπ* low-energy or the ππ* high-energy electronic excited states, Figure 8 is divided into two parts: Part (a) on the left side corresponds to low energy excitations (λ > 425 nm) whereas part (b) on the right depicts the processes activated through high energy excitations (λ < 350 nm). (n) or else located symmetrically above and below this plane (π). They are also clearly localized in just one moiety of the dimer so that orthogonality of the electronic excitations (as seen in dimer 1) is preserved in the proposed new dimer. The corresponding orbitals for the other three isomers of dimer 2 are to be seen in Figures S5–S7. As done for dimer Int. J. Mol. Sci. 2022, 23, 8690 1, the approximate geometries of the localized conical intersections (red horizontal lines 14 of 20 in Figure 6) are presented in Figure S8. HOMO-9 HOMO-6 HOMO-5 Int. J. Mol. Sci. 2022, 23, 8690 15 of 21 HOMO-4 HOMO-3 HOMO-1 LUMO LUMO + 1 Figure Figure 7. Shape 7. Shape of the of the molecular molecular orbitals orbitals involved involved in the in the relevant relevant excited excited electronic electronic states states of the of the EEEE isomer of dimer 2. The drawings correspond to an iso-level of 0.02 a.u. of electronic density. The leftleft isomer of dimer 2. The drawings correspond to an iso-level of 0.02 a.u. of electronic density. The toptop corner corner of the of the Figure Figure depicts depicts thethe molecular molecular structure structure of the of the EE EE isomer. isomer. As we A naïf do not haveofany interpretation theexperimental data to8compare data seen in Figure with, it is would conclude notfor that necessary dimer 2 to specifically analyze each feasible isomerization reaction. A more one should be able to selectively trigger any isomerization process involving just efficient way to one grasp which processes can be photoactivated and which irradiation will monomeric part as, strictly speaking, each process has a different activation wavelength. activate them is the scheme depicted However, in Figure this conclusion 8, which that overlooks shows theallactual the possible single absorption photon bands arereactions known to and bethe wavelength quite broad (asthat is predicted previously to when noted triggerdealing each one of them. with dimerAgain, 1). Thewe will assume problem that is that thethe azobenzene actual extent ofpart that iselectronic a given photoexcited bandisisthe one that difficult is able from to assess to eventually a purely isomerize. theoreticalAs now point ofthere viewareas italways wouldtwo such photoexcitations, be necessary to proceed to involving either a careful (and the time-consuming quite nπ* low-energy or in the ππ* terms ofhigh-energy computer time) electronic analysisexcited of thestates, Figure structure vibrational 8 is divided intoelectronic of the two parts:bands, Part (a) andonthe thevalues left side corresponds given in Figureto low energy 8 (and excitations elsewhere (λ > 425 in the paper) nm) just whereas provide thepart (b) on the calculated position of the maximum right depicts the processes of the electronic activated absorption through band.excitations high energy In addition (λto that,nm). < 350 solvent and external conditions are also going to enlarge the broadening to an extent that is even LUMO + 1 Figure 7. Shape of the molecular orbitals involved in the relevant excited electronic states of the EE isomer of dimer 2. The drawings correspond to an iso-level of 0.02 a.u. of electronic density. The left top corner of the Figure depicts the molecular structure of the EE isomer. Int. J. Mol. Sci. 2022, 23, 8690 15 of 20 As we do not have any experimental data to compare with, it is not necessary to specifically analyze each feasible isomerization reaction. A more efficient way to grasp which to harder processes evaluatecanfrombeaphotoactivated purely theoretical andview. whichAtirradiation will any rate, it activate that is obvious themwhere is the scheme two depicted absorbing in Figure bands have a8,more whichseparated shows allmaxima the possible singlelikely it is more photon reactions that they couldand be the wavelength selectively that is Comparison excited. predicted to trigger each oneexcitations of the allowed of them. Again, between wethewilloriginal assumedimer that the 1 azobenzene and part thathere dimer 2 proposed is photoexcited (Figures 3 and is the one that 6) clearly is ablethat discloses to eventually in dimer 2 the isomerize. calculated As excitation now therewavelengths are always twoof the two such moieties of the involving photoexcitations, dimer are either more separated in both the the nπ* low-energy or high (ππ*) and low (nπ*) energy regimes. Additionally, in dimer 2 all the isomeric the ππ* high-energy electronic excited states, Figure 8 is divided into two parts: Part (a) species have four on the leftallowed excitation to side corresponds bands low (with energyoscillator strengths excitations (λ > 425greater than 0.1), nm) whereas whereas part in (b) on the dimer 1 just three right depicts of them (two the processes for thethrough activated EZF isomer) high are allowed. energy This difference excitations is in favor (λ < 350 nm). of dimer 2 as it opens the door for more photochemical paths to excite a particular isomer. Figure 8. Allowed photo-isomerizations for dimer 2. (a): Activated through low-energy photoexcita- Figure tion (λ >8.425 Allowed photo-isomerizations nm). (b): for dimer 2. photo-excitation Activated through high-energy (a): Activated through low-energy (λ < 350 nm). photoexci- tation (λ > 425 nm). (b): Activated through high-energy photo-excitation (λ < 350 nm). Within the limitations just explained, we can now proceed to a specific analysis of the open A naïfto ways interpretation photoexcite of theisomeric each data seenspecies in Figure 8 would of dimer 2. conclude that for dimer In the following 2 one discussion should be able to selectively trigger any isomerization process involving it will be assumed that an energy separation between two absorption bands larger than just one mono- meric 0.5 eV ispart highas,enough strictlyto speaking, each process permit selective has a different excitation, activation wavelength. whereas differences lower than 0.2 How- eV ever, this conclusion overlooks that the actual absorption bands are known imply that both moieties of the dimer will be jointly photoexcited. As for intermediate to be quite broad separations, energy (as previously thenoted when possible dealingexcitation, selective with dimer 1). The problem as discussed before,iswill thatdepend the actual on extent the of a given unknown shapeelectronic band is difficult of each absorption band. to assess from a purely theoretical point of view as it would be necessary to proceed to a careful (and quite time-consuming in terms • Irradiation of the EE isomer: of computer time) analysis of the vibrational structure of the electronic bands, and the As this is the most stable and, presumably, only present form for dimer 2 in the absence of irradiation, the most relevant photochemical processes are taking place departing from this structure. At the long wavelength side of the spectrum (Figure 8a), it is seen that there are two allowed processes at 489 and 462 nm that should mainly lead to ZP EF and EP ZF isomers, respectively. The 0.14 eV difference between both radiations is not enough to allow a selective excitation of both isomers. As for the EE → ZZ process it will need a sequential process involving an initial photo-excitation to either ZP EF or EP ZF forms that would be, almost immediately, excited by a second photon. These processes are predicted to involve two absorptions of 489 and 464 nm for the ZZ isomerization through ZP EF or 462, and 490 nm if it goes through EP ZF . In both cases, the two irradiations are quite close (0.14/0.15 eV) and it is likely that experimentally their bands will overlap so that a single irradiation near the 500 nm mark (the one used to photoexcite dimer 1) may allow for the double EE to ZZ isomerization process. Things are different for the high-energy irradiation of EE (Figure 8b). In that case, irradiations of 345 and 301 nm will mainly lead to ZP EF and EP ZF , respectively. The noticeable energy difference of these two bands (0.53 eV) now seems enough to guarantee selective excitation of the EE initial structure. As for the double isomerization to ZZ, it would need two photon absorptions either at 345 and 305 nm (if passing through ZP EF ) or 301 and 344 nm (passing through EP ZF ). Both paths have now quite similar energy differences between the two predicted absorption bands of ca. 0.5 eV, a value also high enough to anticipate that, within our theoretical results, the process may not proceed with just one irradiation wavelength. At this point, an alternative procedure, not considered Int. J. Mol. Sci. 2022, 23, 8690 16 of 20 in the experimental setup used for dimer 1, is envisaged: Irradiation of the sample with two different wavelengths that would trigger the consecutive two E/Z isomerizations. • Irradiation of the ZP EF isomer: At the higher wavelength range, the calculations predict that this isomer could be irradiated by close wavelengths of 480 and 464 nm to obtain the EE and ZZ structures, respectively, so it may be impossible to selectively obtain just one form with low-energy irradiation. To directly produce the EP ZF crossed isomer, two sequential absorptions of 480 + 462 nm or 464 + 481 nm are needed. The latter implies just a 0.09 eV difference between both bands, so it should be possible to accomplish the two consecutive photoexci- tations with a single wavelength again near the 500 nm mark. As for the high-energy range of photoexcitations, ZP EF isomer may be photoexcited at 319 and 305 nm, respectively leading to EE and ZZ isomers. Again, the two radiations appear to be too close in energy to guarantee selectivity in the isomerization through irradiation. As for the two possible sequential two-photon processes necessary to obtain the EP ZF isomer, they are predicted at sequences of 319 + 301 nm or else 305 + 319 nm. In both cases, the energy separation between both excitations (0.18/0.23 eV) does not clarify whether the double isomerization could be feasible with just one wavelength or if, depending on the experimental setup, two different irradiation frequencies might be necessary. • Irradiation of the EP ZF isomer: In the long wavelength range of photoexcitations, this isomer can proceed to EE at 429 nm and to ZZ at 490 nm. As for the double photo-absorption leading to the ZP EF isomer, the maxima of the two involved absorption bands are predicted either at 429 and 489 nm or 490 and 429 nm. The energy difference (ca. 0.35 eV) between these two bands does not allow us to assess if selective photoexcitation will be possible within this wavelength range. Clear differences are seen in the high-energy range as EP ZF could revert to EE and ZZ in a single process with irradiations of 280 or 344 nm, respectively, so now a selective photoisomerization should be possible (0.83 eV difference). As for the two absorptions needed to obtain the ZP EF , they would involve quite separate successive irradiations of 280 and 345 nm or else 344 and 267 nm, making the double isomerization process accessible only through a stepwise two-photon absorption process involving two different light sources. • Irradiation of the ZZ isomer: According to our calculations, the two single photoisomerizations from ZZ could be selectively addressed in the high-energy region as the wavelengths predicted to photoexcite ZZ to EP ZF and ZP EF are, respectively, 319 and 267 nm (0.75 eV away). Things are not so clear at the low energy range, where photoexcitation at 481 nm is predicted to lead to EP ZF , whereas irradiation at 429 nm (0.31 eV difference) will produce ZP EF . As for the photochemical paths leading to the double isomerization from ZZ to EE in the low-energy domain, they are predicted at 429 and 480 nm or else 481 and 429 nm, an intermediate separation of around 0.31 eV in both cases that makes it again difficult to predict whether a single irradiation may trigger the process or if two different irradiations will be needed. Things are clearer at the high-energy range as the double isomerization implies well separated (>0.5 eV) sequential absorptions of 267 and 319 nm or else 319 and 280 nm. 3. Materials and Methods All theoretical calculations have been done with the GAUSSIAN09 program [42]. Ace- tonitrile, the solvent used in the experimental work by Zhao et al. [30] is introduced in all the calculations performed through the use of the continuum PCM method as implemented in the GAUSSIAN09 suite of programs [43,44]. The geometries of all the stable isomers in the ground electronic states have been obtained through full geometry optimization at the DFT level within the hybrid long-range corrected CAM-B3LYP functional, which avoids the problems of non-long-range-corrected functionals (such as B3LYP) that grossly under- estimate the excitation energies of excited states of internal charge-transfer type [45,46]. Int. J. Mol. Sci. 2022, 23, 8690 17 of 20 In the recent past, CAM-B3LYP has been proved to perform well for organic molecules of moderate size with a systematic tendency to overestimate excitation energies [47–49]. This error has the attractive feature of being quite constant [47]. Recent benchmarks analyz- ing a large set of molecules confirm CAM-B3LYP among the best functionals to accurately measure vertical excitation energies [50,51]. Additional technical details of the calculations may be found in the Supplementary Material. Frequency calculations have been carried out at the optimized geometries in the ground electronic state to confirm that they are true minima (all eigenvalues of the second derivative Hessian matrix are positive). Transition states for the isomerization reaction in the ground state have also been directly located and fully characterized (with exactly one negative eigenvalue of the Hessian matrix). Electronic excited states have been obtained using the linear time-dependent formalism at the DFT level (TD-DFT) without the Tamm–Dancoff approximation. The basis set used is 6-311+G(d,p), a triple-ζ quality split-valence that also includes polarization and diffuse functions [52,53]. We have tested that the use of the larger def2-TZVP basis set [54,55] does not appreciably modify the geometries and the relative energies of the different isomers and the same holds for the calculated excitation energies as differences between the two basis sets well below 0.1 eV, as posted in Tables S3 and S4 in the Supplementary Material. Geometry optimizations in the excited state were unsuccessful as the optimization systematically turned to points with almost null excitation energies where the optimization failed or entered a loop. This is a clear indication that after photoexcitation the geometry relaxes to a point where it crosses the potential energy surface of the ground electronic state. These points are called conical intersections (CIs) and their presence is known to be ubiquitous in almost every photochemical process involving E/Z isomerizations [56–58]. A more precise localization of the CI structures is not possible within our single-reference DFT method as it would require a multi-determinantal methodology, but, as discussed in the results and discussion section, this precision is also not necessary in the present context. Cartesian coordinates for all the located geometries of interest (minima, transition states and conical intersections for both dimer 1 and dimer 2) are given in Tables S5–S25. 4. Conclusions In this work, we have performed high-level computations devoted to rationalizing the photochemistry of multi-state photoswitches. This task has allowed the design of a conceptual and computational mechanism to propose brand new molecular architectures with specific properties. To do so, we have first considered a bisazobenzene system (dimer 1), well-studied from the experimental side. We have carefully analyzed from a theoretical point of view the relevant features of this molecule. Comparison of the obtained data with the experimentally known facts has allowed us to gain full comprehension of the mechanisms governing the photochemistry of this four-state photoswitch. The main target of such a multi-state photoswitch is that the different isomers could be selectively interconverted using different sets of light irradiations. In dimer 1 this goal was just partially accomplished as only two out of three isomers were obtained from the most stable isomer with irradiation of different wavelengths. Moreover, some other interconversions between isomers were not experimentally observed to occur at any of the considered irradiations. The deeper insight acquired from our computational design has allowed us to attribute these drawbacks to the fact that the energies of the photons needed to excite each azobenzenic unit are not separated enough to allow selective isomerization. Combining this strategy with the huge wealth of knowledge about the influence of different substituents on the electronic excitation energies of single azobenzenes, a new bisazobenzene (dimer 2) that should overcome the shortcomings of the previous one is proposed. Calculations on dimer 2 confirm the expectations (to be experimentally confirmed) of our computationally guided design. Computations predict that each isomer may absorb light of four different wavelengths: Two of high energy (λ < 350 nm) and two of low energy Int. J. Mol. Sci. 2022, 23, 8690 18 of 20 (λ > 425 nm), in each case leading to excitations of a different azobenzene unit. This opens the door to more photochemical paths that may selectively address a given isomerization process. For dimer 1, just one excitation was allowed (f ≥ 0.1) in the low energy region for all the isomeric species. Departing from the EE isomer, the only thermodynamically stable species of dimer 2, our calculations predict that the initial, and more stable, EE structure could be selectively photoexcited to the other three isomers using light of three different wavelengths centered respectively at 301 nm, leading to EP ZF , 345 nm for the ZP EF structure and 464 + 489 nm (probably just one single wavelength given the broadness of the absorption bands) to mainly obtain the ZZ isomer. The fact that for dimer 2 the separation of the absorption bands of both moieties has been enlarged makes the absorption of two consecutive photons of the same energy, needed to isomerize both moieties of the dimer with a single irradiation, a process that in dimer 1 was the key to obtain the ZZ isomer from the initial EE one in the previous experimental work, less likely. The result obtained for dimer 2 has led us to propose a brand- new procedure to selectively access the different isomers of a multiazobenzene system not considered up to now: A simultaneous irradiation with two different light sources. Given that there are four isomeric structures, and so twelve possible interconversions between each pair of metastable structures, finding an ideal molecular system where each transition could be triggered through single irradiations of light with separated-enough wavelengths is an unattainable goal, so that the eventual use of pairs of excitations, as proposed here, may be a clever way to increase the available tools to selectively trigger such multi-step processes through different photoirradiations. Our computational strategy is able to deal with more complex systems with three (or more) “isomerizable” units. The star-like structure with three azobenzene units linked to a central core is a paradigmatic case that has aroused considerable interest in recent years [26–29]. A clever choice of different substituents at each branch should lead to a system with well-separated electronic absorption bands so that selective isomerizations could be accomplished using either one or two different irradiation wavelengths. Additional work towards this goal is in progress in our laboratory. Supplementary Materials: The supporting information can be downloaded at: https:www.mdpi. com/article/10.3390/ijms23158690/s1. Author Contributions: Conceptualization, M.M., R.G. and J.M.L.; investigation, M.M., R.G. and J.M.L.; writing—original draft preparation, M.M., R.G. and J.M.L.; writing—review and editing, M.M., R.G. and J.M.L.; funding acquisition, M.M., R.G. and J.M.L. All authors have read and agreed to the published version of the manuscript. Funding: This study was supported by the Spanish “Ministerio de Ciencia e Innovación” through project PID2020-113764GB-I00. We also acknowledge CSUC (“Consorci de Serveis Universitaris de Catalunya”) for the use of computational facilities. Informed Consent Statement: Not applicable. Data Availability Statement: The data presented in this study are contained within the article. Conflicts of Interest: The authors declare no conflict of interest. References 1. Neilson, B.M.; Bielawski, C.W. Illuminating photoswitchable catalysis. ACS Catal. 2013, 3, 1874–1885. [CrossRef] 2. Beharry, A.A.; Wooley, G.A. Azobenzene photoswitches for biomolecules. Chem. Soc. Rev. 2011, 40, 4422–4437. [CrossRef] [PubMed] 3. Klajn, R. Spyropyran-based dynamic materials. Chem. Soc. Rev. 2014, 43, 148–184. [CrossRef] [PubMed] 4. Katsonis, N.; Lubomska, M.; Pollard, M.M.; Feringa, B.L.; Rudolf, P. Synthetic light-activated molecular switches and motors on surfaces. Prog. Surf. Sci. 2007, 82, 407–434. [CrossRef] 5. Szymanski, W.; Beierle, J.M.; Kistemaker, H.A.; Velema, W.A.; Feringa, B.L. Reversible photocontrol of biological systems by the incorporation of molecular photoswitches. Chem. Rev. 2013, 113, 6114–6178. [CrossRef] [PubMed] 6. Hüll, K.; Morstein, J.; Trauner, D. In vivo photopharmacology. Chem. Rev. 2018, 118, 10710–10747. [CrossRef] Int. J. Mol. Sci. 2022, 23, 8690 19 of 20 7. Manna, D.; Udayabhaskararao, T.; Zhao, H.; Klajn, R. Orthogonal light-induced self-assembly of nanoparticles using differently substituted azobenzenes. Angew. Chem. Int. Ed. 2015, 54, 12394–12397. [CrossRef] 8. Borowiak, M.; Nahaboo, W.; Reynders, M.; Nekolla, K.; Jalinot, P.; Hasserodt, J.; Rehberg, M.; Delattre, M.; Zahler, S.; Vollmar, A.; et al. Photoswitchable inhibitors of microtubule dynamics optically control mitosis and cell death. Cell 2015, 162, 403–411. [CrossRef] 9. Ankenbruck, N.; Courtney, T.; Naro, Y.; Deiters, A. Optochemical control of biological processes in cells and animals. Angew. Chem. Int. Ed. 2018, 57, 2768–2798. [CrossRef] 10. Dhammika Bandara, H.M.; Burdette, S.C. Photoisomerization in different classes of azobenzene. Chem. Soc. Rev. 2012, 41, 1809–1825. [CrossRef] 11. Sadovski, O.; Beharry, A.A.; Zhang, F.; Woolley, G.A. Spectral tuning of azobenzene photoswitches for biological applications. Angew. Chem. Int. Ed. 2009, 48, 1484–1486. [CrossRef] [PubMed] 12. Merino, E.; Ribagorda, M. Control over molecular motion using the cis-trans photoisomerization of the azo group. Beilstein J. Org. Chem. 2012, 8, 1071–1090. [CrossRef] [PubMed] 13. Wegener, M.; Hansen, M.J.; Driessen, A.J.M.; Szymanski, W.; Feringa, B.L. Photocontrol of antibacterial activity: Shifting from UV to red light activation. J. Am. Chem. Soc. 2017, 139, 17979–17986. [CrossRef] 14. Bahrenburg, J.; Sievers, C.M.; Schönbron, J.B.; Hartke, B.; Renth, F.; Temps, F.; Näther, C.; Frank, d.; Sönnichsen, F.D.; Bahrenburg, J. Photochemical properties of multi-azobenzene compounds. Photochem. Photobiol. Sci. 2013, 12, 511–518. [CrossRef] [PubMed] 15. Vapaavori, J.; Goulet-Hanssens, A.; Heikkinen, I.T.S.; Barrett, C.J.; Priimagi, A. Are two azo groups better than one? Investigating the photoresponse of polimer-bisazobenzene complexes. Chem. Mater. 2014, 26, 5089–5096. [CrossRef] 16. Fihey, A.; Perrier, A.; Browne, W.R.; Jacquemin, D. Multiphotochromic molecular systems. Chem. Soc. Rev. 2015, 44, 3719–3759. [CrossRef] 17. Erbas-Cakmak, S.; Leigh, D.A.; McTernan, C.T.; Nussbaumer, A.L. Artificial molecular machines. Chem. Rev. 2015, 115, 10081–10206. [CrossRef] 18. Coskun, A.; Banaszak, M.; Astumian, R.D.; Stoddart, J.F.; Grzybowski, B.A. Great expectations: Can artificial molecular machines deliver on their promise? Chem. Soc. Rev. 2012, 41, 19–30. [CrossRef] 19. Hoffmann, K.; Guentner, M.; Mayer, P.; Dube, H. Symmetric and nonsymmetric bis-hemithioindigos precise visible light controlled shape-shifters. Org. Chem. Front. 2019, 6, 1244–1252. [CrossRef] 20. Cisnetti, F.; Ballardini, R.; Credi, A.; Gandolfi, M.T.; Masiero, S.; Negri, F.; Pieraccini, S.; Spada, G.P. Photochemical and electronic properties of conjugated bis(azo) compounds: An experimental and computational study. Chem. Eur. J. 2004, 10, 2011–2021. [CrossRef] 21. Robertus, J.; Reker, S.F.; Pijper, T.C.; Dezeman, A.; Browne, W.R.; Feringa, B.L. Kinetic analysis of the thermal isomerization pathways in an asymmetric double azobenzene switch. Phys. Chem. Chem. Phys. 2012, 14, 4374–4382. [CrossRef] [PubMed] 22. Floss, G.; Saalfrank, P. The photoinduced E → Z isomerization of bisazobenzenes: A surface hopping molecular dynamics study. J. Phys. Chem. A 2015, 119, 5026–5037. [CrossRef] [PubMed] 23. Slavov, C.; Yang, C.; Scheighauser, L.; Boumrifak, C.; Drew, A.; Wegner, H.A.; Wachtveitl, J. Connectiviy matters—Ultrafast isomerization dynamics of bisazobenzene photoswitches. Phys. Chem. Chem. Phys. 2016, 18, 14795–14804. [CrossRef] [PubMed] 24. Heindl, A.H.; Becker, J.; Wegner, H.A. Selective switching of multiple azobenzenes. Chem. Sci. 2019, 10, 7418–7425. [CrossRef] [PubMed] 25. Reuter, R.; Wegner, H.A. Oligoazobenenophanes-synthesis, photochemistry and properties. Chem. Commun. 2011, 47, 12267–12276. [CrossRef] 26. Koch, M.; Saphiannikova, M.; Santer, S.; Guskova, O. Photoisomers of azobenzene star with a flat core: Theoretical insights into multiple states from DFT and MD perspective. J. Phys. Chem. B 2017, 121, 8854–8867. [CrossRef] 27. Yang, C.; Slavov, C.; Wegner, H.A.; Watchtveitl, J.; Drew, A. Computational design of a molecular triple photoswitch for wavelength-selective control. Chem. Sci. 2018, 9, 8665–8672. [CrossRef] 28. Galanti, A.; Santoro, J.; Mannancherry, R.; Duez, Q.; Diez-Cabanes, V.; Valasek, M.; De Winter, J.; Cornil, J.; Gerbaux, P.; Mayor, M.; et al. A new class of rigid multi(azobenzene) switches featuring electronic decoupling: Unraveling the isomerization in individual photochromes. J. Am. Chem. Soc. 2019, 141, 9273–9283. [CrossRef] 29. Gupta, D.; Gaur, A.K.; Kumar, P.; Kumar, H.; Mahadevan, A.; Devi, S.; Roy, S.; Venkataramani, S. Tuning of Bistability, Thermal Stability of the Metastable States, and Application Prospects in the C3 -Symmetric Designs of Multiple Azo(hetero)arenes Systems. Chem. Eur. J. 2021, 27, 3463–3472. [CrossRef] 30. Zhao, F.; Grubert, L.; Hecht, S.; Bléger, D. Orthogonal switching in four-state azobenzene mixed-dimers. Chem. Commun. 2017, 53, 3323–3326. [CrossRef] 31. Bléger, D.; Schwarz, J.; Brouwer, A.M.; Hecht, S. o-Fluoroazobenzenes as readily synthesized photoswitches offering nearly quantitative two-way isomerization with visible light. J. Am. Chem. Soc. 2012, 134, 20597–20600. [CrossRef] [PubMed] 32. Knie, C.; Utecht, M.; Zhao, F.; Kulla, H.; Kovalenko, S.; Brouwer, A.M.; Saalfrank, P.; Hecht, S.; Bléger, D. ortho-Fluoroazobenzenes: Visible light switches with very long-lived Z isomers. Chem. Eur. J. 2014, 20, 16492–16501. [CrossRef] [PubMed] 33. Bléger, D.; Dokic, J.; Peters, M.V.; Grubert, L.; Saalfrank, P.; Hecht, S. Electronic decoupling approach to quantitative photoswitch- ing in linear multiazobenzene architectures. J. Phys. Chem. B 2011, 115, 9930–9940. [CrossRef] [PubMed] 34. Dennington, R.; Keith, T.A.; Millam, J.M. GaussView, Version 6; Semichem Inc.: Shawnee Mission, KS, USA, 2016. Int. J. Mol. Sci. 2022, 23, 8690 20 of 20 35. Ward, C.L.; Elles, C.G. Cycloreversion dynamics of a photochromic molecular switch via one-photon and sequential two-photon excitation. J. Phys. Chem. A 2014, 118, 10011–10019. [CrossRef] [PubMed] 36. Thum, M.D.; Falvey, D.E. Photoreleasable protecting groups triggered by sequential two-photon absorption of visible light: Release of carboxylic acids from a linked anthraquinone-N-alkylpicolinium ester molecule. J. Phys. Chem. A 2018, 122, 3204–3210. [CrossRef] 37. Ghosh, I.; Ghosh, T.; Bardagi, J.L.; König, B. Reduction of aryl halides by consecutive visible light-induced electron transfer processes. Science 2014, 346, 725–728. [CrossRef] 38. Zapata, F.; Fernández-González, M.A.; Rivero, D.; Álvarez, A.; Marazzi, M.; Frutos, L.M. Toward an optomechanical control of photoswitches by tuning their spectroscopical properties: Structural and dynamical insights into azobenzene. J. Chem. Theory Comput. 2014, 10, 312–323. [CrossRef] 39. Bures, F. Fundamental aspects of property tuning in push-pull molecules. RSC Adv. 2014, 4, 58826–58851. [CrossRef] 40. Silva, D.L.; Krawczyk, P.; Bartkowiak, W.; Mendonça, C.R. Theoretical study of one- and two-photon absorption spectra of azoaromatic compounds. J. Chem. Phys. 2009, 131, 244516. [CrossRef] 41. Corchado, J.C.; Luz Sánchez, M.; Fdez Galvan, I.; Martín, M.E.; Muñoz-Losa, A.; Barata-Morgado, R.; Aguilar, M.A. Theoretical study of solvent effects on the ground and low-lying excited free energy surfaces of a push-pull substituted azobenzene. J. Phys. Chem. B 2014, 118, 12518–12530. [CrossRef] 42. Frisch, M.J.; Trucks, G.W.; Schlegel, H.B.; Scuseria, G.E.; Robb, M.A.; Cheeseman, J.R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G.A.; et al. Gaussian 09; Gaussian, Inc.: Wallingford, CT, USA, 2009. 43. Miertus, S.; Scrocco, E.; Tomasi, J. Electrostatic interaction of a solute with a continuum. A direct utilization of ab initio molecular potentials for the prevision of solvent effects. Chem. Phys. 1981, 55, 117–129. [CrossRef] 44. Improta, R.; Barone, V.; Scalmani, G.; Frisch, M.J. Using redundant internal coordinates to optimize equilibrium geometries and transition states. J. Comput. Chem. 1996, 17, 49–56. 45. Handy, N.C. The molecular physics lecture 2004: (i) Density functional theory, (ii) Quantum Monte Carlo, molecular physics. Mol. Phys. 2004, 102, 2399–2409. [CrossRef] 46. Zhang, M.Y.; Wang, J.Y.; Lin, C.S.; Cheng, W.D. First-principles study of one- and two-photon absorption of the H-bonding complexes from monomeric red fluorescent proteins with large Stokes shifts. J. Phys. Chem. B 2011, 115, 10750–10757. [CrossRef] [PubMed] 47. List, H.N.; Olsen, J.M.; Rocha-Rinza, T.; Christiansen, O.; Kongsted, J. Performance of popular XC-functionals for the description of excitation energies in GFP-like chromophore models. Int. J. Quantum Chem. 2012, 112, 789–800. [CrossRef] 48. Randino, C.; Moreno, M.; Gelabert, R.; Lluch, J.M. Peek at the potential energy surfaces of the LSSmkate1 and LSSmKate2 proteins. J. Phys. Chem. B 2012, 116, 14302–14310. [CrossRef] 49. Moreno, M.; Gelabert, R.; Lluch, J.M. deciphering the grounds of the suitability of acylhydrazones as efficient photoswitches. Phys. Chem. Chem. Phys. 2019, 21, 16075–16082. [CrossRef] 50. Brémond, E.; Savarese, M.; Adamo, C.; Jacquemin, D. Accuracy of TD-DFT geometries: A fresh look. J. Chem. Theory Comput. 2018, 14, 3715–3727. [CrossRef] 51. Liang, J.; Feng, X.; Hait, D.; Head-Gordon, M. Revisiting the performance of time-dependent density functional theory for electronic excitations: Assessment of 43 popular and recently developed functionals from rungs one to four. J. Chem. Theory Comput. 2022, 18, 3460–3473. [CrossRef] 52. McLean, A.D.; Chandler, G.S. Contracted gaussian basis set for molecular calculations I. 2nd row atoms, Z = 11–18. J. Chem. Phys. 1980, 72, 5639–5648. [CrossRef] 53. Raghavachari, K.; Binkley, J.S.; Seeger, R.; Pople, J.A. Self-consistent molecular orbital methods 20. Basis set for correlated wave-functions. J. Chem. Phys. 1980, 72, 650–654. 54. Weigend, F.; Ahlrichs, R. Balanced basis sets of split valence, triple zeta valence and quadruple zeta valence quality for H to Rn: Design and assessment of accuracy. Phys. Chem. Chem. Phys. 2005, 7, 3297–3305. [CrossRef] 55. Weigend, F. Accurate Coulomb-fitting basis sets for H to Rn. Phys. Chem. Chem. Phys. 2006, 8, 1057–1065. [CrossRef] [PubMed] 56. Domcke, W.; Yarkony, D.R.; Köppel, H. Conical Intersections: Electronic Structure, Dynamics and Spectroscopy; World Scientific Publishing: Singapore, 2004. 57. Domcke, W.; Yarkony, D.R.; Köppel, H. Conical Intersections: Theory, Computation and Experiment; World Scientific Publishing: Singapore, 2011. 58. Yarkony, D.R. Nonadiabatic quantum chemistry—Past, present, and future. Chem. Rev. 2012, 112, 481–498. [CrossRef] [PubMed]