Electronic Absorption, Emission, and Two-Photon Absorption Properties of Some Extended 2,4,6-Triphenyl-1,3,5-Triazines

We report herein the linear optical properties of some extended 2,4,6-triphenyl-s-triazines of formula 2,4,6-[(1,4-C6H4)C≡C(4-C6H4X)]3-1,3,5-(C3H3N3) (3-X; X = NO2, CN, OMe, NMe2, NPh2) and related analogues 4 and 7-X (X = H, NPh2), before briefly discussing their two-photon absorption (2PA) cross-sections. Their 2PA performance is discussed in relation to 2PA values previously measured for closely related octupoles such as N,N′,N′′-triphenylisocyanurates (1-X, 5, and 6-X) or 1,3,5-triphenylbenzenes (2-X). While s-triazines are usually much better two-photon absorbers in the near-IR range than these molecules, especially when functionalised by electron-releasing substituents at their periphery, they present a decreased transparency window in the visible range due to their red-shifted first 1PA peak, in particular when compared with corresponding isocyanurates analogues. In contrast, due to their significantly larger two-photon brilliancy, 2,4,6-triphenyl-s-triazines appear more promising than the latter for two-photon fluorescence bio-imaging purposes. Rationalisation of these unexpected outcomes is proposed based on DFT calculations.


Introduction
Planar molecules featuring trigonal symmetry have attracted sustained attention for their second-order nonlinear optical (NLO) properties since the late eighties [1][2][3][4]. Initially aroused by the quest for molecules with large second-order NLO properties, these so-called "octupolar" molecules were likely to exhibit sizeable hyperpolarizabilities due to their peculiar symmetry, in reason of the existence of off-diagonal tensorial elements in the electronic coupling matrix between peripheral branches [2]. It was subsequently shown that this symmetry can also be potentially beneficial to third-order NLO properties such as two-photon absorption (2PA). Given the very appealing societal prospects for dyes presenting large 2PA cross-sections [5][6][7][8], in particular regarding fluorescence bioimaging when the dye also fluoresces [6,9], we have started exploring the 2PA properties of various families of molecules, such as extended 1,3,5-triaryltriazinane-2,4,6-triones (1-X; Scheme 1) [10,11] or 1,3,5-triphenylbenzene (2-X) [12]. Both 1-X (also known as N,N ,Ntriphenylisocyanurates [13]) and 2-X derivatives proved to be good two-photon absorbers, especially when functionalised by electron-releasing groups at their periphery [11,12]. Actually, in line with these findings as well as with independent reports [14], we observed that triphenylisocyanurates [13]) and 2-X derivatives proved to be good two-photon absorbers, especially when functionalised by electron-releasing groups at their periphery [11,12]. Actually, in line with these findings as well as with independent reports [14], we observed that the cross-section of the first 2PA peak for these derivatives increased with the polarisation of the peripheral arms, typically when strongly electron-releasing X substituents were present (X = OMe, NMe2, NPh2). However, we also observed that upon progressing from the isocyanurate core (1-X) to the slightly less electron-deficient 1,3,5-phenylene core (2-X), a slight decrease in the 2PA cross-section occurred for the latter derivatives but only for the most electron-releasing substituents. This suggested that the polarisation induced by the former core was slightly more favourable to 2PA than that induced by the second. In this respect, surmising that the 1,3,5-triazine core was more electron-attracting than the isocyanurate one, it was now interesting to study the 2PA properties and the fluorescence of related 1,3,5-triazine analogues of 1-X such as 3-X (X = NO2, CN, OMe, NMe2, and NPh2). Thus far, emissive triphenyl-s-triazine derivatives and related extended analogues have mainly raised interest in the field of OLEDs or closely related fields [15][16][17][18]; however, very few studies were actually focused on the NLO properties of such derivatives. While related derivatives such as extended trialkynyl-s-triazines [19], trialkenyls-triazines [20,21], or tris(2-thienyl)-s-triazines [22][23][24][25] have given rise to some 2PA investigations, to the best of our knowledge, only one recent theoretical paper deals specifically with the second-order NLO properties of molecules such as 3-X [26], and only one other single paper addresses the 2PA properties of an s-triazine derivative closely related to 7-H [27]. Thus, regarding extended triphenyl-s-triazines, essentially styryl-type analogues of 3-X have been investigated so far for their 2PA properties [28][29][30]. A general experimental and theoretical study focused on the 2PA properties of octupolar compounds such as 3-X and 7-X would therefore be timely. Furthermore, anticipating the well-known propensity of nitro substituents to poison fluorescence in 3-NO2, we also targeted a derivative such as 4, which similar to 3-CN, constitutes another example of a compound featuring electron-withdrawing arms [31]. The optical properties of this new triazine derivative might then be compared with those of its known isocyanurate analogue (5) [32]. Finally, given that the extended 4-fluorenyl derivative 6-NPh2 exhibits the most promising optical properties for bio-imaging purposes among all triarylisocyanurates studied so far [33], the study of triazine analogues such as 7-X (X = H, NPh2) was also required. Accordingly, in the following we will first start with the synthesis of the targeted molecules (3-X, 4, and 7-X) and study the 2PA properties of their emissive representatives via two-photon excited fluorescence (TPEF). Their optical properties of interest will then be discussed with the help of density functional theory (DFT) and time-dependent DFT (TD-DFT) calculations. In this respect, surmising that the 1,3,5-triazine core was more electron-attracting than the isocyanurate one, it was now interesting to study the 2PA properties and the fluorescence of related 1,3,5-triazine analogues of 1-X such as 3-X (X = NO 2 , CN, OMe, NMe 2 , and NPh 2 ). Thus far, emissive triphenyl-s-triazine derivatives and related extended analogues have mainly raised interest in the field of OLEDs or closely related fields [15][16][17][18]; however, very few studies were actually focused on the NLO properties of such derivatives. While related derivatives such as extended trialkynyl-s-triazines [19], trialkenyl-s-triazines [20,21], or tris(2-thienyl)-s-triazines [22][23][24][25] have given rise to some 2PA investigations, to the best of our knowledge, only one recent theoretical paper deals specifically with the second-order NLO properties of molecules such as 3-X [26], and only one other single paper addresses the 2PA properties of an s-triazine derivative closely related to 7-H [27]. Thus, regarding extended triphenyl-s-triazines, essentially styryl-type analogues of 3-X have been investigated so far for their 2PA properties [28][29][30]. A general experimental and theoretical study focused on the 2PA properties of octupolar compounds such as 3-X and 7-X would therefore be timely. Furthermore, anticipating the well-known propensity of nitro substituents to poison fluorescence in 3-NO 2 , we also targeted a derivative such as 4, which similar to 3-CN, constitutes another example of a compound featuring electron-withdrawing arms [31]. The optical properties of this new triazine derivative might then be compared with those of its known isocyanurate analogue (5) [32]. Finally, given that the extended 4-fluorenyl derivative 6-NPh 2 exhibits the most promising optical properties for bio-imaging purposes among all triarylisocyanurates studied so far [33], the study of triazine analogues such as 7-X (X = H, NPh 2 ) was also required. Accordingly, in the following we will first start with the synthesis of the targeted molecules (3-X, 4, and 7-X) and study the 2PA properties of their emissive representatives via two-photon excited fluorescence (TPEF). Their optical properties of interest will then be discussed with the help of density functional theory (DFT) and time-dependent DFT (TD-DFT) calculations.

Synthesis and Characterisation
The desired 3-X, 4, and 7-X derivatives are structurally close to known s-triazine derivatives [15][16][17]27,[34][35][36][37][38][39][40]. The 3-X series corresponds to the simplest derivatives in which the 1,3,5-triphenyl-s-triazine core has been extended with a phenyl-alkynyl linker and terminated with X-groups of varying electron-donor/acceptor power. The second series (7-X) involves replacing the second phenyl unit with a 2-fluorenyl unit, which is luminescent in its own right. All these derivatives were obtained via Sonogashira coupling reactions (Scheme 2) from the known bromo [17,[41][42][43] or iodo [43] precursors 8-Y, after reaction with the corresponding aromatic alkynes. Most of these products required chromatographic purification. The use of 8-I instead of 8-Br allows generally using smother reaction conditions or leads to higher yields of isolated coupling products under similar conditions. Several 3-X derivatives have already been reported, such as 3-CN [17], 3-OMe [15], and 3-NPh 2 [18]. All other compounds were new and were fully characterised by usual techniques. However, synthesis of the nitro derivative 3-NO 2 , due solubility issues, was very problematic by such an approach and had to be attempted from the known triyne 8-C≡CH [16,18], itself obtained in two steps from 8-Br. Regarding IR-characterization, the in-plane ring stretches of the s-triazine ring resemble those of the phenyl ring and are not so characteristic of triphenyl-s-triazines [44], except perhaps for the fully symmetric stretch (only Raman-active), around 990 cm −1 [45,46].

Synthesis and Characterisation
The desired 3-X, 4, and 7-X derivatives are structurally close to known s-triazine derivatives [15][16][17]27,[34][35][36][37][38][39][40]. The 3-X series corresponds to the simplest derivatives in which the 1,3,5-triphenyl-s-triazine core has been extended with a phenyl-alkynyl linker and terminated with X-groups of varying electron-donor/acceptor power. The second series (7-X) involves replacing the second phenyl unit with a 2-fluorenyl unit, which is luminescent in its own right. All these derivatives were obtained via Sonogashira coupling reactions (Scheme 2) from the known bromo [17,[41][42][43] or iodo [43] precursors 8-Y, after reaction with the corresponding aromatic alkynes. Most of these products required chromatographic purification. The use of 8-I instead of 8-Br allows generally using smother reaction conditions or leads to higher yields of isolated coupling products under similar conditions. Several 3-X derivatives have already been reported, such as 3-CN [17], 3-OMe [15], and 3-NPh2 [18]. All other compounds were new and were fully characterised by usual techniques. However, synthesis of the nitro derivative 3-NO2, due solubility issues, was very problematic by such an approach and had to be attempted from the known triyne 8-C≡CH [16,18], itself obtained in two steps from 8-Br. Regarding IR-characterization, the in-plane ring stretches of the s-triazine ring resemble those of the phenyl ring and are not so characteristic of triphenyl-s-triazines [44], except perhaps for the fully symmetric stretch (only Raman-active), around 990 cm −1 [45,46].

One-and Two-Photon Absorption and Emission Studies
Scheme 2. Synthesis of 3-X, 4, and 7-X derivatives.

One-and Two-Photon Absorption and Emission Studies
The UV-Vis absorption spectra of the various triazine derivatives were recorded. Except for 3-CN and 4, most of the extended compounds (3-X and 7-X) absorbed significantly in the visible range and were strongly coloured (deep yellow in solution), with a lowest-energy absorption entailing significantly above 400 nm (Table 1). Upon progressing from 4 to 3-NPh 2 (corresponding to an increase in the electron-releasing nature of the para-substituents), a bathochromic shift of the first absorption was clearly observed (Figure 1). A similar trend could also be observed when progressing from 4 to 3-NO 2 .
For a given substituent in the 3-X series, the intensity of the lowest-energy absorption was comparable with that of 1-X or 2-X analogues. A higher-energy absorption (at ca. 270-325 nm) was also observed for all these derivatives. The latter became of comparable intensity to the one at the lowest energy for the compounds with the most electron-releasing substituents. Except for 3-NO 2 , 3-CN, and 4, all extended derivatives were significantly luminescent in CH 2 Cl 2 solutions (Table 1). Luminescence was maximal (Φ F = 0.80) for the fluorenyl derivative 7-H. The fluorescence of 3-X and 4 was comparable to that of their isocyanurate analogues (1-X and 5) but was overall slightly lower than that of their known 1,3,5-triphenylbenzene analogues (2-X). In contrast, the fluorescence quantum yield of 7-NPh 2 was roughly one-fourth of that previously found for 6-NPh 2 (Φ F = 0.78). In all cases, mirror-symmetry relationships between the first absorption and emission bands and energetic differences between their maxima (see ESI, Figure S7) suggest that the strongly absorbing state at the lowest energy was also the emitting state for all these compounds. Then, as indicated by the corresponding Stokes shifts, larger structural reorganisations and/or solvation energy changes apparently occurred for the compounds featuring the strongest electron-releasing (X = NPh 2 ) or electron-withdrawing (X = NO 2 ) substituents, i.e., 7085 cm −1 and 10,685 cm −1 in THF, respectively (ESI , Table S1).
In line with similar studies made for 1-X, 2-X and related derivatives [11,12,18,27], solvatochromic studies on 3-NPh 2 revealed a relatively solvent-insensitive absorption but a very solvent-sensitive emission for such a symmetrical D 3h molecule [47]. The shift to lower energy observed on proceeding to the most polar solvent suggests that the excited state was more polarised than the ground state. These results are consistent with the localisation of a (more polar) charge-transfer excited state on one arm of the compound (after relaxation) [14,33,[48][49][50].
The 2PA cross-sections of these derivatives were measured in the near-IR (NIR) range (λ = 700-1000 nm) through an investigation of their two-photon excited fluorescence (TPEF). The excitation was performed with femtosecond pulses from a Ti:sapphire laser ( Figure 2 and Table 2). Due to instrumental limitations (1PA below 350 nm for 3-CN), solubility issues (3-NO 2 ), or too weak luminescence (4), no TPEF maxima could be detected for several samples. A comparison of these 2PA bands with the 1PA bands for each compound reveals that the 2PA maxima were situated close to twice that of the 1PA maxima detected at the lowest energy in the UV range (see ESI, Figure S9), suggesting that the excited states at the origin of 2PA in the NIR were also active for 1PA or were close in energy to the first allowed 1PA state.
Photochem 2022, 2, FOR PEER REVIEW 4 The UV-Vis absorption spectra of the various triazine derivatives were recorded. Except for 3-CN and 4, most of the extended compounds (3-X and 7-X) absorbed significantly in the visible range and were strongly coloured (deep yellow in solution), with a lowestenergy absorption entailing significantly above 400 nm (Table 1). Upon progressing from 4 to 3-NPh2 (corresponding to an increase in the electron-releasing nature of the para-substituents), a bathochromic shift of the first absorption was clearly observed (Figure 1). A similar trend could also be observed when progressing from 4 to 3-NO2. For a given substituent in the 3-X series, the intensity of the lowest-energy absorption was comparable with that of 1-X or 2-X analogues. A higher-energy absorption (at ca. 270-325 nm) was also observed for all these derivatives. The latter became of comparable intensity to the one at the lowest energy for the compounds with the most electron-releasing substituents. Except for 3-NO2, 3-CN, and 4, all extended derivatives were significantly luminescent in CH2Cl2 solutions (Table 1). Luminescence was maximal (ΦF = 0.80) for the fluorenyl derivative 7-H. The fluorescence of 3-X and 4 was comparable to that of their isocyanurate analogues (1-X and 5) but was overall slightly lower than that of their known 1,3,5-triphenylbenzene analogues (2-X). In contrast, the fluorescence quantum yield of 7-NPh2 was roughly one-fourth of that previously found for 6-NPh2 (ΦF = 0.78). In all cases, mirror-symmetry relationships between the first absorption and emission bands and energetic differences between their maxima (see ESI, Figure S7) suggest that the strongly absorbing state at the lowest energy was also the emitting state for all these compounds. Then, as indicated by the corresponding Stokes shifts, larger structural reorganisations and/or solvation energy changes apparently occurred for the compounds featuring the strongest electron-releasing (X = NPh2) or electron-withdrawing (X = NO2) substituents, i.e., 7085 cm −1 and 10,685 cm −1 in THF, respectively (ESI , Table S1).       a Wavelength of the one-photon absorption maximum. b Wavelength of the two-photon absorption maximum. c 2PA cross-section at λ 2PA . d Specific cross-sections: figure-of-merit relevant for applications in optical limiting or nanofabrication [51]. e Effective number of π electrons [52]. f Two-photon brightness: figure-of-merit for imaging applications [51]. In these expressions, Φ F represents the luminescence quantum yield and MW the molecular weight. g Computed values using the SAOP functional (see text). h For exp. data, see also [11]. i For exp. data, see also [12] j For exp. data, see also [33]. k See: [53].
To understand better the structural dependence of these nonlinear absorptions on the π electrons, we divided their 2PA cross-sections by the square of the effective number of π electrons (N eff 2 ) [54]. These figures of merit (σ 2 /N eff 2 ) allow normalisation of the cross-sections and should permit a better comparison between them regardless of the different number of π electrons in each compound (see Section 3) [54]. The N eff values were derived according to the method initially proposed by Kuzyk [52], by decomposing the various compounds into a collection of "independently" conjugated π manifolds (ESI, Figure S18).

Density Functional Theory (DFT) Calculations
DFT calculations (see computational details in Section 4.4) were performed on the 3-X (X = NO 2 , CN, H, OMe, NMe 2 , NPh 2 ), 4 and 7-X (X = H, NPh 2 ) derivatives. Similar computations were also performed on the model compounds 9a-b, 10a-b, and 11a-b (Scheme 3) to specifically investigate the impact of the central core on the electronic structure of 1-X, 2-X, and 3-X analogues. The key results of these are disclosed in the next section.

Density Functional Theory (DFT) Calculations
DFT calculations (see computational details in Section 4.4) were performed on the 3-X (X = NO2, CN, H, OMe, NMe2, NPh2), 4 and 7-X (X = H, NPh2) derivatives. Similar computations were also performed on the model compounds 9a-b, 10a-b, and 11a-b (Scheme 3) to specifically investigate the impact of the central core on the electronic structure of 1-X, 2-X, and 3-X analogues. The key results of these are disclosed in the next section. All derivatives adopted a nearly coplanar conformation after geometrical optimisation relative to the central core upon geometry optimisation in CH2Cl2 (ESI, Table S1). As a result, in line with the available crystallographic data for 3-CN [17] or 3-NPh2, 3-X compounds appeared significantly more planar in solution than their 1-X or 2-X analogues (55-80°) [18]. Such planar conformations are likely favoured by weak intramolecular hy-Scheme 3. Model compounds studied for investigating the changes occurring in the FMO when changing the central core.
All derivatives adopted a nearly coplanar conformation after geometrical optimisation relative to the central core upon geometry optimisation in CH 2 Cl 2 (ESI, Table S1). As a result, in line with the available crystallographic data for 3-CN [17] or 3-NPh 2 , 3-X compounds appeared significantly more planar in solution than their 1-X or 2-X analogues (55-80 • ) [18]. Such planar conformations are likely favoured by weak intramolecular hydrogen bonds between ortho hydrogen atoms of the first aromatic ring and the triazine nitrogen atoms (with N· · · H-distances around 2.46 Å in optimised structures) [55,56]. Thus, the permanent dipole moment of 3-X and 4 compounds with axially symmetric X substituents was nearly zero (ESI , Table S2) [11,33].
The HOMO-LUMO gap in these derivatives (Table 3) was smaller than that in the corresponding isocyanurate (1-X) and triphenylbenzene (2-X) analogues. Starting from 3-CN, the calculations revealed that this gap progressively decreased when increasingly electronreleasing substituents were installed at their periphery, mirroring the trend observed experimentally for the lowest-lying intense absorptions of these compounds (Table 1). In line with experimental observations (Figure 1), replacing the peripheral 1,4-phenylene groups with a 2,7-fluorenyl one in 3-X, when a strong electron-releasing X group such as X = NPh 2 was present ( Figure 3), led only to a slight bathochromic shift of the first allowed absorption band computed at lowest energy for 7-X (e.g., 371 nm (f = 3.03) for 3-NPh 2 vs. 375 nm (f = 3.73) for 7-NPh 2 , using CAM-B3LYP). In contrast, computations predicted that a much more pronounced shift could be expected for a less electron-releasing substituent such as X = H (e.g., 329 nm (f = 2.46) for 3-H vs. 352 nm (f = 3.23) for 7-H, using the same functional), reminiscent of observations previously made between 1-X and 6-X [33]. Calculations made for 3-X compounds also indicated that, for X = CN (ESI; Figure S14) and more electron-withdrawing substituents, the direction of the photo-induced chargetransfer process was reversed, leading to a "umpolung" of the polarisation of the peripheral branches in the first allowed excited state, as already observed for the 2-X analogues [12]. Table 3. Calculated (CAM-B3LYP/6-31G* level in CH 2 Cl 2 ) HOMO-LUMO energy gaps in CH 2 Cl 2 for 1-X, 2-X, and 3-X. TD-DFT calculations qualitatively reproduced the energy trends observed for the most intense transitions at the lowest energy for 3-X, 4, and 7-X (Table 1). Transition energies were consistently overestimated when using the CAM-B3LYP functional and underestimated when using the MPW1PW91 functional [57], a better agreement with the experiment being obtained using the CAM-B3LYP functional for the low energy band [58]. However, overall, better results were obtained with MPW1PW91 when the transition moments were also considered (ESI, Figure S13). The nature of the dominant excitations underlying the first absorption band (ESI , Table S4) was the same using either the CAM-B3LYP or the MPW1PW91 functional. It does not change much for all the compounds presently considered. These excitations are a pair of (nearly) degenerate transitions that would correspond to the set of degenerate transitions towards an E-type excited state under strict C 3v symmetry (i.e., an E←A transition). In line with previous findings for 1-X and 2-X [11,12,33], the first absorption band, therefore, corresponds to a π→π* symmetric charge transfer (CT) between the peripheral arms and the central core.

Discussion
As for their 1-X and 2-X analogues [12,33], the first allowed electronic transition at lowest energy for most of the 3-X compounds corresponds to a π→π* transition, with a charge transfer (CT) character directed from the electron-rich peripheral arms towards the central electron-poor core (Figure 3). In line with previous investigations on 1-X [11,33] and 2-X [4,14] derivatives, the available solvatochromic studies on some of these molecules [18,27] strongly suggest localisation of the excited state on one branch in the first excited state after vibrational relaxation. As surmised at the start of this study, DFT calculations on compounds 9a-b, 10a-b, and 11a-b (Figure 4), confirm that the s-triazine central unit in 3-X compounds is more electron-withdrawing than the isocyanurate core in 1-X analogues. Indeed, regardless of the presence of three peripheral phenyl groups (11b) or not (11a), for a HOMO of comparable energy, the LUMO is always more stabilised in 11ab than in 10a-b, resulting in lower HOMO-LUMO gaps for the triazine-cored compounds relative to their isocyanurate analogues. Consistent with the acceptor character of the s-triazine ring, the CT occurs usually from the periphery towards the centre for all compounds, when X is electron-releasing [11,12,33]. The next and weaker absorption at higher energy observed for all these derivatives also corresponds to an allowed π-π* transition with a similar CT character but involves deeperlying occupied MOs. As a result, this transition is more arm-centred than that at the lowest energy. However, it does not correspond to an n-π* transition, as previously proposed [18]. Indeed, according to our calculations, transitions with n-π* character were much weaker (f < 0.1) and remained hidden beneath the dominant bands at lowest energy (especially for 7-X derivatives) but also at higher energy.
In line with our previous studies [53,59], the simulations of 2PA spectra were also carried out for selected compounds using the damped cubic response theory of Jensen et al. [60]. The calculations were performed using the SAOP functional, for sake of consistency with previous computations already reported on these compounds (see computational details in Section 4.5) [53]. The simulated 2PA spectra (ESI, Figure S17) revealed a first 2PA band for these compounds at lower energy than that experimentally determined with significantly higher 2PA cross-section values ( Table 2). While an overestimation of the real 2PA cross-sections by this method was expected [53], the experimental trends were generally qualitatively fairly well reproduced within a given family of compounds [59]. Based on these SAOP calculations, we speculate that the cross-section of the first 2PA band of 3-CN or 3-NO 2 , which could not be experimentally determined, should be lower than that of the other 3-X compounds. Interestingly, these computations revealed the existence of another and much larger 2PA peak at about 620 nm for 3-NPh 2 . This second 2PA band was hypsochromically shifted for 3-NMe 2 and 3-OMe.

Discussion
As for their 1-X and 2-X analogues [12,33], the first allowed electronic transition at lowest energy for most of the 3-X compounds corresponds to a π→π* transition, with a charge transfer (CT) character directed from the electron-rich peripheral arms towards the central electron-poor core (Figure 3). In line with previous investigations on 1-X [11,33] and 2-X [4,14] derivatives, the available solvatochromic studies on some of these molecules [18,27] strongly suggest localisation of the excited state on one branch in the first excited state after vibrational relaxation. As surmised at the start of this study, DFT calculations on  compounds 9a-b, 10a-b, and 11a-b (Figure 4), confirm that the s-triazine central unit in 3-X compounds is more electron-withdrawing than the isocyanurate core in 1-X analogues. Indeed, regardless of the presence of three peripheral phenyl groups (11b) or not (11a), for a HOMO of comparable energy, the LUMO is always more stabilised in 11a-b than in 10a-b, resulting in lower HOMO-LUMO gaps for the triazine-cored compounds relative to their isocyanurate analogues.  Furthermore, 3-X compounds were also significantly more fluorescent than their 1-X counterparts, a trend especially apparent with the weaker electron-releasing substituents (X = OMe, NMe2). As for 2-X derivatives, this might be related to the stronger transition moments observed for the lowest-energy 1PA absorptions in 3-X compared with 1-X analogues. Stronger transition moments should result in larger emission quantum yields, owing to the Strickler-Berg relationship [61], and in the absence of low-lying n-π* states able to efficiently quench the fluorescence in 3-X derivatives [11,62].
As previously observed for 1,3,5-triphenylbenzene-or isocyanurate-cored families of analogues, electron-releasing peripheral (X) substituents favour larger 2PA cross-sections (σ2) [11,12]. Thus, the cross-sections determined by TPEF for the 3-X derivatives were always significantly higher than those found for 1-X and 2-X ( Table 2). In accordance with Furthermore, 3-X compounds were also significantly more fluorescent than their 1-X counterparts, a trend especially apparent with the weaker electron-releasing substituents (X = OMe, NMe 2 ). As for 2-X derivatives, this might be related to the stronger transition moments observed for the lowest-energy 1PA absorptions in 3-X compared with 1-X analogues. Stronger transition moments should result in larger emission quantum yields, owing to the Strickler-Berg relationship [61], and in the absence of low-lying n-π* states able to efficiently quench the fluorescence in 3-X derivatives [11,62].
As previously observed for 1,3,5-triphenylbenzene-or isocyanurate-cored families of analogues, electron-releasing peripheral (X) substituents favour larger 2PA cross-sections (σ 2 ) [11,12]. Thus, the cross-sections determined by TPEF for the 3-X derivatives were always significantly higher than those found for 1-X and 2-X (Table 2). In accordance with this observation, the σ 2 values calculated for 3-NMe 2 and for quite all other 3-X derivatives (except for X = CN) were also higher than those previously computed for 1-NMe 2 or 2-NMe 2 ( Table 2) [53]. In spite of the fact that SAOP calculations always overestimate 2PA cross-sections [53,59], the present study therefore provides further evidence that the qualitative ordering of the experimental σ 2 values is generally retrieved.
Then, the 2PA cross-section presently measured for 7-H (710 GM) is significantly lower than the value of 2210 GM independently reported for its dihexyl analogue (13-H; Scheme 4) at nearly the same wavelength (730 nm vs. 740 nm) [27,63]. Actually, the value reported for 7-H is closer to that reported for compound 14 (395 GM at 790 nm), although the latter value was not recorded at the TPA peak maximum (779 nm) and possibly includes some RSA contribution [64]. Comparison of the photophysical data presently measured for 3-NPh2 with those reported for structurally related s-triazine derivatives such as 15 [19], 16 [22], and 17 [29] (Scheme 4) revealed larger fluorescence yields in chlorinated organic solvents for 3-NPh2 (69% for the latter compound vs. 52%, 51%, and 27%, respectively). Regarding 2PA, removing the first 1,4-phenyl ring significantly reduced the 2PA cross-section (1500 GM for 3-NPh2 vs. 910 GM reported for 15), while replacing it with a 2,5-thienyl unit apparently did not impact this figure significantly (1508 GM reported for 16) apart from slightly redshifting the 2PA maximum (850 nm vs. 830 nm for 3-NPh2). Then, changing the alkynyl linkers for alkenyl ones in 3-NPh2 seemed to reduce the 2PA cross-section. However, care should be given here since the value reported for 17 (495 GM) was obtained at a single wavelength (800 nm) which did not exactly match that of the 2PA maximum of this compound.
In terms of NLO activity per volume/mass unit, specific cross-sections [51] (Table 2) revealed that 3-X derivatives are much more active than 1-X and 2-X ( Table 2). More precisely, among all triazine derivatives investigated, 3-NPh2 appears to be the best-suited two-photon absorber for elaborating molecular materials. As confirmed by other relevant figures of merit (σ2/(Neff) 2 ) [54], these compounds optimise 2PA for a given number of effective electrons (Neff), a feature possibly connected to their enforced planarity [65] but also certainly to their more pronounced (multi)polarity, compared to those of their 1-X and 2-X analogues. Both of these features ultimately translate into lower HOMO-LUMO gaps for 3-X derivatives. Based on a simple perturbational approach, such a reduction in HOMO-LUMO gaps when progressing from 1-X or 2-X to 3-X derivatives is expected to favour 2PA (at least at wavelengths above that of the lowest-energy absorption) [9]. Finally, the near-perfect spectral overlap between the 1PA spectrum plotted at half wavelength and the 2PA spectrum for each of these octupolar compounds (ESI, Figure S9) suggests a near degeneracy of the A and E excited states, in line with a small electronic/excitonic coupling between the peripheral branches [5,14,66]. As a result, the contribution of the octupolar symmetry to the 2PA improvement in these compounds should not be determining [67,68].
In terms of applied uses of these 2PA absorbers, both the significant bathochromic shift of 1PA and 2PA peaks and the larger intensity of the first absorption band result in Comparison of the photophysical data presently measured for 3-NPh 2 with those reported for structurally related s-triazine derivatives such as 15 [19], 16 [22], and 17 [29] (Scheme 4) revealed larger fluorescence yields in chlorinated organic solvents for 3-NPh 2 (69% for the latter compound vs. 52%, 51%, and 27%, respectively). Regarding 2PA, removing the first 1,4-phenyl ring significantly reduced the 2PA cross-section (1500 GM for 3-NPh 2 vs. 910 GM reported for 15), while replacing it with a 2,5-thienyl unit apparently did not impact this figure significantly (1508 GM reported for 16) apart from slightly redshifting the 2PA maximum (850 nm vs. 830 nm for 3-NPh 2 ). Then, changing the alkynyl linkers for alkenyl ones in 3-NPh 2 seemed to reduce the 2PA cross-section. However, care should be given here since the value reported for 17 (495 GM) was obtained at a single wavelength (800 nm) which did not exactly match that of the 2PA maximum of this compound.
In terms of NLO activity per volume/mass unit, specific cross-sections [51] (Table 2) revealed that 3-X derivatives are much more active than 1-X and 2-X ( Table 2). More precisely, among all triazine derivatives investigated, 3-NPh 2 appears to be the best-suited two-photon absorber for elaborating molecular materials. As confirmed by other relevant figures of merit (σ 2 /(N eff ) 2 ) [54], these compounds optimise 2PA for a given number of effective electrons (N eff ), a feature possibly connected to their enforced planarity [65] but also certainly to their more pronounced (multi)polarity, compared to those of their 1-X and 2-X analogues. Both of these features ultimately translate into lower HOMO-LUMO gaps for 3-X derivatives. Based on a simple perturbational approach, such a reduction in HOMO-LUMO gaps when progressing from 1-X or 2-X to 3-X derivatives is expected to favour 2PA (at least at wavelengths above that of the lowest-energy absorption) [9]. Finally, the near-perfect spectral overlap between the 1PA spectrum plotted at half wavelength and the 2PA spectrum for each of these octupolar compounds (ESI, Figure S9) suggests a near degeneracy of the A and E excited states, in line with a small electronic/excitonic coupling between the peripheral branches [5,14,66]. As a result, the contribution of the octupolar symmetry to the 2PA improvement in these compounds should not be determining [67,68].
In terms of applied uses of these 2PA absorbers, both the significant bathochromic shift of 1PA and 2PA peaks and the larger intensity of the first absorption band result in a diminished transparency relative to 1-X analogues, making them less favourable for selected NLO applications [6] such as optical limiting [69] or second harmonic generation in the visible range [70]. However, both the increased two-photon brightness [51] and the bathochromically-shifted 2PA maxima for the various 3-X compounds ( Table 2) points to a larger potential for two-photon fluorescence imaging. Such uses were already reported for structurally different s-triazine derivatives presenting much poorer figures of merit than 3-X [71], but also for nanoparticles obtained from extended analogues of 17, and giving rise to aggregation-induced emission (AIE) in water-THF mixtures [30]. Compared with 1-X or 2-X derivatives, the best candidate for fluorescence imaging is 3-NPh 2 . This compound has indeed a two-photon brightness far above those of its known triphenylbenzene and isocyanurate analogues (1-NPh 2 and 2-NPh 2 , respectively) and a 2PA peak at 820 nm, i.e., also significantly bathochromically shifted relative to them (Table 2) [6,9]. In contrast, contrary to the trend previously observed with triarylisocyanurates derivatives [33], the tris(2-fluorenyl)triazine 7-NPh 2 presents neither a higher 2PA cross-section value nor a larger two-photon brightness than 3-NPh 2 .

General
All manipulations were carried out under an inert atmosphere of argon with dried and freshly distilled solvents [72]. Transmittance FTIR spectra were recorded using a Perkin Elmer Spectrum 100 spectrometer (Waltham, MA, USA) equipped with a universal ATR sampling accessory (400-4000 cm −1 ). Raman spectra of the solid samples were obtained by diffuse scattering on a Bruker IFS 28 spectrometer and recorded in the 400-2500 cm −1 range (Stokes emission), with a laser excitation source at 1064 nm or on a Raman LabRAM HR 800 spectrometer with a laser excitation source at 785 nm. Nuclear magnetic resonance spectroscopy was performed using a Bruker AV-300 (300 MHz for 1 H, 75 MHz for 13 C) a Bruker AV-400 (400 MHz for 1 H, 101 MHz for 13 C) or a Bruker AV-500 (500 MHz for 1 H, 101 MHz for 13 C) spectrometers at ambient temperature. 1 H and 13 C spectra were calibrated using residual solvent peaks [73,74]. MS analyses were performed at the "Centre Regional de Mesures Physiques de l'Ouest" (CRMPO, Université de Rennes) on high resolution Bruker Maxis 4G or Thermo Fisher Q-Extractive Spectrometers. Elemental analyses were also performed at CRMPO. Chromatographic separations (rapid suction filtration (RSF), column chromatography, or flash chromatography) were performed on Merck silica gel (40-63 µm), Aldrich basic Alumina (act 1), or Aldrich neutral Alumina (act 1), with the eluants indicated. Commercial reagents and (pre/co)catalysts were used as received. The trisbromo s-triazine precursor 8-Br was synthesised using a procedure inspired by the literature [42], which was subsequently extended to the known trisiodo analogue 8-I (ESI). The triyne 8-C≡CH was also obtained following modifications of the reported literature procedure (ESI) [16]. The known triazine derivatives 3-CN [17], 3-OMe [15], and 3-NPh 2 [18] were obtained according to published procedures and fully characterised (ESI), as were the required alkynes [33,[75][76][77][78].

2,4,6-tris{4 -[(4 -dimethylamino)-2 -phenylethynyl]phenyl}-1,3,5-triazine (3-NMe 2 ).
A dry Schlenk flask was charged with 8-I (203 mg, 0.295 mmol), 4-ethynyl-N,N-dimethylan iline (211 mg, 1.453 mmol, 5 eq.), Pd(PPh 3 ) 4 (35 mg, 0.030 mmol, 10 mol%), and CuI (11 mg, 0.058 mmol, 20 mol%), and then degassed (4 × vacuum/argon cycles). A degassed mixture of DMF/ i Pr 2 NH (2/1 mixture, 30 mL) was added using a cannula, and the flask was sealed and heated at 50 • C for 2 days. The solvent was removed in vacuo and the residue triturated with Et 2 O to remove the unreacted starting material. The residue was dissolved in a mix-ture of pentane/CH 2 Cl 2 /NEt 3 (200 mL/200 mL/20 mL) and filtered through a short plug (2 cm × 2 cm) of neutral alumina (deactivated with NEt 3 ). The bright orange filtrate was evaporated under reduced pressure, giving the title product as an orange powder (141 mg, 65%). MP: > 150 • C (dec.). . The compound was isolated using a workup similar than that described above from 8-C≡CH (110 mg, 0.289 mmol) and 4-bromonitrobenzene in excess (293 mg, 1.45 mmol, 5 eq.) with Pd(PPh 3 ) 4 (10 mol%) and CuI (20 mol%) as catalysts in an DMF/NEt 3 mixture at 75 • C for 2.5 days. The volatiles were removed in vacuo and the residue triturated with small portions of CH 2 Cl 2 and Et 2 O to remove the unreacted 4-bromonitrobenzene and other soluble byproducts that formed during the reaction. The residue (ca. 90 mgs) was then dissolved in THF and filtered through a short plug (2 cm × 2 cm) and recrystallised several times to eventually yield pure fractions of yellow product that were used for characterisations. then degassed (4 × vacuum/argon cycles). The flask was wrapped in foil before degassed DMF (20 mL) and NEt 3 (9 mL) were added using a cannula. The reaction mixture was then stirred at 25 • C for 2 days. The solvent was removed in vacuo, and the residue was dissolved in CH 2 Cl 2 (100 mL), washed with water (3 × 50 mL), and dried (MgSO 4 ). The solvent was removed under reduced pressure, and the crude material purified using column chromatography (neutral alumina, 5 cm × 6 cm, eluting with a methanol/CH 2 Cl 2 [2:98] mixture). The resulting solid was then precipitated from methanol/CH 2 Cl 2 , with the slow evaporation of the CH 2 Cl 2 , and collected on a glass frit, washed with methanol, and dried under reduced pressure (high vacuum) for 12 h giving the title product as an off-white coloured powder (261 mg, 75%). MP: 310 • C (dec.). then degassed (4 × vacuum/argon cycles). A mixture of dry, degassed DMF/ i Pr 2 NH (1/1 mixture, 30 mL) was added using a cannula. The flask was sealed and the reaction mixture was heated at 70 • C for 3 days. The solvent was removed in vacuo, and the residue was dissolved in CH 2 Cl 2 (100 mL), washed with water (3 × 30 mL) and dried (MgSO 4 ). The solvent was removed under reduced pressure, and the crude material was purified using flash column chromatography (silica gel, 3.5 cm × 15 cm, eluting with an hexanes/CH 2 Cl 2 [9:1] mixture), giving the title product as a pale yellow solid (198 mg, 89%). MP: 146 • C. A mixture of dry, degassed DMF/ i Pr 2 NH (2/1 mixture, 20 mL) was added using a cannula. The flask was sealed, and the reaction mixture was heated at 70 • C for 2.5 days. The solvent was removed in vacuo, and the residue was dissolved in CH 2 Cl 2 (70 mL), washed with water (2 × 20 mL), and dried (MgSO 4 ). The solvent was removed under reduced pressure, and the crude material was purified using flash column chromatography (neutral alumina deactivated with NEt 3 , eluting with mixture of ether/hexanes/NEt 3 gradient from 50/450/5 mL up to 245/250/5 mL mixture), giving the title product as a bright yellow waxy solid (

Fluorescence Measurements
All photophysical measurements were performed with freshly prepared air-equilibrated CH 2 Cl 2 (or THF) solutions (HPLC grade) at room temperature (298 K). UV-Vis absorption spectra were recorded on dilute solutions (ca. 10 −5 M) by using a Jasco V-570 spectrophotometer (Mary's Court, Easton MD, USA). The samples used to make the solutions were freshly recrystallised or thoroughly washed with cooled ether/pentane prior to the measurements to remove any organic impurity. Steady-state fluorescence studies were performed in diluted air-equilibrated solutions in quartz cells of 1 cm path length (ca. fluorescence quantum yield of each compound was calculated using the integral of the fully corrected emission spectra relative to a standard, quinine bisulphate (QBS, λ ex = 346 nm, Φ F = 0.546) [79,80]. UV-Vis absorption spectra used for the calculation of the fluorescence quantum yields were recorded using a double-beam Jasco V-570 spectrometer.

Two-Photon Absorption Experiments
To span the 790-920 nm range, an Nd:YLF-pumped Ti:sapphire oscillator (Chameleon Ultra, Coherent) was used, generating 140 fs pulses at an 80 MHz rate. The excitation power was controlled using neutral density filters of varying optical density mounted in a computer-controlled filter wheel. After a fivefold expansion through two achromatic doublets, the laser beam was focused with a microscope objective (10×, NA 0.25, Olympus, Shinjuku, Tokyo, Japan) into a standard 1 cm absorption cuvette containing the sample. The applied average laser power arriving at the sample was typically between 0.5 and 40 mW, leading to a time-averaged light flux in the focal volume on the order of 0.1-10 mW/mm 2 . The fluorescence intensity was measured at several excitation powers in this range owing to the filter wheel. For each sample and each wavelength, the quadratic dependence of the fluorescence intensity (F) on the excitation intensity (P), i.e., the linear dependence of F on P 2 was systematically checked (see ESI, Figure S8). The fluorescence from the sample was collected in epifluorescence mode, using the microscope objective, and reflected by a dichroic mirror (Chroma Technology Corporation, Bellows Falls, VT, USA; "red" filter set: 780dxcrr). This made it possible to avoid the inner filter effects related to the high dye concentrations used (10 −4 M) by focusing the laser near the cuvette window. Residual excitation light was removed using a barrier filter (Chroma Technology; "red": e750sp-2p). The fluorescence was coupled into a 600 µm multimode fibre with an achromatic doublet. The fibre was connected to a compact CCD-based spectrometer (BTC112-E, B&WTek, Newark DE, USA), which measured the two-photon excited emission spectrum. The emission spectra were corrected for the wavelength dependence of the detection efficiency using correction factors established through the measurement of reference compounds having known fluorescence emission spectra. Briefly, the setup allowed for the recording of corrected fluorescence emission spectra under multiphoton excitation at variable excitation power and wavelengths. Further, 2PA cross-sections (σ 2 ) were determined from the twophoton excited fluorescence (TPEF) cross-sections (σ 2 ·Φ F ) and the fluorescence emission quantum yield (Φ F ). TPEF cross-sections of 10 −4 M dichloromethane solutions were measured relative to fluorescein in 0.01 M aqueous NaOH using the well-established method described by Xu and Webb [81] and the appropriate solvent-related refractive index corrections [82]. To check the absence of aggregation, UV-Vis absorption spectra of the dyes were recorded at this concentration in cells of 1 mm pathlength and compared with those obtained with diluted solutions in cells of 1 cm pathlength.

DFT Computations
DFT and TD-DFT calculations reported in this study were performed using the Gaus-sian09 [83] program. The geometries of all compounds were optimised without symmetry constraints using the CAM-B3LYP [84] or the MPW1PW91 [57] functionals and the 6-31G* basis set. The solvent effects were taken into account by means of the polarizable continuum model (PCM) [85]. Calculations of the normal modes of vibration were carried out to confirm the true minima character of the optimised geometries. TD-DFT calculations were performed at the same level of theory using the previously optimised geometries. A value of 0.1 was considered for the damping parameter when simulating the electronic spectra (ESI). Swizard [86] was used to plot the simulated spectra, and GausView [87] was used for the MO plots. Subsequently, the 2PA properties were calculated for selected compounds using the SAOP model potential [88,89] (statistical average of orbital model exchange-correlation potential) using the damped cubic response theory module of Jensen et al. [60] implemented in the ADF program package [90], considering the different molecules in the gas phase with the optimised geometries obtained in solution (CH 2 Cl 2 ). The lifetime of the electron-ically excited states was included in the theory using a damping parameter of 0.0034 au (∼0.1 eV ∼ 800 cm −1 ) value, which was found suitable for 2PA computations [91,92]. The 2PA cross-section σ 2 was obtained from the imaginary part of the third-order hyperpolarizability γ using ad hoc expressions (see ESI) [59]. Cross-section (σ 2 ) values are usually given in Göppert-Mayer units (1 GM = 10 −50 cm 4 ·s·photon −1 ), so we first evaluated them in atomic units and then multiplied them by (0.529177 × 10 −8 cm/a.u.) 4 × (2.418884 × 10 −17 s/a.u.) to obtain values in the conventional units (cm 4 ·s·photon −1 ).

Conclusions
The two-photon absorption properties of several 2,4,6-triaryl-s-triazines derivatives such as 3-X (X = NO 2 , CN, H, OMe, NMe 2 , NPh 2 ), 4, and 7-X (X = H, NPh 2 ) were studied. After characterising the new members of these families (X = 3-NO 2 , 3-NMe 2 , 4, 7-H, and 7-NPh 2 ), we uncovered evidence that significant 2PA occurs at the near-IR edge (730-820 nm) for most of them. Similar to what was previously shown for 1,3,5-triarylbenzene or N,N ,Ntriarylisocyanurate analogues, this nonlinear absorption process apparently populates the π-π* charge-transfer excited state(s) at the lowest energy. As confirmed by DFT calculations, the charge transfer occurring during 2PA corresponds to a shift of electron density from the periphery (arms) toward the central ring, except for strongly electron-withdrawing X substituents (X = NO 2 and CN) for which the charge transfer direction reverts. Based on our calculations, compared with 1-X and 2-X, the improved 2PA properties of 3-X derivatives find their origin in the quasi-planar conformation adopted by these molecules in solution and also in the increased polarisation of their π manifold, two features directly resulting from the presence of the central s-triazine unit. The latter favours π-π interactions between peripheral arms and the central core, resulting in a smaller HOMO-LUMO gap for these molecules. We also showed that all 3-X and 7-X derivatives featuring neutral to electronreleasing peripheral groups were sufficiently emissive for performing two-photon imaging purposes in solution (Φ F > 0.2). Actually, in line with extent data for other s-triazines in the literature, they presented larger two-photon brightness than all 1-X, 2-X, and 6-X fluorophores investigated so far. This statement, coupled to the fact that their first 2PA peak is always bathochromically shifted relative to that of their 1,3,5-triarylbenzene or N,N ,N -triaryl isocyanurate analogues, points to a larger potential for 3-X derivatives, with 3-NMe 2 and 3-NPh 2 being the most promising molecules in this respect. Provided that such compounds can now be made water-soluble without changing the observed trends, they should give rise to appealing new two-photon dyes for fluorescence bioimaging. Research along these lines is in progress.