Rafał Flamholca,
Damian Plażuka,
Janusz Zakrzewski*a,
Rémi Métivier*b,
Keitaro Nakatanib,
Anna Makalc and
Krzysztof Woźniakc
aDepartment of Organic Chemistry, Faculty of Chemistry, University of Łódź, Tamka 12, 91-403 Łódź, Poland. E-mail: janzak@uni.lodz.pl
bPPSM, ENS Cachan, CNRS, UniverSud, 61 av President Wilson, 94230 Cachan, France. E-mail: metivier@ppsm.ens-cachan.fr
cDepartment of Chemistry, Warsaw University, Pasteura 1, 02-093 Warszawa, Poland
First published on 15th July 2014
Friedel–Crafts acylation of pyrene with alkynoic acids in the presence of trifluoroacetic anhydride and triflic acid constitutes a direct and efficient route to 1-pyrenyl ynones. These compounds in chloroform solution emit fluorescence at longer wavelengths, with higher quantum yields and longer lifetimes than a typical saturated acylpyrene derivative, 1-acetylpyrene. Moreover, in contrast to 1-acetylpyrene, they are moderate solid-state emitters. Comparative DFT studies revealed strong stabilization of the LUMOs of 1-pyrenyl ynones in comparison to the LUMO of 1-acetylpyrene. The single-crystal X-ray structure of 1-(pyren-1-yl)but-2-yn-1-one showed π-interactions of pyrenyl moieties in the crystal lattice. Investigations of the solid-state fluorescence of this compound revealed emission from long-lived excited states, including excimer species.
In a continuation of our research programme, which has focused on the use of functionalised carboxylic acids as acylating agents in Friedel–Crafts reaction, we recently elaborated an efficient method of synthesis of ferrocenyl ynones via a direct reaction of ferrocene with alkynoic acids in the presence of trifluoroacetic anhydride (TFAA) and trifluoromethanesulfonic acid (TfOH).24 Herein we report that this approach may be used for simple and efficient synthesis of 1-pyrenyl ynones. We have also studied the fluorescence properties of these compounds which were compared to those of the simplest pyrenyl alkyl ketone, 1-acetylpyrene. Unexpectedly, we found that, in contrast to the latter compound, the synthesized 1-pyrenyl ynones display fluorescence not only in solution but also in the solid state. An X-ray diffraction study (including topological analysis of experimental charge density) performed for 1-(pyren-1-yl)but-2-yn-1-one revealed face-to-face π-stacking of the pyrene moieties in the crystal, thus suggesting that emission may originate from solid-state excimers. Finally, we carried out comparative DFT calculations on this compound and 1-acetylpyrene.
The Friedel–Crafts reaction of pyrene with 2-alkynoic acids (Scheme 1) was carried out under conditions described earlier for ferrocene.24 The isolated yields of compounds 1a–c were in the range of 69–74%. Similarly, as in the case of ferrocene, the reaction of pyrene with propiolic acid (RH) led to an intractable reaction mixture. However, compound 1d was prepared in an almost quantitative yield (99%) by fluoride-promoted desilylation of 1c.
Structures of synthesized ynones 1a–d were confirmed by spectroscopic and elemental analysis data (see ESI†). The simplest compound of this series, 1d, was already reported in the literature. It was synthesised 40–44% overall in a reaction of pyrene-1-carboxaldehyde with ethynyl magnesium bromide22 (or TMS-ethynyl magnesium bromide followed by desilylation23), and subsequent oxidation of the alcohol formed with Jones reagent. Compared to those methods our synthesis was simpler (using pyrene as the starting material) and more efficient (71% overall yield).
Fig. 1 ORTEP representation of 1a with an atom numbering scheme. Atomic displacement parameters are drawn at 50% probability level. |
In the experimental structure of 1a the pyrenyl moiety is not ideally planar, but slightly bent along its longer (C7 → C14) axis. The angle between the plane of ring C5, C6, C7, C8, C9, and C16 (ring 1) and that of ring C1, C2, C12, C13, C14, C15 (ring 2) is 3.7(3) degrees. The propynoyl substituent is twisted out of the plane of adjacent ring 2, as indicated by the C2–C1–C17–C18 and C14–C1–C17–O1 torsion angles which were significantly different from 180° (172.37(7) degrees and C14 C1 C17 O1 168.68(9) degrees, respectively). The angle between the plane of ring 2 and the plane of the carbonyl moiety was 9.9(5) degrees. This conformation enables the formation of two weak intermolecular C–H⋯O hydrogen bonds in the crystal lattice: C4–H4⋯O1_i O1–H4 where the distance is 2.552(3) Å and C20–H20A⋯O1_ii O1–H20A where the distance is 2.579(3) Å (where (i) denotes the following symmetry transformation: 1 − x, 2 − y, 1 − z and (ii) denotes the symmetry transformation: x, 1 1/2 − y, 1/2 + z) (Fig. 2).
The tilt of the carbonyl group out of the ring 2 plane does not prevent the formation of a weak intramolecular C3–H3⋯O1 hydrogen bond, with an O1⋯H3 distance of only 2.156(1) Å. Viewed along the crystallographic [001] direction, the crystal packing of 1a shows a characteristic herringbone motif (Fig. 3a) with distinct layers of molecules stacked along the [100] direction. Within the layers the carbonyl moieties are oriented almost parallel to the [001] direction; in consecutive layers the carbonyl groups point in opposite directions (i.e. the carbonyl moieties in one layer are almost parallel to [001] direction, while the carbonyl moieties from the next layer are almost antiparallel to this direction). The angle between the C17–O1 carbonyl bond and the crystallographic [001] direction is 25.8(5) degrees (for a molecule in the alternative layer this would be 205.8(5) degrees).
The crystal structure of 1a can be considered as composed of layers of molecules with parallel dipole and transition moments; with the directions of these moments alternating between layers. Each layer can be further viewed as composed of strands of molecules, packed along the [001] direction.
Molecules from consecutive layers along the [100] direction (Fig. 3b), related by crystallographic centres of inversion, are also involved in π-stacking interactions, thus building columns of symmetry-related molecules along [100] (Fig. 3a). Due to the ‘bend’ in the pyrene moiety, the closest contacts of an independent molecule (I) and the symmetry-related molecule II (above) are different from the closest contacts of an independent molecule (I) and the symmetry-related molecule III (below) (Fig. 4). For the former the closest contacts are C2_I–C12_II (3.377(3) Å) and C14_I–C16_II (3.419(3) Å), while for the latter the closest contacts are C1_I–C9_III (3.361(3) Å) and C15_I–C15_III (3.433(3) Å).
Interlayer interactions are stabilised by a weak C–H⋯O hydrogen bond: C4–H4⋯O1_i, while the separate layers are stabilised by a weak C–H⋯O hydrogen bond: C20–H20A⋯O1.
Because the scattering power of the 1a crystals was good, aspherical atomic scattering factors could be applied in structure refinement. Therefore, we could also perform a preliminary topological analysis of experimental charge density for this compound (see ESI†). It supports the structural analysis, thus confirming the presence and significance of all of the already mentioned C–H⋯O interactions. The strongest hydrogen bond that is present in the crystals of 1a is the intramolecular C3–H3⋯O1 bond, according to electron density and energy density criteria. A ring critical point was found within the C1–C2–C3–H3–O1–C17 ring, closed by this bond, thus confirming the significance of this interaction according to Koch and Popelier criteria.31 This analysis also demonstrated bond paths for the π-stacking interactions between symmetry-related molecules of 1a, which may support the ‘excimer’ hypothesis (vide infra). (Table 1)
Interaction | ρ (rBCP) [e Å−3] | Lap (rBCP) [e Å−5] | G (rBCP) [H a0−3] | V (rBCP) [H a0−3] | H (rBCP) [H a0−3] | G (rBCP)/ρ (rBCP) [H e−1] | H (rBCP)/ρ (rBCP) [H e−1] | |V (rBCP)|/G (rBCP) |
---|---|---|---|---|---|---|---|---|
O1⋯H3 | 0.110 (5) | 2.076 (4) | 0.017 | −0.013 | 0.004 | 1.043 | 0.245 | 0.765 |
H4⋯i_O1 | 0.056 (5) | 0.684 (3) | 0.006 | −0.004 | 0.001 | 0.723 | 0.121 | 0.667 |
H20A⋯ii_O1 | 0.045 (4) | 0.745 (3) | 0.006 | −0.004 | 0.002 | 0.900 | 0.300 | 0.667 |
C2⋯iii_C12 | 0.031 (1) | 0.295 (1) | 0.002 | −0.002 | 0.001 | 0.435 | 0.218 | 1.000 |
C14⋯iii_C16 | 0.029 (1) | 0.272 (1) | 0.002 | −0.002 | 0.001 | 0.465 | 0.233 | 1.000 |
C1⋯iv_C9 | 0.036(2) | 0.314 (1) | 0.003 | −0.002 | 0.001 | 0.562 | 0.187 | 0.667 |
C15⋯iv_C15 | 0.021 (2) | 0.303 (1) | 0.002 | −0.001 | 0.001 | 0.643 | 0.321 | 0.500 |
The molecule of 1a in the crystal is strongly polarized with a negative charge of over 0.5e localised on the butynoyl group and an equivalent positive charge residing on the pyrene moiety. This results in a significant molecular dipole moment, oriented in between the C1–C17 and C17–O1 bond axes and coplanar with the pyrene moiety. According to the DFT B3PW91 calculations, an isolated molecule of 1a has the dipole moment of 3.8 D, while the dipole moment obtained from the experimental charge density model is almost 4 times larger (15.8 D). It must be stressed that the absolute value of the molecular dipole moment in the crystal cannot be reliably derived from the current experimental charge density model, and that the value only indicates a tendency for increased polarisation of the molecule in the crystalline environment. Quantum chemical calculations in the periodic lattice performed at the B3LYP level of theory with a 6-31+g(d) basis set yielded an optimised geometry that was identical to the geometry from X-ray diffraction within an experimental error. These calculations confirm enhancement of the molecular dipole moment in the crystal lattice. The resulting dipole moment has a direction that is in agreement with the results of experimental charge density analysis and the results of theoretical calculations for an isolated molecule and a value of 4.7 D. The dipole moment vector is almost perpendicular to the [100] crystallographic direction (∼86°) and coplanar with the pyrene moiety.
Fig. 5 Normalised absorption (left) and fluorescence (right) spectra of compounds 1a–d and AcPyr in CHCl3 (λexc = 390 nm for 1a–d and 330 nm for AcPyr). |
1a | 1b | 1c | 1d | AcPyr | |
---|---|---|---|---|---|
λabs (nm) (CHCl3) | 408 | 418 | 416 | 415 | 391 |
εmax (M−1 cm) (CHCl3) | 15100 | 45500 | 18800 | 24800 | 15400 |
λem (nm) (CHCl3) | 449 | 471 | 462 | 460 | 409 |
ΦF (CHCl3) | 0.02 | 0.06 | 0.05 | 0.07 | <0.005 |
λem (nm) (solid) | 569 | 530 | 550 | 565 | 486 |
ΦF (solid) | 0.13 | 0.04 | 0.12 | 0.07 | — |
τ1 (ns)/contribution (CHCl3) | 0.11/0.67 0.29/0.29; 0.73/0.04 | 0.24/0.44; 0.77/0.56 | 0.20/0.650.60/0.35 | 0.24/0.250.68/0.75 | 0.06/0.890.27/0.11 |
The spectra reveal significant (up to ∼60 nm) bathochromic shifts of the bands of 1a–d in comparison to those of AcPyr. This suggests that the alkynoyl substituents are more efficiently conjugated with the pyrenyl moiety than the acetyl group. Furthermore, the absorption bands of 1a–d enter the visible region and these compounds can be excited with violet light. All of the investigated ketones are practically non-fluorescent in a nonpolar solvent, hexane (see ESI†). In a medium polarity solvent, chloroform, compounds 1a–d showed fluorescence at 449–471 nm with quantum yields in the range of 0.02–0.07, whereas AcPyr emitted weakly at 409 nm (quantum yield was lower than 0.005). In a more polar aprotic solvent, DMSO, all 1a–d were emissive, whereas in a polar but hydroxylic solvent, methanol, only 1a was a strong emitter. Surprisingly, 1a–d were stronger emitters in chloroform than in methanol, whereas the opposite effect was observed for AcPyr. Time-resolved fluorescence investigations of 1a–d and AcPyr were performed in chloroform solutions and revealed multiexponential decays in all cases (Fig. 6a, Table 2). A fast decay was observed for AcPyr, intermediate behavior was observed for 1a and 1c, whereas much slower decays were recorded for 1b and 1d. Three sets of decay time constants could be distinguished: a fast contribution (τ1 < 0.15 ns), an intermediate contribution (τ2 = 0.2–0.3 ns), and a slow contribution (τ3 > 0.5 ns).
These three contributions may be related to three populations of conformers having different relaxation times and different emission efficiencies. The fraction of intensity related to the slowest component τ3 is very much variable depending on the compound. It is absent for AcPyr, represents a small proportion for 1a (ϕ3 = 0.15) and its contribution becomes predominant for compounds 1b (ϕ3 = 0.80), 1c (ϕ3 = 0.59), and 1d (ϕ3 = 0.88). This fraction of intensity is well-correlated to the overall fluorescence quantum yields of compounds AcPyr and 1a–d (Fig. 6b). Therefore, we could conclude that the slowest decay time corresponds to a conformer which is much more fluorescent than the others. Consequently, the fluorescence quantum yields measured in CHCl3 reflect the various proportions of the different conformers, which are variable from one compound to another.
Fig. 7 Photographs of 1a–d and AcPyr in the solid state under visible light (top row) und under 254 nm UV light (bottom row) illumination. |
Fig. 8 (Left) normalized excitation and (right) emission spectra of compounds 1a–d recorded in the solid state (powder inserted in an integration sphere). λexcit = 430 nm for 1a and 460 nm for 1b–d. |
Fluorescence time-resolved experiments of 1a were also performed in the solid state under ambient atmosphere. Fluorescence decay curves monitored at three different emission wavelengths (λem = 540, 590 and 640 nm) are shown in Fig. 9.
Global analysis was successfully applied to the three decays with multiexponential fitting and revealed three common components: two long decay times, τ1 = 13.2 ns and τ2 = 27.4 ns (which represented more than 99% of the fraction of intensity) and a short component, τ3 ∼ 0.6–1.2 ns, which appeared either as a decay time (for λem = 540 nm) or as a rise-time (for λem = 590 and 640 nm), as shown in Fig. 9. Such a short decay time in the blue-edge of the spectrum corresponding to a short rise-time in the red part of the spectrum is a typical signature of excimer formation which occurs within the crystal with a fast kinetic rate. This mechanism is rather consistent with the face-to-face orientation of the pyrene moieties in the crystal lattice, as revealed by X-ray diffraction (vide supra).
It is generally believed that such an aggregation leads to non-emissive species (H-aggregates). However, some examples of emissive H-aggregates have been reported.32–34 In such aggregates, emission may arise from restriction of intramolecular rotation (RIR), blocking of nonradiative decay channels and from the formation of emissive solid-state excimers (meaning not only “excited dimers” but also “excited oligomers”). Consequently, 1a–d emitted in the solid-state as monomers (a green emission), but they also interacted very quickly in the excited state and led to strongly emissive excimer species (a red emission), and then following a very slow decay rate. This situation is different from the one observed in solution, where only fast monomer decays were observed.
It should be emphasized that solid-state fluorescence, which is essential for various industrial applications, is still a relatively rare phenomenon because of the ubiquitous aggregation-caused quenching (ACQ) effect. Solid-state fluorophores exhibiting the aggregation-induced emission enhancement (AIEE) effect were discovered only very recently.35–37 Moreover, such fluorophores bearing a flat π-conjugated pyrene moiety are relatively rare.38–40 Strong yellow solid-state emission of 1a–d, with long fluorescence decays, makes them promising candidates for the design of luminescent materials and devices.
First we scanned the potential energy vs. the dihedral angle between the pyrenyl moiety and the CO group (approximated by the C2–C1–C17–O1 angle), for 1d and AcPyr (Fig. 10).
Fig. 10 Plots of calculated energy vs. C2–C1–C17–O1 dihedral angle for 1d (a) and AcPyr (b) and optimized geometries of the most stable conformations of these compounds (c) and (d). |
The most stable geometry of 1d is planar with the carbonyl oxygen interacting with C3–H (Fig. 10a and c). The same conformation is present in the crystals of 1a (vide supra), a small twist of the CO group being attributable to the packing effects.
It is more stable by 2.34 kcal mol−1 than the conformation with the carbonyl oxygen directed towards C14–H and having the C2–C1–C17–O1 dihedral angle 150°. This means that the equilibrium amount of the latter conformer at room temperature is <3%. The calculated barrier to rotation of the carbonyl group is 5.03 kcal mol−1. On the other hand, the most stable conformation of AcPyr (Fig. 10b and d) is nonplanar with a C2–C1–C17–O1 dihedral angle 25° and with the carbonyl oxygen also interacting with C3–H (the corresponding planar conformation is slightly less stable by 0.18 kcal mol−1). However, in this case the conformation with the dihedral CO/pyrene angle equal to 145° is less stable only by 1.49 kcal mol−1, which means that the equilibrium amount of this conformation at room temperature may reach 8–10%. The energy barrier to rotation of the acetyl group was found to be equal to 2.72 kcal mol−1. It is also worth noting that the pyrenyl-CO bond in 1d is significantly shorter than the analogous bond in AcPyr (1.483 and 1.495 Å, respectively). This reveals a more efficient pyrenyl-CO conjugation in 1d, which is in line with the higher energy barrier for rotation around this bond in this compound than in AcPyr.
The calculated molecular orbitals of 1d and AcPyr are shown in Fig. 11
Fig. 11 DFT-calculated molecular orbitals for 1d and AcPyr. Orbital energies in eV in vacuum and (in parentheses) in CHCl3. |
The HOMO-1 and HOMO of 1d and AcPyr differ only slightly in energy (∼0.1 eV), presumably because in these orbitals the electron density is localized mostly at the pyrene moiety. In contrast, the HOMO-2 orbitals of these compounds are localized on the acyl groups, which results in a stronger (0.4 eV) stabilization of this orbital in 1d. A significant difference in localization of electron density is observed for unoccupied LUMO, LUMO + 1 and LUMO + 2 orbitals. In the case of AcPyr these orbitals are localized on the pyrenyl moiety and the carbonyl group, whereas delocalization is observed on the ethynyl group in the case of 1d. The strongest stabilization (∼0.4–0.5 eV) is observed for the LUMO and LUMO + 2 orbital of 1d, in comparison to the same orbital in AcPyr.
We have also calculated the electronic absorption spectra of 1d and AcPyr for isolated molecules and for a chloroform solution using time-dependent DFT and a polarizable continuum model (PCM) using integral equation formalism variant (IEFPCM). The data are gathered in Table 3.
Compound | λmax/ nm (eV) | f | Main contribution(s) |
---|---|---|---|
(a) Isolated molecules | |||
1d | 407.4 (3.04) | 0.3985 | H → L (0.95) |
356.1 (3.48) | 0.0527 | H-1 → L (0.79) | |
292.1 (4.24) | 0.1297 | H → L + 1 (0.71) | |
261.9 (4.73) | 0.1231 | H-4 → L (0.79) | |
AcPyr | 377.0 (3.29) | 0.3523 | H → L (0.92) |
283.5 (4.37) | 0.1909 | H → L + 1 (0.45) | |
H-1 → L (0.19) | |||
247.6 (5.01) | 0.1743 | H-4 → L (0.62) | |
(b) CHCl3 solution | |||
1d | 429.3 (2.89) | 0.5538 | H → L (0.97) |
362.3 (3.42) | 0.0774 | H-1 → L (0.85) | |
296.4 (4.18) | 0.2104 | H → L + 1 (0.70) | |
291.7 (4.25) | 0.0510 | H → L + 2 (0.73) | |
266.9 (4.64) | 0.1416 | H-4 → L (0.84) | |
249.9 (4.96) | 0.1003 | H-5 → L (0.89) | |
AcPyr | 386.8 (3.21) | 0.4962 | H → L (0.95) |
298.7 (4.15) | 0.0525 | H-3 → L (0.64) | |
H → L + 1 (0.22) | |||
287.0 (4.32) | 0.2435 | H → L + 1 (0.37) | |
H-3 → L (0.32) |
The lowest energy band may practically be considered a pure HOMO–LUMO transition for both compounds. The calculated wavelengths of this band for the chloroform solution are in good agreement with the experimental values (429 vs. 415 nm for 1d and 387 vs. 391 nm for AcPyr). It is generally believed that face-to-face H-aggregates are non-emissive in terms of Kasha's theory of exciton coupling. However, some examples of emissive H-aggregates have been reported.32–34 In such aggregates, emission may arise from restriction of intramolecular rotation (RIR), blocking nonradiative decay channels or formation of emissive solid-state excimers (meaning not only “excited dimers” but also “excited oligomers”). A time-resolved fluorescence study clearly demonstrated that solid-state emission of 1a originated from dynamic excimers formed during and shortly after the laser pulse and emitting at longer wavelengths than monomeric molecules. Some contribution from the RIR effect is also possible since the intermolecular hydrogen bond network along with π-stacking may severely inhibit rotational motions in individual fluorophores in the crystals.
Footnote |
† Electronic supplementary information (ESI) available: Syntheses of 1a–d. Electronic absorption and emission spectra of 1a–d in various solvents. Details of X-ray diffraction and photophysical studies. CCDC 996918. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c4ra03961k |
This journal is © The Royal Society of Chemistry 2014 |