Synthesis and Solid State Structures

The main objective was the preparation of new intercalation structures of SPAnPS for structural and solid state fluorescence comparison to determine the influence of intermolecular interactions and fluorophore deformation on emission properties. In this context the preparation of lattice solvent free crystals – which had not been previously achieved – was of particular interest.

9,10-bis(diphenylthiophosphoryl)anthracene (SPAnPS) was prepared according to literature procedures.^{[44b, 44c]} 9,10-Dibromoanthracene was di-lithiated using nBuLi at – 15°C and then reacted with two equivalents of chlorodiphenylphosphane. Removal of lithium chloride by filtration and oxidation with elemental sulfur in toluene at 110°C gave SPAnPS at high yields (Scheme 3-5). Crystallization of SPAnPS from toluene

yielded the host/guest complex containing two toluene molecules per SPAnPS molecule. This complex will be referred to as SPAnPS@tol (15) from here on.

Scheme 3-5: Synthesis of SPAnPS.

SPAnPS@tol (15) was dissolved in – and crystallized from – each of the following solvents/solvent mixtures: EtOAc, MeCN, Acetone/hexane 2:1, DCM/MeOH 3:1, and toluene d8. Hot saturated solutions were prepared and slowly cooled to ambient temperature, –3°C, or –30°C, respectively. Crystallization conditions for each solvent/solvent mixture are listed in Table 3-4. moieties from getting close enough to one another to form π stacked arrangements in the solid state.

Hence, the observed π-π distances are regularly between 8 and 10 Å, which can be considered negligible in terms of π-π interaction. Interactions between SPAnPS molecules are therefore generally rare, interactions mostly occur between intercalated molecules and SPAnPS. Due to these large intermolecular distances, cavities or “pockets” within the structure are formed which are suitable in size and shape for the intercalation of aromatic solvents. The

deformation of the fluorophore.[44b, 44c, 59b]

If the intercalated solvent is not aromatic (→ non-planar) or carries too large substituents to fit into the cavities of the structure, a cisoid conformation is formed with a large folding angle and smaller twist angle.^[59b]

Also the transoid conformation mostly leads to intercalation of two guest molecules per host molecule, while the cisoid conformation may intercalate one or two guest molecules.

When SPAnPS is crystallized from acetonitrile, SPAnPS@MeCN (16) is formed.

Despite the linear geometry and the low steric demand of the acetonitrile molecule, the cisoid conformation is generated. One guest molecule of acetonitrile is co-crystallized in the structure of 16 (Figure 3-22).

Figure 3-22: crystal structure of SPAnPS@MeCN (16), hydrogen atoms are omitted for clarity.

It is located above the plane of the anthracene moiety at a distance of well over 4 Å.

16 crystallizes in the hexagonal space group P63/m, which is very unusual and has not been previously observed for phosphoryl anthracenes. It is the first compound of this kind which is completely symmetric and has a mirror plane located directly down the middle of the molecule. The acetonitrile molecule is located on this mirror plane. This unique geometry leads to the typical “cis”-folding of the anthracene moiety, but no twist deformation whatsoever. The folding angle measures 27.8°. In this respect SPAnPS@MeCN (16) is the direct opposite of SPAnPS@tol (15), which shows only twist deformation but no folding. The P–S and P–C-bond distances are very similar to those found in SPAnPS@tol (15). As expected, there is no π-π-overlap in the structure of

SPAnPS@MeCN (16). The only noteworthy intermolecular interactions present are two C–H^…π bonds of the 2,3-hydrogen atoms to the central ring of an adjacent anthracene moiety (Figure 3-23).

Figure 3-23: C–H^…π bonding in the structure of SPAnPS@MeCN (16), phenyl substituents are omitted for clarity.

Due to the symmetry of the molecule both interactions show identical distances and angles of 2.667 Å / 56.9°. Though this distance is fairly short, the hydrogen atoms are located above the very edge of the central C6-perimeter, which makes the π-density in this region at least questionable and therewith makes this interaction weaker than the distance and angle imply.

Crystallization of SPAnPS from an acetone/hexane mixture again affords a structure with a cisoid conformation of the SPAnPS molecule. SPAnPS@Ace (17) crystallizes in the monoclinic space group P21/c and the asymmetric unit contains one molecule as well as disordered fragments of acetone and hexane molecules. The observed bond distances do not differ significantly from those observed in the structure of 16. A fairly strong folding angle of the anthracene moiety of 28.3° is found, as well as a twist deformation of 3.1°. As in 15 and 16, no noteworthy π-π interactions are present in the structure of 17. A single C–H^…π interaction between a para C-H of a phenyl substituent and a peripheral anthracene ring is observed (Figure 3-24). It measures 2.714 Å and encloses a moderate angle of 60.5° with the π system.

Figure 3-24: C–H^…π bonding in the structure of SPAnPS@Ace (17), phenyl substituents are omitted for clarity.

When SPAnPS is crystallized from DCM/MeOH 3:1, SPAnPS@DCM (18) is obtained.

Two molecules of DCM are co-crystallized in this structure, MeOH is not intercalated.

18 crystallizes in the monoclinic space group P21/c and the asymmetric unit contains one molecule of SPAnPS@DCM. The P–S and P–C bond distances are in the expected range, showing only marginal deviations compared the previous structures of SPAnPS.

A cisoid conformation of the SPAnPS molecule is found which is nearly perfectly symmetric. SPAnPS@MeCN (16) – which is perfectly symmetric – shows a twist angle of 0°. SPAnPS@DCM (18) which is structurally very closely related exhibits a twist angle of 1.1° which deviates only minimally. The symmetry of the structure of SPAnPS@DCM is lowered by the intercalated DCM molecules.

Figure 3-25: left: crystal structure of SPAnPS@DCM (18), hydrogen atoms are omitted for clarity; right:

C–H^…π bonding in the structure of 18, phenyl substituents are omitted for clarity.

While the twist angle is very small, a folding angle of 25.0° is found, which is slightly smaller than that of SPAnPS@MeCN (16). Again, no π-π interaction is present in the structure of 18, but a C–H^…π bond of the para-C-H of a phenyl substituent to a peripheral ring of the anthracene moiety is found. The distance to the π-system measures 2.796 Å at an angle of 59.0°.

So far, all solvents/solvent mixtures used were non-aromatic. All intercalation structures resulting from crystallization from these solvents showed cisoid conformation of the SPAnPS molecule with large folding angles and small twist angles.

Crystallization of SPAnPS from EtOAc surprisingly does not lead to an intercalation of solvent molecules (Figure 3-26). When crystallized at 22°C, the first completely lattice solvent free structure of SPAnPS is obtained, which will be referred to as SPAnPS_pure (19). Crystallization temperatures below 0°C lead to a structure which contains strongly disordered EtOAc molecules. However the quality of the acquired X-ray diffraction data was too poor for structure refinement, which is why this structure is not included in this thesis. The quantitative formation of solvent-free crystals at 22°C crystallization temperature was also confirmed by NMR experiments.

Figure 3-26: Crystal structure of SPAnPS_pure (19), hydrogen atoms are omitted for clarity.

SPAnPS_pure (19) affected by the stronger distortion of the molecule.

A single C–H^…π bond from the para C-H of a phenyl substituent to the adjacent π-system is found, which is very similar in orientation and geometry to the interaction found in SPAnPS@DCM (18). It measures only 2.613 Å, which is quite short, and encloses an angle of 53.0° with the π-system.

Finally, SPAnPS was crystallized from toluene d8. The obtained compound SPAnPS@tol_d8 (20) was prepared to determine whether the change from hydrogen to deuterium would affect the C-H^…π interactions present in SPAnPS@tol (15). Also a sufficient amount of 20 was prepared for acquisition of solid state fluorescence spectra. This way comparison of C-H^…π bond distances from the crystal structures, as well as of emission properties was possible. Because an identical mode of intercalation in the host/guest complex was proposed, the contribution of C-H^…π bonding to fluorescence emission by weighing up of small structural alterations induced by differences between C-H^…π and C-D^…π bonding was sighted.

Surprisingly, the unit cells of SpAnPS@tol (15) and SPAnPS@tol_d8 (20) differed significantly, the latter showing a unit cell volume of 9682.8 ų, which is more than the fourfold volume observed for SPAnPS@tol (15). A second dataset using a new crystal was collected which confirmed the cell parameters. To eliminate uncertainties, the

Figure 3-27: C-H^…π bonding of SPAnPS_pure (19), phenyl substituents are omitted for clarity.

crystals were re-dissolved and crystallized again. Matrix scans of several new crystals were recorded and the cell parameters were identical every time, confirming the results of the first batch of crystals.

While 15 crystallizes in the monoclinic space group P21/n, 20 crystallizes in the triclinic space group P . The symmetry of 20 is lowered due to the disorder of toluene guest molecules. The asymmetric unit contains four whole SPAnPS molecules and two half SPAnPS molecules. Additionally, seven toluene d8 molecules are intercalated, 5 of them in disordered positions. Though all SPAnPS molecules have very similar transoid geometries, they all differ slightly in terms of folding and twist angles which are listed in Table 3-8.

Figure 3-28: Superposition of SPAnPS@tol (15) and SPAnPS@tol_d8 (20), molecule 4 (dashed).

Moreover, all molecules found in the structure of 20 also minimally differ from SPAnPS@tol (15) (Figure 3-28). While all molecules of 20 exhibit slightly smaller twist angles than 15, four of the molecules show weak folding around 2°. Hence, the weaker twisting of the molecules is compensated by slight folding of the fluorophores.

Exceptions are the two molecules which are generated by symmetry operations, similar to SPAnPS@tol (15). Due to symmetry reasons their folding angle is also 0° and the observed twist angle is at approximately 18°, resulting in an overall weaker deformation than found in SPAnPS@tol (15). Several phenyl substituents of 20 show slight disorder. Although the toluene d8 molecules are also located above and below

the anthracene planes, they are with few exceptions strongly disordered and do not adopt fixed positions as the normal toluene molecules do in 15. Even the non-disordered toluene d8 molecules are not positioned in accordance with the normal toluene molecules in 15. Figure 3-29 shows the structures of SPAnPS@tol (15) (left) and the structure of SPAnPS@tol_d8 (20), molecule 3 (right) including the respective C-H^…π/C-D^…π bonded toluene molecules. While in 15, the toluene molecules are associated by symmetry, they are not in 20, which is why they are marked with A and B. A is not disordered and adopts a fixed position, while B is one of three stationary disordered positions. It is obvious at first sight that the toluene molecules in 15 assume an almost orthogonal position relative to the anthracene plane, which is often referred to as a T-shaped or “herringbone” arrangement. In contrast, the toluene molecules in 20 are arranged in a more flat angled manner.

Figure 3-29: comparison between SPAnPS@tol (15) (left) and SPAnPS@tol_d8 (20), molecule 3 (right);

A = toluene d8 molecule in fixed position, B = disordered toluene d8 molecule (only one position depicted).

This also reflected by the lengths and angles of the C-H^…π/C-D^…π interactions. The two sp² C-H^…π bonds found in 15 measure 2.816 Å at 76.8°to the ring plane. For the disordered toluene d8 molecules in 20, an accurate determination of C-D^…π distances is difficult, but the angles to the π-system can be acquired at least roughly. For the disordered toluene molecules these angles range from 35° to 65°, depending on the examined molecule. For the toluene molecules in fixed positions a more accurate determination is possible. Here the sp² C-D^…π distances range from 2.850 Å to 3.076 Å.

Moreover the angles enclosed with the π-system deliver values between 55° and 65°.

Some toluene d8 molecules are displaced from the “optimum” position so far that the aromatic hydrogen atoms are not positioned above the π-system, making C-D^…π bonding unlikely. In some cases also C-D^…π interactions of the toluene methyl group and the π-system are found ranging from ~2.75 to ~3.00 Å.

Overall, the C-D^…π interactions found in the structure of SPAnPS@tol_d8 (20) appear to be weaker than those found in SPAnPS@tol (15), which can be deduced from the distances and angles of these interactions. While normal toluene is only found in fixed positions within the structure of 15, toluene d8 mostly appears in disordered positions, and only few fixed positions. This indicates that the interactions of carbon bound deuterium atoms with aromatic π-systems seem to be weaker than those of hydrogen atoms. The stronger interaction produces nearly orthogonal arrangements in fixed positions, the weaker interaction produces flat-angled arrangements and is not capable of binding the toluene d8 molecules in fixed positions.

These differences in bonding should also become manifest in the solid state emission properties of 15 and 20.

3.2.3 Solid State Fluorescence

The solid state fluorescence properties of 15-20 were investigated to fathom whether structural features and intermolecular interactions gathered in 2.3.2 are also reflected by the observed fluorescence emission.

For this purpose, sufficient amounts of single crystals of 15-20 were removed from the mother liquor by filtration and immediately ground and filled into the solid state sample cell for measurement. This way evaporation of lattice solvent was largely suppressed, which is important for consistent results. The evaporation of solvent leads to destruction of the crystalline structure of the sample, which is accompanied by fluorescence quenching. Depending on the volatility of the lattice solvent, this can be a slow or fast process. Although the resulting decomposition is at first only limited to the surface of the microcrystalline particles of the sample, the impact on fluorescence emission is not negligible. To quantify this effect a sample of SPAnPS@tol (15) was filled into the solid state sample cell and repeatedly subjected to identical fluorescence measurements over the course of ten hours. Though the sample cell is a closed device,

it is not completely air proof and allows the diffusion of evaporated lattice solvent from the sample. Nevertheless this process most likely proceeds even faster in an open vessel.

Figure 3-30 shows the decay of fluorescence intensity over elapsed time. It clearly indicates the quenching of emission upon loss of lattice solvent and therewith of crystalline structure. The speed of crystal decomposition is also dependent on the particle size, because finely ground particles possess a larger surface than larger particles and allow more lattice solvent to evaporate in a defined time interval.

Figure 3-30: Time dependent fluorescence decay of SPAnPS@tol (15) by gradual evaporation of lattice solvent.

If the evaporation process were monitored over an even longer period of time, eventually the emission intensity of solvent free SPAnPS powder would be reached, which was produced by Fei et al. by drying of SPAnPS@tol crystals under heating and reduced pressure.^{[44b, 44c]}

The observations made by Schwab (which were founded exclusively on optical inspection) that only structures of SPAnPS in a transoid conformation show strong solid state fluorescence were generally confirmed by solid state fluorescence experiments. Because the weakly fluorescent cisoid structures require different measurement conditions than the strongly fluorescent structures in order to obtain suitable data, comparison is only reasonable among compounds of similar

1,00E+06

conformation. The only transoid structures obtained were those of SPAnPS@tol (15) and SPAnPS@tol_d8 (20). As a crystalline analogue to the vacuum dried crystals which were used by Fei et al. for comparison, the lattice solvent free SPAnPS_pure (19) was employed. It will therefore serve as a comparison for molecules in both the transoid and the cisoid conformation.

Figure 3-31: Left: normalized excitation (red) and emission (green) spectra of SPAnPS@tol (15); right:

normalized excitation spectra of SPAnPS@tol (15) (red) and SPAnPS@tol_d8 (20) (green).

Figure 3-31 (left) shows the normalized solid state excitation and emission spectra of SPAnPS@tol (15). In a broad range from 350 nm to 480 nm 15 shows strong fluorescence, reaching a sharp maximum at an excitation wavelength of 467 nm. The emission band is not quite as broad but also spans over 50 nm. It reaches its maximum at 519 nm and does not exhibit the typical anthracene vibrational structure, which has been previously observed for several phosphorylanthracenes in solution (c.f. 3.1). The excitation spectra of SPAnPS@tol (15) and SPAnPS@tol_d8 (20) (Figure 3-31, right) are of nearly identical shape and show identical maxima, but also deviate slightly in some regions, which already indicates that the differences observed in the crystal structures may also be reflected by the fluorescence properties of both compounds. Also the maximum emission wavelengths are affected and are surprisingly not identical (Figure 3-33, left). Though they only differ by 4 nm, the differences induced co-crystallized toluene and toluene d8 are undeniable. The lattice solvent free SPAnPS_pure (19) shows even stronger deviation from 15 in its maximum emission wavelength, which exhibits a bathochromic shift of 8 nm. Though π-π overlap has repeatedly been named as an influential factor on maximum emission wavelengths,^{[50, 53]} the large π-π distances within the structures of 15, 19, and 20 virtually preclude this option.

Figure 3-32: Sample of SPAnPS@tol_d8 (20) in daylight (left) and under exposure to UV light, λEx = 366 nm (right).

The differences between these three compounds become even more evident when comparing their emission intensities (Figure 3-33, right). The emission of the lattice solvent free 19 is nearly completely quenched compared to SPAnPS@tol (15) and SPAnPS@tol_d8 (20). The intensity ratio of 1 : 0.018 between the emission of 15 and 19 (equates to factor 56) is in the range of the intensity ratio reported by Fei et al.

between SPAnPS@tol and vacuum dried crystals.[44b, 44c]

But also 15 and 20 differ significantly in terms of emission intensity. Independent of the excitation wavelength, the structure containing toluene d8 as lattice solvent only reaches between 60% and 70% of the emission intensity produced by the structure containing regular toluene (Table 3-9).

Table 3-9: Compiled maximum emission wavelengths and relative emission intensities of 15, 19, 20.

λEm (max) [nm] Irela

Irelb

Irelc

15 519 1 1 1

20 515 0.62 0.66 0.70

19 527 0.018 n/A n/A

a) λ_Ex = 380 nm; b) λ_Ex = 449 nm; λ_Ex = 467 nm.

Figure 3-33: Left: normalized solid state emission spectra of SPAnPS_pure (19) (red), SPAnPS@tol (15) (green), and SPAnPS@tol_d8 (20) (blue); right: solid state emission spectra of SPAnPS_pure (19) (red), SPAnPS@tol (15) (green), and SPAnPS@tol_d8 (20) (blue).

Explanations for the observed phenomena can be deduced from the structural properties discussed in 3.2.2. Compared to both compounds with a transoid conformation of the SPAnPS molecule (15 and 20), the cisoid structure of 19 features an enormous folding angle of the anthracene moiety. It measures 32.1°, while the transoid structures exhibit folding angles of 0° (15) and an averaged angle of 1.15° (20), respectively. Thus, the folding of the anthracene moiety appears to be obstructive for fluorescence emission, while twist deformation (which is strong for both 15 and 20) does not appear to have a similarly strong quenching effect. On the other hand, the small deviations in fluorophore deformation between SPAnPS@tol (15) and SPAnPS@tol_d8 (20) do not seem to justify the loss of over one third of emission intensity from 15 to 20. Fei et al. have assigned the strong emission of SPAnPS@tol (15) to the formation of an exciplex between the fluorophore and the co-crystallized toluene molecules. The T-shaped arrangement and the formation of it by C-H^…π bonding were shown play a key role in enabling of fluorescence emission. While the toluene molecules in the structure of 15 are in fixed positions in the T-shaped complex and exhibit no disorder, the toluene d8 molecules in the structure of 20 show strong disorder and a much flatter averaged angle to the fluorophore. In some cases the orientation of the toluene d8 molecules even prevents interaction with the fluorophore altogether. Hence, the weaker C-H^…π bonding of the toluene d8 molecules to the fluorophore leads to flat angled arrangements and in consequence to weaker fluorescence emission. This clearly demonstrates the importance of C-H^…π interactions for the formation of strongly emitting molecular arrangements.

Figure 3-34: Left: normalized solid state emission spectra of SPAnPS@Ace (17) (red), SPAnPS@DCM (18) (green), SPAnPS@MeCN (16) (blue), and SPAnPS_pure (19) (cyan); right: solid state emission spectra of

SPAnPS@Ace (17), SPAnPS@DCM (18) (green), SPAnPS@MeCN (16) (blue), and SPAnPS_pure (19) (cyan).

Although by a magnitude weaker fluorescent, the emission properties of the cisoid structures can also be compared. While SPAnPS@MeCN (16), SPAnPS@Ace (17), and SPAnPS@DCM (18) all exhibit very similar maximum emission wavelengths, the emission maximum of the lattice solvent free SPAnPS_pure (19) is again red-shifted by approximately 10 nm (Figure 3-34, left). As pointed out before, this effect cannot be produced by π-π overlap which is completely absent in all four structures. Though all four compounds are weakly fluorescent, SPAnPS_pure (19) shows the clearly strongest

Although by a magnitude weaker fluorescent, the emission properties of the cisoid structures can also be compared. While SPAnPS@MeCN (16), SPAnPS@Ace (17), and SPAnPS@DCM (18) all exhibit very similar maximum emission wavelengths, the emission maximum of the lattice solvent free SPAnPS_pure (19) is again red-shifted by approximately 10 nm (Figure 3-34, left). As pointed out before, this effect cannot be produced by π-π overlap which is completely absent in all four structures. Though all four compounds are weakly fluorescent, SPAnPS_pure (19) shows the clearly strongest

Im Dokument Synthesis and Fluorescence Properties of Anthracene Derivatives and their Metal Complexes (Seite 66-83)