Although pure DLCs generally show a poor absorption in the visible spectral domain, mixtures of the electron-donating discoids with non-discogenic electron acceptors could exhibit absorption bands in the visible due to the formation of a CT complex. In many cases CT complexation even causes a considerable increase in the stability of the columnar mesophase. For good performance of a photovoltaic device the donor and acceptor molecules must form separate columns, i. e., enable charge separation and subsequent charge transport along the columnar wires. The position of the electron acceptors within the columnar mesophases is still contro­versial. Acceptor molecules such as TNF have been reported to be sandwiched between discotic molecules within the same column [31—34] but also inter­columnar, i. e., between the columns within the aliphatic tails of the discotic mol­ecules [29, 35]. Only the inter-columnar juxtaposition could provide a morphology with separate continuous columns for electron and hole transport. Another issue is that the characterization of (photo-induced) electron transfer and relaxation pro­cesses in self-assembled aggregates such as DLCs and DLC-CTs is in its infancy. The addition of electron acceptors such as TNF has been shown to increase the conductivity of DLCs. On the other hand, it has been proposed that recombination processes limit the hole photocurrent in DLC-CT compounds. Charge carriers in CT compounds are supposed to be trapped and readily annihilated through rapid, phonon-assisted relaxation and recombination processes.

In this section the morphology issue is elucidated by considering the prototypical discotic CT compound HAT6-TNF (Fig. 6.8), where HAT6 is used as electron — donating discoid and TNF as electron acceptor. HAT6-TNF forms a CT compound exhibiting a stable columnar phase from below room-temperature and up to 237 °C. The high symmetry and moderate molecular size of discogens such as HAT6 make these systems attractive for exploring the effects of increasing molecular complexity by comparing their photo-physical properties with those of the fundamental building blocks: benzene, and large poly-aromatic hydrocarbons. For discotic liquid crystal­line CT-complexes it is generally accepted that intermolecular charge-transfer occurs in the excited state, but not in the ground state. Mixtures of the electron-donating discoids with non-discogenic electron-acceptors exhibit absorption bands in the visible region due to excited-state charge-transfer. Support for these indications can be found from a combination of NMR and Raman spectroscopy measurements. Furthermore, the electronic transitions involved in the CT-band of HAT6-TNF can be characterized by combining UV-visible absorption and resonant Raman

spectroscopies. Additionally, the UV-visible and Raman measurements are accompanied by DFT calculations, to identify the vibrational modes that assist charge-carrier relaxation in the hot-band of HAT6 and in the CT-band of HAT6-TNF.

The ND pattern of HAT6-TNF (Fig. 6.16) is characteristic for a columnar mesophase, with sharp reflections in the small 20 region ((100), (010), etc.) origi­nating from the two-dimensional columnar lattice, a broad liquid-like band from the distribution in tail-tail distances, and a broad (001) peak from the intra-columnar distances [36]. Two observations are important. First, there is no superstructure peak visible with a double co-facial distance as would result in an intra-columnar juxta­position of TNF where TNF and HAT6 alternate, doubling the cell dimensions in the z-direction. This essentially makes such configurations unlikely. Second, the large decrease in lattice parameters is not accompanied by a similar increase in density that would result when simply shrinking the HAT6 cell to the new dimensions.[10]

These observations indicate that the columnar morphology has drastically changed in the charge-transfer compound. For a hexagonal columnar-structure with the TNF sandwiched between the HAT6 molecules, the column-column distance should be about 17 % smaller compared to pure HAT6. This only appears possible if the HAT6 and TNF are also alternately packed in the hexagonal plane. Other distributions result in energetically unfavourable inter-digitation of the aliphatic tails, such as the often suggested alternating intra-columnar packing with the dis — cotic molecules positioned in the same hexagonal plane. The intra-columnar jux­taposition of TNF, on the other hand, is only consistent with the experiments if the HAT6 columns are tilted on an oblique lattice. Such an arrangement, with discotic molecules slid laterally, has already been observed for highly polar HAT2-NO2 molecules. However, the tilted HAT6 columns leave such small spaces within the tail region that the TNF molecules should mainly adopt a vertical orientation.

The sandwich (with HAT6 and TNF alternating in the horizontal plane) and intra-columnar (with tilted HAT6) models were further analysed with Rietveld refinement. Remarkably, the refinement does not favour either of the juxtapositions of TNF over the other. Both models reproduce the characteristic features of the experimental diffraction patterns with comparable agreement. In addition, the effect of deuteration on the diffraction patterns also fits well with the measurements in both cases, particularly considering that no extra refinement step has been made; i. e., only the deuterium atoms of TNF were replaced with protons. Even the ori­entation dependences of the diffraction patterns on macroscopically aligned samples show little difference between the two models. As expected, the two-dimensional lattice peaks have maximum intensity when the diffraction beam is perpendicular to the column director, and the intra-columnar peak is at a maximum for the parallel orientation.

Clearly, it is difficult to determine the juxtaposition of TNF using only diffrac­tion. Nevertheless, the sandwich and inter-columnar models only fit with the observed density and diffraction patterns under the conditions illustrated in Fig. 6.17. As anticipated, the HAT6 columns in the inter-columnar model are tilted, with a co-facial slide of about 3.5 A between two neighbouring molecules in a column. We defined d column-column as the average separation between the column directors of two neighbouring columns, which is significantly larger for the inter-column juxtaposition of TNF. The minimal distance dHAT6core-TNF between the core of HAT6 and TNF predominantly determines the charge-transfer

#HAT*i>-TNFd

HAT. d TNFp

Fig. 6.18 Left Solid state CP-MAS 13C NMR spectra for HAT6 (a), HAT6-TNFD (b), TNF (c), and HAT6-TNF (d). The temperature and CP mixing time are 358 K, 10 ms for (a, b), 300 K, 2 ms for (c), and 300 K, 5 ms for (d). The HAT6 carbon assignment in (a) follows the labelling of Fig. 6.8. The blue (red) arrows indicate the downfield (up-field) shifts of HAT6 (TNF) peaks in the composite. Right 1H13C two-dimensional hetero-nuclear correlation spectra for HAT6-TNFD (coloured contours) and HAT6D-TNF (grey contours) at 290 K with a CP mixing time of 10 ms. The red numbers indicate cross-polarization between protons on the tail of HAT6 (H2H5 and H6) and specific TNF carbons shown in the inset. The circles in the inset surrounding the numbered carbons illustrate the strength of these interactions and the possible HAT6 hydrogens involved (pink for H2H5, green for H6). The HAT6 Cc and Co carbon and the Hcore hydrogen chemical shifts are indicated by the green, blue, and brown dashed lines, respectively. The signals marked with yellow arrows are due to imperfect deuteration of TNF behaviour of the complex. For the inter-columnar arrangement dHAT6core-TNF is difficult to estimate, since there is considerable freedom left in the refinement of the TNF position. We refined several inter-column models with different initial vertical positions of TNF from which we estimated that dHAT6core-TNF should be within the range of 4-10 A. Typically, the closest distance between TNF and HAT6 for the intra-columnar juxtaposition involved a CH C atom of HAT6 and TNF NO2 group.

Figure 6.18 (left) shows the solid state 13C cross-polarization (CP) magic-angle spinning nuclear magnetic resonance spectra of liquid-crystalline HAT6, TNF, and the CT complex. On the basis of these spectra of the un-complexed samples, all the peaks in the CT-complex spectrum are assigned to specific HAT6 or TNF carbons. In the CT compounds, however, the chemical shifts of the HAT6 and TNF carbons are changed significantly. All the TNF carbon signals are shifted up-field, reflecting a stronger local magnetic-field for the TNF in the mixture compared with the pure compound. In contrast to TNF, the HAT6 lines show both downfield and up-field shifts.

Chemical-shift changes in charge-transfer complexes have been attributed to partial electron-transfer from donor to acceptor molecules in the electronic ground — state. According to Mulliken’s theory, partial transfer of electron density occurs from the HOMO of the donor to the LUMO of the acceptor in the electronic GS. This is observed for the HAT6 donor. The anticipated general shift-sign (lower
electron density, lower field) that would result from partial electron-transfer to the acceptor TNF. For HAT6, the largest down-field shifts are observed for 13C nuclei in the outer part of the aromatic core, which is consistent with the spatial distri­bution of the HOMO [37].

A clear conclusion on the juxtaposition of TNF can be drawn from the two­dimensional 1H13C hetero-correlation nuclear magnetic resonance measurements. The signals indicated with red arrows in Fig. 6.18 (right) result from coherence transfer between the HAT6 tail-proton spins H1H6 and specific TNF carbons labelled in the inset. The TNF should be, at least for the major part, within the tail region of the HAT6 molecules. On the other hand, no interaction between TNF and the core of HAT6 was observed. The proton Hcore attached to the HAT6 core only shows cross-polarization with HAT6 carbons. For the differently deuterated sample, HAT6D-TNF, very little coherence transfer between TNF protons and any of the HAT6 carbons was observed. The absence of a significant reverse CP from TNF to the HAT6 tails is likely due to the different number of protons involved (12 for HAT6 tail protons H2H5, 1 for TNF protons). The combination of a rapid rigid-like CP build-up and the narrowing of the 1H line widths are indicative of selective averaging or quenching of weak longer range homo-nuclear 1H1H dipolar inter­actions by anisotropic motion of the HAT6 and TNF molecules in the liquid crystalline phase. Larger molecular displacements of HAT6, related to dynamic defects in the liquid-crystalline phase, occur on time scales up to milliseconds. These motions can quench the long-range dipolar interactions and contribute to the narrowing of the 1H lines in the data sets without decoupling, while allowing at the same time for the high CP rates for the 13C in the aromatic core by strong short — range hetero-nuclear dipolar interactions.

The nuclear magnetic resonance analyses seem to indicate that charge transfer from the HAT6 core to TNF already takes place in the ground state of the complex. Excited-state or ground-state charge-transfer from the HAT6 core to TNF requires that the electron acceptor should not be too far from the aromatic core. For the inter­columnar CT-complex structure the diffraction analyses resulted in HAT6core-TNF distances of 4-10 A, mostly involving the optically active NO2 groups of TNF. To ensure a sufficient orbital overlap for CT electron delocalization, most of the TNF molecules should be in the lower part of this range. The consistent analysis reveals a CT-complex morphology with dynamically disordered TNF molecules that are vertically oriented between the HAT6 columns, i. e., within the aliphatic-tail region. What does this mean for PV applications?

A promising observation is that there is a hole-conducting column present that is well separated from the electron acceptors. The columnar morphology has changed drastically in the composites, with a time-averaged tilted orientation and smaller average distances between the neighbouring HAT6 molecules within the column than in pure HAT6. In the CT complex the hole transport through the column will thus still be possible, while the CT process enables efficient charge separation. The liquid-crystalline structure and its facile alignment over macroscopic distances is an important asset for device realization.

The future PV application of CT complexes such as HAT6-TNF thus relies on improving the electron transport channel (Fig. 6.19). For instance, by searching for better acceptors that self-assemble into a separate channel, designing molecularly — connected donor and acceptor groups. The persistence of the hole conducting HAT6 column in the CT complex is promising for future application in organic PV systems. The major challenge towards such an application is to achieve a mor­phology that enables a BHJ PV device architecture.