Long-range exciton transport in conjugated polymer nanofibers prepared by seeded growth

See allHide authors and affiliations

Science  25 May 2018:
Vol. 360, Issue 6391, pp. 897-900
DOI: 10.1126/science.aar8104

A longer exciton pathway

Organic semiconductors typically exhibit exciton diffusion lengths on the order of tens of nanometers. Jin et al. prepared nanofibers from block polymers consisting of emissive polyfluorene cores surrounded by coronas of polyethylene glycol and polythiophene (see the Perspective by Holmes). Excitons generated in the polyfluorene cannot enter the polyethylene glycol layer and so diffuse more than 200 nm. This distance can be tuned by varying the length of the polyethylene glycol—a feature that could potentially be exploited in the development of organic devices such as photovoltaics.

Science, this issue p. 897; see also p. 854


Easily processed materials with the ability to transport excitons over length scales of more than 100 nanometers are highly desirable for a range of light-harvesting and optoelectronic devices. We describe the preparation of organic semiconducting nanofibers comprising a crystalline poly(di-n-hexylfluorene) core and a solvated, segmented corona consisting of polyethylene glycol in the center and polythiophene at the ends. These nanofibers exhibit exciton transfer from the core to the lower-energy polythiophene coronas in the end blocks, which occurs in the direction of the interchain π-π stacking with very long diffusion lengths (>200 nanometers) and a large diffusion coefficient (0.5 square centimeters per second). This is made possible by the uniform exciton energetic landscape created by the well-ordered, crystalline nanofiber core.

The ability to transport excitation energy over length scales comparable to the optical absorption depth (100 nm and beyond) is central to the function of a range of devices, including solar cells. Most thin-film organic semiconductor structures, such as those formed from conjugated polymers, show short exciton diffusion lengths (LD) of around 10 nm (1) that are primarily constrained by energetic disorder. As a result, these are fabricated as blends of electron donor and acceptor materials with length scales for the partly demixed materials targeted around 10 nm, so that all excitons can reach the charge-generating hetero-interface (1, 2). In contrast, diffusion ranges for singlet excitons in purified single crystals are known to be considerably larger (up to 220 nm) (1, 36). However, in these cases the materials are generally polydisperse in dimensions and are problematic to incorporate into useful devices. Device development is thus dependent on the ability to develop uniform nanostructures that are amenable to processing and able to support long-range exciton diffusion.

Self-assembly of molecular and polymeric amphiphiles in solution has recently emerged as a promising route to core-corona nanoparticles (micelles) suitable for device applications. For instance, long-range exciton diffusion has been observed in dye molecule H and J fiber-like aggregates (where H and J characterize a shift to shorter or longer wavelength upon aggregation, respectively) by single-molecule spectroscopy (7, 8) and in more detailed investigations performed on single conjugated polymers isolated in matrices or dilute solution (9, 10). These latter studies, however, necessarily exclude the role of interchain transport in exciton diffusion. Such transport is possible in solution-processed bulk conjugated polymers, which have been reported to show a singlet exciton diffusion length of 70 nm (11), but this is still short of the lengths exhibited in molecular crystals. Here, we report exceptional exciton diffusion ranges in uniform nanofibers formed from the seeded-solution self-assembly of block copolymers with a crystallizable π-conjugated poly(di-n-hexylfluorene) (PDHF) block.

To facilitate the study of exciton diffusion within π-conjugated crystalline PDHF, we prepared segmented nanofibers comprising a continuous PDHF core with a discrete region of energy-accepting quaternized polythiophene (QPT) covalently attached as a corona at each end (Fig. 1, A and B). The central segment contained an electronically insulating poly(ethylene glycol) (PEG) corona, and the length of this region was varied to afford a size series of nanofibers. These segmented nanostructures were prepared using living crystallization–driven self-assembly, a recently developed seeded-growth method for producing one-dimensional (1D) and 2D objects of controlled dimensions (12, 13). The resulting structures are generally typified by highly crystalline cores, which can comprise a number of building blocks (1417). We, Faul, and co-workers have previously demonstrated the use of this method with polythiophene-containing block copolymers to yield fibers with a π-conjugated core, which form ensembles with promising transport properties but weak fluorescence due to aggregation-induced quenching (18).

Fig. 1 Formation of segmented PDHF nanofibers by multistep self-assembly.

(A) Schematic diagram illustrating the seeded growth process and the structures of PDHF14-b-PEG227 and PDHF14-b-QPT22. (B) Illustration of the segmented B-A-B nanofiber structure with separate donor and acceptor domains. (C) Normalized absorption of QPT homopolymer in THF:MeOH (1:1) (orange dashes), and unsegmented PDHF nanofibers (Ln = 1605 nm) (blue dashes), and photoluminescence (PL) emission of unsegmented PDHF nanofibers (Ln = 1605 nm) in the same solution (solid blue line). The I0–0 and I0–1 peaks in the PL are at 425 nm and 455 nm, respectively. The inset shows the energy levels of the PDHF and QPT.

PDHF exhibits bright blue fluorescence in the solid state (19), in electrospun microfibers (20), and in polydisperse nanofibers self-assembled from triblock copolymers in solution (21). The uniform examples used in the present study, however, were prepared in a multistep approach, so as to achieve the desired segmented coronal structure and varied length of the PEG component. Initially, nanofibers ~5 to 10 μm in length, derived from the block copolymer PDFH14-b-PEG227 (where the numbers refer to the number-average degree of polymerization of each block; see figs. S1, S3 to S5, and S10), were prepared by homogeneous nucleation in a 10:8 mixture of tetrahydrofuran and methanol (THF:MeOH). Analysis by transmission electron microscopy (TEM), atomic force microscopy (AFM), and wide-angle x-ray scattering (WAXS) revealed the presence of a crystalline PDHF core with a rectangular cross section (number-average width Wn = 12.9 nm, height = 4.5 nm) surrounded by a PEG corona (figs. S10 to S12). A solution of micelle seeds (number-average length Ln = 30 nm; polydispersity Lw/Ln = 1.03, where Lw is the weight-average length; height = 4.5 nm) was then prepared by sonication of the multimicrometer-long fibers (figs. S13, A and B, and S14A). Subsequent addition of different volumes of a solution of unimeric (molecularly dissolved) PDHF14-b-PEG227 copolymer in THF led to the formation of uniform nanofibers of controlled length (Fig. 1A and figs. S14 and S15).

To create nanofibers with a segmented corona, we prepared an all-π-conjugated donor-acceptor diblock copolymer, PDHF14-b-QPT22, comprising the same PDHF core-forming block and a QPT corona–forming block (figs. S2 and S6 to S9). This material was added in a molecularly dissolved unimeric state (in THF:MeOH 3:1) to the PDHF14-b-PEG227 nanofibers in THF:MeOH 1:1, leading to growth from the two PDHF core termini. The resulting uniform nanofibers had a B-A-B structure of controlled overall and segment length with a crystalline PDHF core present over the entire length, but with a π-conjugated corona-forming block located only on the terminal (B) segments (Fig. 1A). This was used to produce a size series of near-uniform nanofibers with Ln = 180 ± 40 nm, 300 ± 70 nm, 505 ± 100 nm, 945 ± 240 nm, and 1840 ± 540 nm, with each QPT segment comprising 35 to 120 nm of this length, as measured by TEM and supported by AFM (figs. S13, C and D, S16, and S17 and table S1). The dimensional control is illustrated further in Fig. 2A (and fig. S18), which shows a laser scanning confocal microscopy (LSCM) image of model nanofibers with central A-segments of 1.6 µm and QPT-corona–containing segments of 1.8 µm in length.

Fig. 2 Photoluminescence of segmented PDHF B-A-B nanofibers in solution.

(A) LSCM image of the uniform segmented PDHF nanofibers with a crystalline PDHF core (blue emission) and two terminal segments with QPT coronas (orange emission in THF:MeOH 1:1). Values of Ln for the central and terminal segments were 1.6 μm and 0.9 μm, respectively. (B) PL spectra of segmented PDHF nanofibers with different A-segment lengths, normalized to peak maxima. In each case, the solutions of segmented nanofibers (~0.5 mg/ml) in THF:MeOH 1:1 were excited at the PDHF absorption peak of 380 nm. Emission arising from direct excitation of the QPT in the 1605-nm sample was unresolved.

Figure 1C shows the absorption and photoluminescence (PL) spectra of unsegmented PDHF14-b-PEG227 nanofibers, the absorption spectrum of the QPT homopolymer (see fig. S19 for emission spectrum), and the energy levels of the two conjugated species. The overlap of absorption and emission in the PDHF indicates a small Stokes shift, whereas the PL illustrates a high degree of vibronic structure. The good overlap of QPT absorption and PDHF emission gives a PDHF-to-QPT Förster transfer radius of 4 nm.

We performed further steady-state optical measurements on the PDHF14-b-PEG227 as both nanofiber (of average length 435 nm) and unimer to probe the nature of exciton transport in the PDHF core. The nanofibers show more vibrational structure (Fig. 1C) than the unimer (fig. S20), and there is no pronounced red or blue shift upon aggregation. This is consistent with a π-stacked polymer possessing a transition dipole along the polymer backbone (22), which would be expected to exhibit both H- and J-like aggregate characteristics (23, 24). The ratio of the I0–0 to I0–1 PL bands is larger in the nanofiber than in the unimer (fig. S20), indicating a decreased Huang-Rhys parameter and hence a smaller configurational relaxation in the excited state (19).

We studied the energy transfer from PDHF to QPT for the aforementioned size series of segmented nanofibers; Fig. 2B shows the normalized PL when the PDHF is selectively excited (at 380 nm). We observed quenching of the PDHF peak relative to the QPT peak for average A-segment lengths below 775 ± 150 nm. This indicates energy transfer from the A-segment PDHF core to the B-segment QPT corona, although reabsorption of the PDHF PL is expected at this concentration. Further evidence for energy transfer comes from the excitation PL scan (fig. S21), which maps the absorption (fig. S22), and from the change in PL quantum efficiency with A-segment length (fig. S23). The latter matches that of pure QPT (13 ± 5%) below the critical length (775 nm) and approaches that for the unsegmented structure (73 ± 10%, a very high value) beyond this length.

Time-resolved PL measurements enable better quantification of energy transfer, because donor quenching kinetics are not affected by reabsorption effects. We used time-correlated single photon counting (TCSPC; instrument response time ~300 ps) to access low-excitation density regimes, and transient grating PL spectroscopy to probe the 1- to 100-ps time scale (see supplementary materials).

From TCSPC, the predominant natural lifetime of the unsegmented PDHF14-b-PEG227 nanofibers (with average length 435 nm) in solution is 430 ps (fig. S24), shorter than for the corresponding unimer (700 ps), as expected for a J-aggregate. Figure 3A shows the transient grating PL time slices of the segmented nanofiber with an A-segment core length of 775 nm. We see a reduction of the PDHF emission (~480 nm) and a concurrent rise in the broader QPT emission at longer wavelengths (530 to 630 nm). Figure 3B shows the kinetics of this transfer. We note that the QPT emission must be integrated over a broad wavelength range to account for the slow energy transfer that excitations undergo within the QPT itself (fig. S25).

Fig. 3 Transient grating PL spectra and kinetics of segmented PDHF B-A-B nanofibers.

(A) Transient grating PL time slices of a segmented nanofiber solution (0.5 mg/ml) with an average A-segment length of 775 nm, showing energy transfer from the PDHF to the QPT acceptor corona. There is a decay of the core PDHF I0–1 peak (which appears at 480 nm because of a filter cutting off the blue edge of the spectrum) due to exciton annihilation and quenching to the acceptor. This is accompanied by a concurrent rise in the broad QPT PL peak from 550 to 630 nm in the first tens of picoseconds (materials are excited with a 200-fs laser pulse at 400 nm with an equivalent excitation density of ~5 × 1017 cm–3). (B) Normalized PL kinetics of PDHF decay and rise of QPT signal for the spectra shown in (A). The green line shows the PDHF signal (integrated from 430 to 460 nm); the blue line shows the QPT PL (integrated from 530 nm to 630 nm). The solvent used was THF:MeOH 1:1.

We modeled the kinetics of the PDHF and QPT PL with a 1D diffusion model (25) that includes a contribution from Förster resonance energy transfer at the ends of the PDHF nanofiber into the QPT (see supplementary materials). For the TCSPC data (fig. S26) on the 775-nm A‑segment length nanofiber solution, the best fit gives a diffusion length of LD = 210 ± 100 nm and a diffusion constant of Embedded Image = 0.5 ± 0.2 cm2 s–1, with errors estimated from the polydispersity of the nanofiber solution and robustness of the fit. Such an exciton diffusion constant is higher than any currently reported for organic semiconductors (1).

Figure 4, A and B, shows the transient grating PL kinetics for the PDHF and QPT for a range of shorter segmented nanofibers as a function of A-segment length. We see faster quenching of PDHF emission and faster rises in the QPT emission as the A-segment length is decreased. To fit the transient grating data, and hence a larger sample of quenching lengths than are available to TCSPC measurements, we add a time-dependent exciton-exciton annihilation term to the diffusion model (see supplementary materials and figs. S27 and S28) to account for second-order decay at the higher excitation densities used in this measurement (between ~5 × 1017 and ~10 × 1017 cm–3). Using a global fit over multiple PDHF14-b-PEG227 segment lengths and fluences (fig. S29), we obtain a best fit to our data for an exciton diffusion length of 380 nm. A residual analysis (fig. S30) confirms the robustness of these values for LD greater than ~150 nm and less than ~600 nm. This agrees within error with the TCSPC result, adding support to our initial observation of large diffusion coefficients. We consider that the diffusion length most likely falls toward the smaller end of these values, because of the possibility of a small amount of exciton-charge annihilation in the higher fluence measurements, as well as the potential for a small amount of intermolecular energy transfer (fig. S31). Nonetheless, these results taken together indicate an exceptional diffusion constant for a conjugated polymer structure.

Fig. 4 Size-dependent transient PL kinetics and corresponding diffusion length model fits.

(A) Transient grating PL kinetics (squares) of PDHF PL signal (integrated from 430 to 460 nm) in segmented PDHF B-A-B nanofibers of different A-segment lengths. The PL decay time decreases with decreasing segment length, showing efficient transfer. Solid lines are example fits of a 1D diffusion model with a diffusion length of LD = 340 nm. The system was excited with a 400-nm, 200-fs laser pulse, at ~1018 cm–3 equivalent excitation densities for the samples in solution. (B) The corresponding PL kinetics (squares) of the rise in the QPT signal in the segmented PDHF B-A-B nanofibers, fitted with the same 1D exciton diffusion model (solid lines) and diffusion length in the PDHF of LD = 340 nm. (C) Transient grating spectra of PDHF-b-PEG nanofibers in solution. Time slices show spectra at early times after excitation, with excitation density of ~5 × 1017 cm–3. Spectral red-shifting is present until ~200 fs, and the ratio of the first and second vibronic peaks continues to reduce until ~700 fs (fig. S32). The solvent used was THF:MeOH 1:1.

In conjugated polymers, efficient exciton diffusion is correlated with an increased degree of structural order (4). Our results are consistent with this, as corroborated by the WAXS data (fig. S12), which show appreciable structural order in the unsegmented PDHF14-b-PEG227 nanofibers. This is supported by the pronounced vibronic structure and narrow spectral linewidths in the PL. Ultrafast transient grating PL measurements show evidence of both excitonic movement (during the instrument response time, ~200 fs) and some small degree of localization [in the first 700 fs after photoexcitation (Fig. 4C and fig. S32)]. We believe that this arises from rapid migration of excitons from the disordered to the ordered regions (evident from the WAXS data) and that exciton diffusion then occurs within these ordered regions in the nanofiber core.

Further understanding can be gained from the self-Förster radius, which is a useful tool for quantifying exciton diffusion lengths (26, 27), although the description has limitations (28, 29). We calculate a self-Förster radius in our PDHF fibers of 2.5 ± 0.2 nm (see supplementary materials). This is large in part because the parallel alignment of polymer chains gives ideal dipolar orientations for energy transfer (we have set κ2, our dipole orientation factor, to 1). From this radius we can calculate (27) a diffusion length, Embedded Image = 75 ± 15 nm, where d is the nearest-neighbor distance of 0.46 nm (fig. S12). Applying a Förster theory beyond the point-dipole approximation (29) would likely reduce this number slightly; however, this gives an estimate of expected exciton diffusion length for an incoherent, nearest-neighbor hopping regime. The fact that we measure diffusion lengths beyond this value is clear evidence that excitons are not hindered by the presence of localization sites, nor by local energy minima (30) in their transport. This provides evidence of a remarkably uniform energy landscape in these materials, enabled by their structural order. This uniformity would also benefit transport properties through a narrowing of the excitonic density of states (31).

Our measurement of a diffusion length greater than our estimate for nearest-neighbor Förster transport implies that we must look beyond this picture to explain our data. An example of a model that includes some degree of coherence but reduces to diffusive Förster transport as a limiting case is that proposed by Barford and Duffy (31). Their model uses a modified Redfield equation to show that a degree of interchain coherence in a polyfluorene film leads to an increase in the mean exciton hopping range, and hence gives a larger diffusion length. This requires interchain overlap, and the factor of 2.5 increase needed to account for our measured diffusion range gives a physically reasonable energy transfer integral of around 20 meV. Therefore, our measured exciton diffusion rate is high because of a combination of a physical packing structure optimizing the self-Förster radius, a lack of energetic trap sites, and a coherent component to exciton motion.

We have synthesized segmented nanofibers of controlled length with a PDHF core and containing a QPT corona at each terminus. Spectroscopic measurements in solution show that these nanostructures exhibit long-range exciton transport on the critical length scale comparable to the optical absorption length in conjugated polymers, and that this is enabled by the high degree of structural order in the PDHF core. In context, a 200-nm-thick film (the depth of our diffusion length) of conjugated polymer of average absorption coefficient (32) would absorb 98% of incoming photons. Such diffusion lengths could enable light-harvesting devices that use these polymer structures as antennae coupled to photodetector materials of limited absorption [such as monolayer transition metal dichalcogenides (33)] and would also enable much simpler bilayer design of organic photovoltaics relative to those based on the bulk heterojunction.

Supplementary Materials

Materials and Methods

Figs. S1 to S32

Table S1

References (3453)

References and Notes

Acknowledgments: We thank A. Chin for helpful discussions, R. Harniman for AFM measurements, and D. Hayward and O. Gould for WAXS experiments. Funding: Supported by EPSRC grants EP/K017799/1 (I.M.), EP/K016520/1 (R.H.F.), and EP/M005143/1 (R.H.F. and M.B.P.) and by a KAUST Competitive Research Grant (S.M.M. and R.H.F.). Author contributions: The project was devised by X.-H.J., M.B.P., R.H.F., G.R.W., and I.M., and these authors contributed to discussion of the data and writing of the manuscript; S.M.M. and A.R. contributed to the manuscript and project conception; X.-H.J., M.B.P., J.M.R., J.R.F., and C.E.B. performed the experimental work. Competing interests: The authors have no competing interests. Data and materials availability: All data are available in the manuscript or the supplementary materials.
View Abstract

Navigate This Article