A Low Reabsorbing Luminescent Solar Concentrator Employing π‐Conjugated Polymers

Gregory D. Gutierrez, Igor Coropceanu, Moungi G. Bawendi, Timothy M. Swager

Advanced Materials · 2015 · 79 citations · 41 references

Concepts

Abstract

A highly efficient thin-film luminescent solar concentrator (LSC) utilizing two π-conjugated polymers as antennae for small amounts of the valued perylene bisimide Lumogen F Red 305 is presented. The LSC exhibits high photoluminescence quantum yield, low reabsorption, and relatively low refractive indices for waveguide matching. A Monte Carlo simulation predicts the LSC to possess exceptionally high optical efficiencies on large scales. To increase the viability of solar radiation as a widespread and accessible class of renewable energy, researchers are actively developing solutions to lower the cost of deploying highly efficient photovoltaic (PV) devices. One widely investigated approach is to use a platform capable of focusing solar energy onto a set of small, but efficient PV cells. The most convenient example of this approach is the luminescent solar concentrator (LSC), which has the advantage that it is compatible with typical infrastructures.1-4 LSCs consist of transparent plastic or glass waveguides that channel luminophore photoemission from the absorption of sunlight to much smaller PV cells attached at their edges.5 Although a promising architecture as a result of the ability to effectively collect sunlight without tracking the Sun, conventional LSCs are often plagued by a multitude of unfavorable processes that curb their ability to deliver light to PV cells.3 Of these limitations, nonradiative reabsorption by the luminophore is regarded as one of the most severe problems and results from significant overlap of the molecule's absorption and emission spectra.6 Thus, although luminophores with high photoluminescence quantum yields (ΦPL) such as perylene bisimides are desired,7, 8 their performance in LSCs will suffer if they possess small Stokes shifts. A general approach to reducing reabsorption by molecules with narrow Stokes shifts, such as a perylene bisimide, is to employ minimal concentrations of luminophore such that its probability of encountering an emitted photon en route to the edges of the LSC is low.9 This tactic, however, comes at the cost of poor light harvesting by the emitter. To overcome this drawback and power the coupled PV cells, one or more strongly absorbing luminescent species can be added to the LSC to enhance the minority perylene bisimide's emission through energy transfer processes. Materials that have been recently investigated in an analogous manner are semiconducting heterostructured quantum dots10-12 and rods,13 wherein the excitons of a higher gap CdS shell are efficiently shuttled to a highly luminescent lower gap CdSe core with a much smaller absorption cross-section, resulting in a separated absorption spectrum primarily resembling that of the shell and an emission entirely attributed to the core. Although organic luminophore-based LSCs that exploit energy transfer schemes are also well documented,14-18 those aimed at specifically reducing reabsorption through the use of minority emitters have relied on a limited set of dyes19, 20 and biological motifs.21 The extension of this strategy to discover the untapped potential of any powerful organic emitter with a very small Stokes shift, especially one as prized as a perylene bisimide, remains as a challenge. As highly delocalized semiconducting macromolecular wires, π-conjugated polymers accommodate fast exciton migration along their backbones that may be easily disrupted by smaller quantities of lower gap emitters.22 Such energy transfer may be accomplished through dipole–dipole coupling mechanisms (i.e., Förster resonance energy transfer or FRET) between a luminescent polymer and a minority emitter with significant spectral overlap between the emission of the donor and the absorption of the acceptor.23 However, our laboratory has demonstrated that thin films of poly(arylene ethynylene)s (PAEs) may participate in electron exchange energy transfer processes (i.e., the Dexter mechanism) that circumvent this requirement and amplify the emission of spectrally mismatched fluorescent far-red dyes.24 We hypothesized that a combination of two luminescent PAEs of different optical bandgaps (Eg) may cooperate in an excitonic relay with a red-emitting perylene bisimide through both FRET and electron exchange processes (Figure 1a,b). Furthermore, through bandgap tuning of the polymers and judicious selection of the minority red emitter, the absorption spectra of both polymers can be decoupled from the emission of the terminal species while still exhibiting significant absorption within the visible spectrum. To demonstrate the promise of this proposed downconversion, we describe herein a highly emissive, low reabsorbing thin-film LSC that makes use of a migratory excitonic cascade from PAEs P1 (Eg ≈ 2.7 eV) and P2 (Eg ≈ 2.3 eV) to minority perylene bisimide Lumogen F Red 305 (Red 305, Eg ≈ 2.0 eV), the structures of which are displayed in Figure 1c. As the efficiencies of both FRET and electron exchange highly depend on the distance between the donor and the acceptor,25 a core requirement necessary to effectively activate all modes of energy transfer in this ternary scheme is to keep all participants in the cascade proximate to each other. An ideal configuration involves hosting the minority emitter in a layer of P1 and P2. To accomplish this, we needed to address the tendency of π-conjugated polymers to undergo aggregation-induced luminescence quenching that impedes energy transfer to the high ΦPL minority luminophore.26 To overcome this problem, we utilized bulky tert-butylated pentiptycene units to render the polymers' backbones incapable of simple and direct π–π stacking. P127 and P228 are structurally modified derivatives of PAEs previously studied by our laboratory and were synthesized via the Sonogashira cross-coupling reaction (see the Supporting Information). As a result of their inability to undergo interchain π–π stacking and alkyl substituents in their repeat units, these PAEs have high solubility for mixing with other polymers and casting into thin films. The tert-butylated pentiptycene repeat units were also used to result in highly amorphous and thus morphologically stable29 films to maintain efficient energy transfer between P1, P2, and Red 305 over time. The absorption and emission spectra of thin films of Red 305 in poly(methyl methacrylate) (PMMA) and the individual PAEs are displayed in Figure 1d. Although both P1 and P2 possess small Stokes shifts, there is no overlap between their absorption and emission of Red 305, thereby supporting the directed energy transfer cascade. The schematic of the LSC is illustrated in Figure 2a. The device is a 50/50 blend of P1 and P2 that functions as the host matrix for small amounts of Red 305 as the terminal isotropic emitter. The resulting thin-film composite lies on a thick transparent square substrate that acts as the primary waveguiding medium. The latter feature is best realized with a host layer refractive index that is nearly equal to or lower than the substrate at the relevant emission wavelengths.21 Consequently, the corresponding light guided to the edges of the LSC will be primarily confined to the substrate, thereby avoiding reabsorption. To obtain the refractive index n as a function of wavelength (Figure 2b), we applied spectroscopic ellipsometry to model the dielectric function of a host layer consisting of 1:1 (by weight) P1:P2 (see the Supporting Information). The wavelength profile of n exhibits wide dispersion that is typical of thin films of semiconducting π-conjugated polymers:30 n ranges as low as 1.58 in the absorptive regime and as high as about 1.70 in the emissive region of the blend. Within the photoluminescence of Red 305 (λ = 545–800 nm), n ranges from 1.65 to 1.61. Guided by these data, we used N-SF10 glass (n ≈ 1.7) as a compatible substrate, although polycarbonate is also a potential candidate (n ≈ 1.6). It is worth noting that the refractive indices of π-conjugated polymer films often depend on the thickness and processing conditions of the film.31 Thus, the data in Figure 2b are best translated to the LSC if similar processing conditions are observed. Nonetheless, by design n is relatively low for π-conjugated polymer films containing highly polarizable benzo[2,1,3]thiadiazoles as repeat units32 and this is the result of the large internal free volume provided by the tert-butylated pentiptycene scaffold.33 LSC1 was constructed by spin casting a blend of 1:1 (by weight) P1:P2 containing 1.5 wt% Red 305 on top of a square piece of N-SF10 glass with dimensions 17.5 mm × 17.5 mm × 1.5 mm (Figure 3). We found the resulting layer to be optically clear and observed no evidence of large aggregates of Red 305 via polarized optical microscopy. The geometric gain (G), or the ratio of the area of a single face of the LSC to the total area of edges coupled to a PV device, represents the concentrating capability of the LSC. Assuming all four edges will be completely covered by PV cells, the value of G for LSC1 is approximately 3. The measured film thickness of LSC1 (186.2 nm) is close to that of the ellipsometry sample and thus correlates with the data in Figure 2b. As a result of its small amount, Red 305 was assumed to not affect n. The absorption and photoluminescence spectra of LSC1 are displayed in Figure 3a.34 The two major bands that span λ = 400 nm to λ = 540 nm are attributed to P1 and P2, whereas the much smaller peak centered at λ = 564 nm is assigned to the minority Red 305 present. To directly quantify the amount of light absorbed at a given wavelength, the absorption spectrum is displayed in the form of its absorptance, which corresponds to the fraction of incident light absorbed after correcting for reflection (i.e., absorptance = 1–10−A, where A is the measured absorbance). LSC1 absorbs approximately 70% of light at its absorption maximum (λmax = 422 nm) and as much as 40% around λ = 500 nm. In contrast to these larger magnitudes, the peak absorption of Red 305 is only about 2%. However, upon excitation of LSC1 in a regime where absorption is mainly attributed to P1 (i.e., λex = 405 nm), the photoluminescence spectrum almost completely resembles that of Red 305 with minor residual emission from P2 present. The key metric of an LSC's performance is its optical efficiency (ηopt), defined as the fraction of incident photons emanating from the edges of the device. As a key goal of LSC research is to scale developed prototypes to commercially relevant sizes, we applied a Monte Carlo simulation to assess the promise of larger versions of our designed LSC by calculating ηopt as a function of G through an analysis of unproductive loss channels. Figure 4 summarizes these results using the photophysical data in Figure 3a, the total ΦPL, and an assumed refractive index of 1.7 across all emission wavelengths as input parameters for LSC1. For LSC1, ηopt was calculated as ≈40%. A plot of ηopt versus G (Figure 4a) exhibits a slow and steady decrease as G increases, an expected trend due to nonradiative reabsorption by the trace absorption of Red 305. Despite this drop, scaled versions of LSC1 are projected to maintain high ηopt at large G and is well above 20% even at G = 400, on par with predictions for high-performing nanocrystal LSCs.36 The concentration factor, or the product of G and ηopt, is also plotted as a function of G and represents the actual magnitude of photon concentration by the LSC, which rises in a nearly linear fashion as G increases. For example, at G = 400, a 90-fold concentration of photons toward PV cells at the edges is expected. Figure 4b summarizes the simulation's calculated impact of the possible loss mechanisms on the performance of LSC1 at G = 200 and a single excitation wavelength (λex = 405 nm). The displayed chart tabulates the fates of incident photons that strike the face of the scaled LSC. Prior to absorption, the fixed Fresnel coefficient for the input refractive index directly contributes to a 7% loss of photons via reflection off the surface of the LSC. By taking into account that some incident photons are lost due to incomplete absorption by the LSC after this reflection occurs (“Not Absorbed,” 31%), the simulation predicts nonradiative events (“Not Reemitted”) to result in a modest 21% loss in which minor reabsorption processes play a large role. Additionally, a leakage of emitted photons through the waveguide's escape cone (“Facial Loss”) is projected to result in a 15% overall loss. The remaining 27% of incident photons, albeit downconverted by Red 305 (“Collected”), represents ηopt. To summarize, we have demonstrated a blend of π-conjugated polymers as amplifying antennae for a perylene bisimide to create a low reabsorbing thin-film LSC with significant absorption in the visible spectrum and performance similar to state-of-the-art analogs. The participants of the described cascade all possess solubilities that allow for facile and low-cost solution processing of an amorphous, highly luminescent thin film. Through the development of even lower gap luminescent π-conjugated polymers, we envision creating additional high-performance LSCs capable of more complete absorption of the visible spectrum with efficient energy transfer directed toward spectrally separated minority terminal luminophores that emit in the far-red or near-infrared for improved spectral matching with silicon PV cells. This work was sponsored by Eni S.p.A. as part of the efforts of the MIT-Eni Solar Frontiers Alliance. Both G.D.G. and I.C. gratefully acknowledge support from the National Science Foundation Graduate Research Fellowship Program (Grant No. 1122374). The authors thank Lionel Moh for assistance with spectroscopic ellipsometry and Phil Reusswig and Tony Wu for initial assistance with integrating sphere measurements. As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

References

41