Nanodroplet‐Containing Polymers for Efficient Low‐Power Light Upconversion

Sensitized triplet–triplet‐annihilation‐based photon upconversion (TTA‐UC) permits the conversion of light into radiation of higher energy and involves a sequence of photophysical processes between two dyes. In contrast to other upconversion schemes, TTA‐UC allows the frequency shifting of low‐intensity light, which makes it particularly suitable for solar‐energy harvesting technologies. High upconversion yields can be observed for low viscosity solutions of dyes; but, in solid materials, which are better suited for integration in devices, the process is usually less efficient. Here, it is shown that this problem can be solved by using transparent nanodroplet‐containing polymers that consist of a continuous polymer matrix and a dispersed liquid phase containing the upconverting dyes. These materials can be accessed by a simple one‐step procedure that involves the free‐radical polymerization of a microemulsion of hydrophilic monomers, a lipophilic solvent, the upconverting dyes, and a surfactant. Several glassy and rubbery materials are explored and a range of dyes that enable TTA‐UC in different spectral regions are utilized. The materials display upconversion efficiencies of up to ≈15%, approaching the performance of optimized oxygen‐free reference solutions. The data suggest that the matrix not only serves as mechanically coherent carrier for the upconverting liquid phase, but also provides good protection from atmospheric oxygen.

A first series of nanodroplet-containing polymers with a cross-linked glassy polymer matrix was prepared under ambient conditions by the redox-initiated free-radical polymerization of a mixture of the hydrophilic monomers 2-hydroxyethyl methacrylate (HEMA), methacrylic acid (MAA), and triethylene glycol dimethacrylate (TEG-diMA, cross-linker), a small amount (10 wt%) of the lipophilic, low-volatility liquid 1-tert-butyl-3,5-dimethyl benzene (BMB), and the surfactant cetyltrimethylammonium bromide (CTAB, Figure 2c). The mixture also contained ethylene glycol as refractive-index modifier and plasticizer for the polymer matrix, and the well-known TTA-UC sensitizer/emitter pair PdOEP (2 × 10 −5 m) and DPA (1.5 × 10 −2 m) (Figure 1). [2,3,17] The sensitizer/emitter concentrations were selected on the basis of solution experiments and the goal of achieving a significant absorptance at 543 nm. [3] The reaction afforded a rigid, tough material that was almost colorless if the dyes were omitted (Figure 2b; Figure S1 and Movie S1, Supporting Information). The dye-free nanodropletcontaining glassy polymer shows a transmittance of >75% over a path of 1 cm in the entire visible regime, which confirms that the matching of the refractive indices of the amorphous  . Energy level diagram and schematic explanation of the TTA-UC mechanism that exploits Pd(II) octaethylporphyrin (PdOEP) as light-harvester/sensitizer and 9,10-diphenylanthracene (DPA) as annihilator/emitter for green-to-blue photon upconversion. ISC: intersystem crossing; TTET: triplet-triplet energy transfer; GS: ground state; 1 S*, 3 S, 1 E*, 3 E: first singlet excited ( 1, *) or triplet state ( 3 ) of the sensitizer and emitter.
www.advmat.de www.advancedsciencenews.com matrix and the dispersed liquid phase allows suppressing interfacial scattering effects ( Figure S2, Supporting Information). Scanning electron microscopy (SEM) images of the dye-free ( Figure 2d) and dye-containing materials ( Figure S3a, Supporting Information) reveal similar nanophase-separated structures. The images show clearly that the continuous polymer phase contains the solvent (BMB) in the forms of nanodroplets that have feature dimensions of ≈20-100 nm and occasionally appear to merge into elongated channels. The absence of phase segregation observed in a reference material prepared without the surfactant ( Figure S3b, Supporting Information) and the differences in the small-angle X-ray scattering data ( Figure S4, Supporting Information) indicate that the surfactant is essential for the formation of the liquid nanodroplets even though no detailed structural information could be extracted from the latter measurement. The differential scanning calorimetry (DSC) trace displays only one glass transition at ≈48-52 °C ( Figure S5, Supporting Information), which is consistent with the formation of an amorphous, ethylene-glycol-plasticized poly(HEMA-co-MAA-co-TEG-diMA) copolymer.
On account of their nonpolar nature, the upconverting dyes were expected to preferably reside in the hydrophobic liquid nanodroplets (accounting for 10-15 wt% of the material) and exhibit solution-like TTA-UC. To test this hypothesis, the optical properties of dye-doped nanodroplet-containing polymer glasses were studied by means of continuous wave and timeresolved optical spectroscopy, employing BMB solutions of the dyes as reference system. The absorption and steady-state photoluminescence (PL) spectra of samples containing either PdOEP or DPA, or the DPA/PdOEP mixture match the spectra of corresponding solutions well (Figure 3a and Figure S6, Supporting Information), [17] demonstrating that the inclusion process did not affect the electronic properties of the chromophores. The dynamics of the characteristic blue DPA fluorescence at 431 nm and red PdOEP phosphorescence at 666 nm were studied using time-resolved PL spectroscopy and provided insights about the location of the dyes. The time-resolved intensity decay of the blue DPA emission shows a biexponential behavior in nanodroplet-containing polymers comprising either both dyes or DPA only (Figure 3b). This suggests the coexistence of two subpopulations of emitters having characteristic PL lifetimes of ≈6 and ≈14 ns, respectively. The faster lifetime is similar to the one observed for a DPA solution in BMB (7.2 ns, Figure 3b), whereas the longer decay appears to be associated with aggregated DPA molecules embedded in the rigid matrix. The relative contributions of the two decay components obtained from biexponential fits of the data indicate that ≈35% of the emitter molecules reside in the liquid phase. Their majority appears to be included in the solid matrix, where they neither contribute to nor hamper the TTA-UC process because the sensitizer molecules appear to reside predominantly in the liquid phase (vide infra).
Upon irradiation with green laser light at 543 nm (275 mW cm −2 ), the DPA/PdOEP doped nanodroplet-containing polymer displayed, even under ambient conditions, bright blue emission (Figure 3c  www.advmat.de www.advancedsciencenews.com containing either PdOEP or DPA only displayed red phosphorescence or scattered green light, respectively (Figure 3c, right). While dominated by the upconverted blue emission, the PL spectrum of the DPA/PdOEP-containing material shows a weak emission band at 666 nm (Figure 3d), which is associated with residual PdOEP phosphorescence and suggests that the sensitizer-emitter TTET is not quantitative. Time-resolved phosphorescence experiments revealed a single exponential decay with a lifetime τ ph = 1.50 ms, which is much higher than the value of 20 µs observed in the degassed BMB solution, but comparable to the radiative decay of PdOEP. [48] Thus, these findings suggest that the weak red luminescence belongs to a small portion of the sensitizer that is embedded in the polymer phase, where it is well protected from oxygen and solvent quenching ( Figure S7, Supporting Information). [3] No fast emission component could be detected, which indicates that for the sensitizer molecules incorporated in the liquid phase, the TTET yield is 100% and that emitter-to-sensitizers back energy transfer is negligible. [3] The same excitation conditions were used to acquire the upconversion emission spectra of degassed and non-degassed DPA/ PdOEP reference solutions in BMB (Figure 3d (2) showing bright blue upconverted emission. In a polymer that was only doped with PdOEP (3), red PL is observed through a 600 nm long-pass filter, while no PL is observed if only DPA is present (4). d) PL spectra of the DPA/PdOEP doped polymer (solid line) as well as degassed (dotted line) and non-degassed (dashed line) BMB solutions containing the same concentration of dyes (1.5 × 10 −2 m/2 × 10 −5 m). The laser stray light has been removed for clarity. The inset shows the decay of the upconverted emission from the polymer at 435 nm under modulated 532 nm excitation (30 mW cm −2 ). The red and the blue solid lines show fits of data with the analytical functions used to describe the decay dynamic at high and low densities of triplet exciton, respectively. e) TTA-UC quantum yield (Φ UC ) of the DPA/PdOEP doped nanodroplet-containing polymer (triangles) and a standard system in solution (dots) as a function of the excitation intensity. The vertical line marks the excitation intensity threshold I th , where Φ UC is half of its maximum value.
www.advmat.de www.advancedsciencenews.com emission, whereas the degassed solution emits bright upconverted blue light, indicating triplet quenching by molecular oxygen in the former. [11] The PL spectrum of the DPA/PdOEP nanodroplet-containing polymer mirrors the features observed for the degassed solution, and the upconverted emission intensities of the two samples are comparable, although the polymer had been prepared under ambient conditions. This material retained ≈50% of its initial upconversion efficiency after storage for six weeks or after 13.5 h of continuous operation under 10 mW cm −2 irradiation at 543 nm ( Figures S8-S10, Supporting Information). A comparison of plots of the upconverted emission intensity versus time ( Figures S9 and S10, Supporting Information) recorded for polymers prepared under ambient conditions and inert atmosphere, respectively, suggests that the former contain traces of oxygen, which are, however, rapidly consumed upon irradiation. Taken together, these results highlight that the glassy polymer matrix protects the liquid nanophase surprisingly well from atmospheric oxygen and that the use of a volatile liquid component is no fundamental obstacle for longevity, although the stability needs to be clearly improved before technological exploitation in devices is feasible. Strategies to achieve this include the use of dyes with better (photo) stability (DPA is well-known to undergo (photo)oxidation to an endoperoxide and dimerize, respectively [2,49,50] ), the addition of stabilizers, [34,51,52] a less volatile solvent, [53,54] and perhaps additional barrier layers. [7] The fact that the upconversion observed in the dye-doped nanodroplet-containing polymer is the result of TTA-UC was confirmed by time-resolved PL measurements (inset of Figure 3d). Under modulated excitation, the blue PL shows the expected complex decay dynamics that range from the nanosecond to the millisecond time scale. As commonly observed in TTA-UC systems, in the short-time range, the upconverted light decay is fast, on account of the initially high density of emitter triplets, which enhances their annihilation rate. As a result, the TTA yield (Φ TTA ) in this time domain is often close to unity. Conversely, at longer times, the triplet density becomes so low that TTA is a negligible dissipation channel. Thus, in this regime, the delayed emission decay follows a single exponential function, with a characteristic lifetime τ UC that follows the spontaneous recombination time of DPA triplets τ T = 2τ UC . [55] The analysis of the long-time component afforded a τ T of 726 µs. This value is typical for DPA in organic solvents, while a much higher τ T would be expected for TTA-UC in the rigid phase, where the spontaneous recombination of DPA triplets is significantly less efficient. [55] Importantly, τ T can also be used to analyze the short-time component; in this domain, the analysis of the decay dynamics results Φ TTA = 99% (see the Supporting Discussion Section in the Supporting Information). The large value of τ T and the fact that Φ TTA is close to unity are features observed in DPA/PdOEP solutions, and indicate that the efficient TTA-UC occurs only in the liquid phase of the nanodroplet-containing glassy polymer. This was further evidenced by cooling the material using liquid nitrogen, which led to the disappearance of the upconverted emission and to the appearance of strong PdOEP phosphorescence, due to suppression of the TTET in the frozen solvent. As expected, subsequent thawing completely restored the upconversion ability of the material (Figures S11 and S12, Supporting Information).
The TTA-UC quantum yield, Φ UC , of the DPA/PdOEP doped nanodroplet-containing glassy polymer, defined as the ratio between the numbers of emitted and absorbed photons, was first measured as function of the excitation intensity at 532 nm, using an air-free TTA-UC standard solution (Φ UC = 26%) as reference (see the Experimental Section and the Supporting Information). In order to compare the performance of the two systems, we show in Figure 3e the measured Φ UC as function of the excitation intensity. The log-log plots of the data reveal two distinct regimes; below a threshold value, the efficiency depends on the excitation power density, but at high excitation intensity, Φ UC of the DPA/PdOEP nanodroplet-containing glassy polymer is constant at ≈15%. This behavior is common and related to the bimolecular nature of the TTA step, which makes the efficiency dependent on the density of annihilating triplets and therewith the excitation intensity. However, at a sufficiently high excitation intensity, the TTA rate is much faster than the spontaneous decay of triplet states and TTA becomes their main recombination pathway with a Φ TTA of 100%. In this case, Φ UC reaches a constant value provided that the efficiencies of the intersystem crossing, TTET, and of the emitter fluorescence remain constant. [3,21] An instructive figure of merit is the threshold excitation intensity I th where the spontaneous decay rate of the emitter triplet equals the TTA rate and thus Φ UC reaches half of its maximum. [3,9,32,56] The DPA/PdOEP nanodroplet-containing glassy polymer exhibits an I th of ≈1.4 mW cm −2 , which corresponds to ≈0.9 suns (vertical line in Figure 3e and Figure S13, Supporting Information). This subsolar intensity threshold matches that of the reference solution well, and corroborates that the TTA-UC process in the nanodroplet-containing polymer occurs in the liquid phase. Another important figure of merit for solar harvesting applications is Φ UC at an excitation density of one solar irradiance; in the case of the DPA/PdOEP nanodroplet-containing polymer, this value is ≈9%, the highest ever reported for TTA-UC in a rigid matrix. [26,29] Indeed, upconversion was observed when the material was excited with noncoherent green light at 543 nm obtained from a filtered high-pressure xenon lamp (20-30 mW cm −2 ) or with collimated solar light passed through a 515 nm long-pass filter ( Figure S14, Supporting Information). Taken together, the present material concept displays characteristics that are useful for solar harvesting technologies and appears to be a promising framework for further development into stable devices, especially by replacing the model UC dyes used here by more (photo)stable chromophores and/or the addition of stabilizers. [4,7] To further explore the efficiency, Φ UC was measured under high excitation intensity at 532 nm (100 mW cm −2 ) using both an integrating sphere and a quasi-collinear PL setup for the detection of the emitted light (see the Experimental Section and the Supporting Information) and averaging data collected for three independently made samples. [3] This yielded Φ UC values of 16 ± 2% and 14 ± 3%, respectively, confirming the data shown in Figure 3e. To our best knowledge, this is by far the highest TTA-UC efficiency ever reported for a rigid polymer prepared and operated under ambient conditions. The fact that the efficiency is substantially lower than that of a degassed DPA/PdOEP reference solution in THF (Φ UC = 26%) is partially related to the reduced DPA fluorescence quantum yield www.advmat.de www.advancedsciencenews.com in BMB (78%) vis-à-vis THF (96%) ( Figure S26, Supporting Information) and likely due to quenching/photodegradation related to some residual oxygen, suggesting room for further improvement.
The generality and versatility of the material design concept were demonstrated by varying several parameters. PdOEP was replaced with PtOEP, another sensitizer that can be paired with DPA. This resulted materials with a maximum green-toblue Φ UC of ≈8%. A detailed spectroscopic study shows that the characteristics are very similar to the DPA/PdOEP-doped system (see the Supporting Information). Three other dye pairs were incorporated in the nanodroplet-containing glassy polymer in order to extend the process to other spectral ranges (Figure 4a,b). The sensitizer PdTPBP was combined with the emitters 9,10-bis(phenylethynyl) anthracene (BPEA) and tetratert-butylperylene (TBPe) to achieve red-to-green and red-to blue upconversion (Figure 4a), respectively. [2,21] The integration of the sensitizer/emitter pair palladium(II)-octabutoxyphthalocyanine (Pd(OBu) 8 Pc) and rubrene afforded a material that displayed red-to-yellow upconversion with a maximum Φ UC of 15 ± 3% (Figure 4b and the Supporting Information). Finally, the composition and mechanical properties of the matrix were varied. A thermoplastic upconverting material that could be reshaped upon heating above 50 °C was made by using the original recipe, but omitting the cross-linker TEG-diMA (Figure 4c). A rubbery material was made by replacing the methacrylic monomers with their acrylic counterparts, i.e., 2-hydroxyethyl acrylate, acrylic acid, and tetraethylene glycol diacrylate (Figure 4d). Other variations, such as the substitution of BMB with toluene, xylenes, or 1-[(2-ethylhexyl)oxy]-4-methoxybenzene also resulted in transparent upconverting materials ( Figure S16, Supporting Information).
In summary, we have developed a simple and versatile onestep approach to create upconverting nanostructured polymers that combine the desirable and tailorable mechanical, thermal, and optical properties of polymer matrices with photophysical properties that are normally only observed in degassed solutions of upconverting dyes. The glassy polymer matrix investigated provides surprisingly good protection from atmospheric oxygen and permits operation under ambient conditions. Some of the materials made displayed record upconversion efficiencies under conditions that would be experienced in solar energy harvesting applications. Considering the generality of this material design strategy, and the robustness of the morphology formation, the framework presented here appears not only to be useful for the fabrication of upconversion materials and devices, but should be exploitable to access other types of advanced optical materials.

Experimental Section
Synthesis of Upconverting Nanodroplet-Containing Glassy Polymers: Chemicals were purchased from Sigma-Aldrich, ABCR, Acros Organics, TCI (Tokyo Chemical Industry Co., Ltd.), or Inochem, Ltd. (Frontier Scientific, Inc.) and were used as received. All nanodropletcontaining polymers were prepared under ambient conditions without prior deoxygenation of the mixture or its single components, unless specifically stated otherwise. A 20 mL vial equipped with a stir bar was charged under ambient conditions with CTAB (250 mg, 5 wt%), HEMA (2.66 g, 53.2 wt%), TEG-diMA (175 mg, 3.5 wt%), BMB (500 mg, 10 wt%), ethylene glycol (750 mg, 15 wt%), and MAA (665 mg, 13.3 wt%) (total weight: 5 g). Benzoyl peroxide (10 mg, 0.2 wt%) was subsequently added and the mixture was heated to 80 °C and stirred for 10-15 min to appear clear and homogeneous, before it was filtered warm through a 0.2 µm polytetrafluoroethylene filter into either an optical glass cuvette or a glass vial containing N,N-dimethyl aniline (10 mg, 0.2 wt%). The still clear mixture was briefly shaken and left to stand at 20 °C until gelation (≈10-15 min), before it was placed into a water bath (15 °C), which served as a cooling medium. After  www.advmat.de www.advancedsciencenews.com the polymerization was complete and a transparent, hard material was obtained, which was either kept in a polymer film sealed cuvette and used for quantitative optical measurements or released from the glass vial by breaking the latter with a hammer. Materials of similar appearance were also obtained when BMB was replaced with toluene, xylenes, or 1-[(2-ethylhexyl)oxy]-4-methoxybenzene.
Green-to-blue upconverting nanodroplet-containing glassy polymers were prepared by substituting the BMB with a solution of PdOEP or PtOEP (2 × 10 −4 m) in BMB (500 mg, 10 wt% of total weight) and adding DPA (25 mg). Assuming a final density of 1 g cm −3 for the final material, the chromophore concentrations were 2 × 10 −5 m for PdOEP or PtOEP and 1.5 × 10 −2 m for DPA. A reduction of the DPA concentration by a factor of 2 resulted in a virtually unchanged upconversion intensity ( Figure S29, Supporting Information).
Material Characterization: Differential scanning calorimetry measurements were performed using a Mettler-Toledo DSC-1 equipped with a Huber TC100 cooling regulation system. Smallangle X-ray scattering spectra were recorded using an SMAX3000 pinhole camera from Rigaku Innovative Technologies, Auburn Hills, USA. The ≈2.5-mm-thick samples were measured in vacuum at room temperature. The scattering data are presented as a function of the momentum transfer q = 4πλ −1 sin(θ/2), where θ is the scattering angle and λ = 0.1524 nm is the photon wavelength. Raw data were processed according to standard procedures. Dynamic mechanical analyses were conducted on a TA Instruments DMA Q 800 using a frequency of 1 Hz and a heating rate of 10 °C min −1 .
SEM: The ultrastructure was investigated with a MIRA 3 LMH fieldemission electron microscope (Tescan, Brno, Czech Republic). To prevent charging, the samples were sputter coated by a thin layer of Pd/ Pt alloy (80:20) prior to imaging.
Optical Measurements: UV-vis absorption spectra were recorded on a Shimadzu UV-2401PC or on a Cary Varian 50 spectrometer. Steady-state PL spectra were acquired with a Photon Technology International C720 spectrophotometer equipped with a Hamamatsu R928P photomultiplier or with a nitrogen cooled charge-coupled device (Spex ≈2000) coupled to a polychromator Triax 190 from J-Horiba. Green-to-blue upconversion spectra were recorded using a 2 mW nonpolarized 543 nm helium-neon (HeNe, 543.5 ± 2 nm) laser (Thorlabs HGR020) or a focused doubled Nd:YAG (neodymium-doped yttrium aluminium garnet) diode pumped Coherent Verdi TEM 00 CW laser at 532 nm for excitation. 1/e 2 beam diameters of 0.83 mm (HeNe) or 0.80 mm (Nd:YAG) were measured by the knife blade method. The laser intensity was varied using reflective power density neutral filters and measured with an optical power meter (Thorlabs PM100USB with photodiode power sensor S120VC). The excitation laser light was further cleaned by using line filters of Edmund Optics (532 ± 5 nm) or Thorlabs (543.5 ± 10 nm), while for detection, the laser stray light was attenuated with the corresponding Edmund Optics notch filter. For the red-to-blue, red-to-green, and dark-red-toyellow upconversion, a 635 or a 670 nm solid-state laser diode from Roithner Lasertechnik was used as excitation sources. All spectra were corrected for the instrumental optical response.
For time-resolved upconversion measurements, samples were excited at 532 nm by modulating a Nd:YAG laser with a TTi TG5011 wavefunction generator. Spectra were recorded by nitrogen cooled photomultiplier (Hamamatsu R5509-73) coupled with a high-speed amplifier (Hamamatsu C5594), a 74100 Cornestone 2601/4 (ORIEL) monochromator, and a PCI plug-in multichannel scaler ORTEC 9353 100 ps time digitizer/MCS in a photon counting acquisition mode. Fluorescence and phosphorescence time-resolved spectra were obtained by using the III harmonic of Nd:YAG Continuum Minilite laser (10 ns pulse width), detecting the luminescence decay by an Edinburgh LP90 flash photolysis setup.
The upconversion quantum efficiency Φ UC of the nanodropletcontaining polymers was determined relative to the maximum upconversion quantum yield of the DPA/PtOEP UC pair in deaerated tetrahydrofuran ([DPA] = 10 −2 m, [PtOEP] = 10 −4 m) used as secondary standard in quasi-collinear and integrating sphere PL setups. This method has practical advantages over using a primary standard and is an accepted practice in the field. [28,55,57] The theoretical maximum UC quantum efficiency is 0.50, since the TTA-UC process uses two low energy photons to produce one high-energy photon. The maximum Φ UC of the secondary standard is 0.26, under excitation at 532 nm and a power density of 100 mW cm −2 ; it was measured and calculated by using the laser dye rhodamine 6G (10 −5 m in ethanol) as reference material. [3] The recorded spectra were corrected for the setup optical response. The secondary standard (or "reference") solution was freshly prepared in a glovebox under nitrogen atmosphere (O 2 -level <0.1 ppm) and the cuvette was sealed with a hot melt adhesive before being characterized prior to each set of measurements outside of the glovebox. To confirm the results obtained, Φ UC measurements were also performed using an integrating sphere on a series of independently made samples (see the Supporting Experimental Section in the Supporting Information for further details). [37]

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.