Triplet–Triplet Annihilation Upconversion in Broadly Absorbing

Research Article www.acsami.org

Cite This: ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Triplet−Triplet Annihilation Upconversion in Broadly Absorbing Layered Film Systems for Sub-Bandgap Photocatalysis Anna L. Hagstrom,†,§ Seunghyun Weon,‡,§ Wonyong Choi,*,‡ and Jae-Hong Kim*,† †

Department of Chemical and Environmental Engineering, Yale University, New Haven, Connecticut 06511, United States Division of Environmental Science and Engineering, Pohang University of Science and Technology (POSTECH), Pohang 37673, Republic of Korea

‡

Downloaded via UNIV OF LOUISIANA AT LAFAYETTE on April 2, 2019 at 07:21:46 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: Upconversion (UC) of sub-bandgap photons extends the effective light absorption range of photovoltaic and photocatalytic devices, allowing them to reach higher conversion efficiencies. Recent advances in polymer host materials make it possible to translate triplet−triplet annihilation (TTA)-UC, the UC mechanism most suitable for this purpose, to solid films that can be integrated into devices. The promise of these films is currently limited by the narrow light absorption of TTA-UC sensitizer chromophores, but incorporating multiple sensitizers into layered film systems presents a promising strategy for producing UC materials with broadened light absorption. This strategy is herein applied for photocatalytic air purification, demonstrating its use in a realworld application for the first time. We superimpose optimized red-to-blue and green-to-blue UC films within dual-layer systems and develop a new photocatalyst compatible with their fluorescence emission. By integrating the dual-layer UC film systems with films of this photocatalyst, we produce the first devices that use TTA-UC to harness both red and green subbandgap photons for hydroxyl radical generation and photocatalytic degradation of gaseous acetaldehyde, a model volatile organic compound (VOC). Under white light-emitting diode excitation, the dual-layer film systems’ broadened light absorption enhances their devices’ photocatalytic degradation efficiency, enabling them to degrade twice as much acetaldehyde as their single-sensitizer counterparts. We show that as a result of the different absorption profiles of the two sensitizers, the film order significantly impacts UC fluorescence and VOC degradation. By probing the influence of the excitation light source, excitation geometry, and chromophore spectral overlap on the film systems’ UC performance, we propose a framework for the design of multilayer TTA-UC film systems suitable for integration with a variety of photovoltaic and photocatalytic devices. KEYWORDS: triplet−triplet annihilation, upconversion, broadband absorption, sub-bandgap photocatalysis, photocatalytic VOC degradation

■

INTRODUCTION Photon upconversion (UC) enables photovoltaic and photocatalytic devices to harvest low-energy sub-bandgap light that would otherwise be wasted, thereby circumventing conventional limits on the efficiencies that they are able to reach.1 Because the semiconductors responsible for the light absorption of these devices can only absorb photons with energy above that of their band gap, Eg, they inherently transmit and waste a large fraction of the energy present in white-light sources such as sunlight. As Eg increases, this subbandgap transmission takes a progressively greater toll on device performance, becoming the dominant constraint on the energy conversion efficiency when Eg > 1.3 eV.2 UC is able to reduce these transmission losses by converting pairs of subbandgap photons into higher-energy photons that a device can harvest, thereby extending its effective absorption range without the need to modify its semiconductor absorber. Easily integrated UC materials could lead to dramatic improvements © XXXX American Chemical Society

in the performance of wide-band gap semiconductor technologies ranging from emerging thin-film solar cells (e.g., perovskite and polymer photovoltaics) to photocatalytic devices for environmental remediation. In recent years, significant strides have been made toward translating triplet−triplet annihilation (TTA)-UC,3 the UC mechanism with the greatest promise in such devices,4 into suitable materials.5 Under low-intensity, noncoherent excitation such as solar radiation, TTA-UC operates orders of magnitude more efficiently than lanthanide-based UC,6,7 the UC mechanism that was first considered for this purpose.8 The superior efficiency of TTA-UC is due in part to the strong light absorption of the organometallic sensitizer chromophores (e.g., metalloporphyrins) that are employed to harvest low-energy Received: January 29, 2019 Accepted: March 20, 2019

A

DOI: 10.1021/acsami.9b01945 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces photons.6,9 These sensitizers funnel their excitation energy to acceptor chromophores, typically polycyclic aromatic hydrocarbons, which collaborate to emit upconverted anti-Stokes fluorescence. Since this takes place through a series of shortrange (≤1 nm) Dexter energy transfers involving at least one chromophore in a triplet excited state,10 studies of TTA-UC were initially conducted in thoroughly deoxygenated organic solvents. Solution-based systems facilitate diffusive collisions between TTA-UC chromophores, and deoxygenation protects the chromophores’ triplet excited states from being quenched by oxygen (3O2) before they can undergo the necessary energy transfers.11 However, sealed containers of deoxygenated chromophore solutions are far from ideal for device integration. Of the solid-state materials that have been studied as alternative chromophore hosts,12 rubbery polymers with low glass transition temperatures currently show the greatest promise. Their matrices provide some intrinsic protection from oxygen, and the mobility of their polymer chains facilitates enough chromophore diffusion to enable highly efficient UC.13,14 To date, rubbery polymerse.g., polyurethane (PU) and polyacrylatedoped with TTA-UC chromophores have been successfully employed in proof-ofconcept studies to extend the light absorption of photovoltaics,15 photoelectrochemical water-splitting cells,14,16 and photocatalysts for environmental remediation.17 Moreover, films of these polymers can be fabricated using simple, open-air procedures,13,15,17,18 including solution-processing techniques such as spin-coating,19 potentially providing a path toward scalable production of UC materials for device integration. At present, the promise of these materials is primarily limited by the narrow spectral response of their sensitizer chromophores. According to detailed energy-balance calculations, the ability to upconvert sub-bandgap light could enable single-junction solar devices to reach energy conversion efficiencies approaching 50%.20−22 However, these calculations assume idealized UC systems with light absorption spanning hundreds of nanometers,22 whereas the absorption bands of typical metalloporphyrin sensitizers are only ∼20 nm wide.23 As a result, conventional TTA-UC systems harvest only a miniscule fraction of the incident sub-bandgap light, drastically limiting their potential to increase device efficiencies.24,25 To address this limitation, some researchers have synthesized broadly absorbing sensitizers by covalently binding multiple chromophore units.26 More recently, several studies have proposed using broadly absorbing semiconductor nanocrystals as alternative sensitizers for TTA-UC,27−30 but the extensive spectral overlap between the nanocrystals’ broad light absorption and the emission of their paired acceptors enables them to reabsorb and quench UC fluorescence. In a more promising approach, films of these nanocrystals laid beneath TTA-UC systems have recently been shown to broaden the sensitizers’ effective light absorption by recycling the energy of the transmitted sub-bandgap photons into fluorescence that the sensitizers can absorb.16,31 Certain fluorescent dyes mixed into solution-based TTA-UC systems can play a similar role.32 Simultaneous use of multiple sensitizers is arguably the most straightforward strategy for producing TTA-UC systems with a broadened spectral response. This approach capitalizes on the inherent adaptability of the TTA-UC mechanism, which allows for independent tuning of absorption and emission wavelengths through choice of sensitizer and acceptor chromophores, respectively. Although studies of TTA-UC have thus far employed over 100 different sensitizers and over 30

different acceptors,9 only a handful have employed more than one sensitizer simultaneously.18,33−35 It is noteworthy that most of these multisensitizer studies focused on solution-based TTA-UC systems that, as discussed above, are inherently illsuited to device integration.33−35 Because solid-state systems typically contain chromophore concentrations an order of magnitude higher than their solution-based counterparts, trends in their TTA-UC performance can differ considerably. A recent study by our group, which introduced the first dualsensitizer solid-state system, showed that as a result of the high sensitizer concentrations required for effective light absorption within thin TTA-UC films, adopting a “multijunction” approach by isolating the two sensitizers within distinct superimposed film layers was far more effective than doping both sensitizers into a single film.18 The ability to overlay multiple films presents a distinct advantage over solution-based systems, which would require intricate custom vessels to achieve a similar layered effect. Herein, we develop improved dual-layer red/green-to-blue UC film systems and demonstrate their potential in photocatalytic air purification, employing layered UC film systems in a real-world application for the first time. Moving beyond laser excitation toward more realistic conditions, we characterize their performance under white light-emitting diode (LED) excitation, thoroughly exploring the effects of the individual films’ sensitizer concentrations as well as their order within the dual-layer systems (i.e., which of the two films faces the excitation light source). We develop a new photocatalyst compatible with the fluorescence of the benchmark acceptor perylene, integrate this photocatalyst with the optimized duallayer UC film systems, and demonstrate the resulting devices’ enhanced photocatalytic volatile organic compound (VOC) degradation efficiency under white LED excitation. Taking advantage of the versatility of this multilayer approach, we then characterize the performance of dual-layer UC film systems using a range of different excitation light sources, excitation geometries, and chromophore systems. Collectively, the results presented in this study provide a framework for the development of multilayer TTA-UC film systems suitable for integration with a wide variety of photocatalytic and photovoltaic devices.

■

RESULTS AND DISCUSSION Selection of Single-Sensitizer UC Film Layers. Using benchmark TTA-UC chromophores, we produced singlesensitizer red-to-blue and green-to-blue TTA-UC films for incorporation into dual-layer systems. Sensitizers palladium(II) meso-tetraphenyltetrabenzoporphyrin (PdTPBP) and palladium(II) octaethylporphyrin (PtOEP) harvest red and green photons, respectively, through Q-band (S0 → S1) absorption peaks centered at approximately 630 and 535 nm and then transfer their energy to the acceptor perylene, which emits blue anti-Stokes fluorescence. Figure 1a illustrates the mechanism by which this occurs. Photon absorption excites the sensitizers to singlet excited states (1S*) that rapidly undergo intersystem crossing (ISC) to longer-lived triplet excited states (3S*) as a result of the spin−orbit coupling induced by the porphyrins’ metal centers. Triplet−triplet energy transfer (TTET) from 3S* to ground-state perylene (1A0), which is present in excess, forms a reservoir of perylene molecules in long-lived triplet excited states (3A*). When pairs of these 3A* undergo TTA, they combine their excitation energy to produce high-energy perylene singlet excited states B

DOI: 10.1021/acsami.9b01945 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

longpass (LP) filter to limit its emission to λ > 500 nm (Figure 2a). This allowed us to excite PtOEP and PdTPBP without directly exciting perylene, thereby isolating the films’ antiStokes UC fluorescence. As shown in Figure 2b, films containing 0.50 mM PtOEP or PdTPBP emitted UC fluorescence comparable in peak intensity, but increasing the concentration to 1.0 mM improved the TTA-UC performance of the PtOEP films and worsened that of the PdTPBP films, consistent with previous results obtained using red and green laser excitation.18 Although PdTPBP and perylene are a highly effective chromophore pairin fact, dilute solutions of these two chromophores exhibit the highest TTA-UC efficiencies reported to date36high concentrations of PdTPBP consistently prove problematic. Moving forward toward device integration, we therefore fixed the PtOEP and PdTPBP concentrations of all films at 1.0 and 0.50 mM, respectively, capitalizing on the ease of optimizing individual sensitizer concentrations within layered film systems. To account for inevitable variation between individual films, we present all spectra herein as the average obtained using several different films. The stark difference between the UC performance of the 1.0 mM PtOEP and 1.0 mM PdTPBP films stemmed from the overlap between the wavelengths of perylene fluorescence and PdTPBP Soret-band absorption (Figure 1b). This spectral overlap promotes both acceptor-to-sensitizer energy backtransfer and UC fluorescence reabsorption, quenching processes that thwarted our previous attempts to achieve strong UC performance within concentrated dual-sensitizer films and inspired our turn to the current layered approach.18 To probe the impact of PtOEP and PdTPBP concentration on the films’ fluorescence, we compared the Stokes fluorescence of the UC films under the direct excitation of perylene at 340 nm to that of sensitizer-free control films containing only perylene. As shown in Figure 2c, both sensitizers quenched the fluorescence of perylene to some extent. This quenching, which increased with increasing sensitizer concentration (Figure 2c), is consistent with the nonradiative transfer of energy from perylene singlet excited states (1A*) back to ground-state PdTPBP and PtOEP, though slight absorption of incident 340 nm excitation by the sensitizers could also contribute. This effect was far more severe in the PdTPBP films, suggesting that the strong overlap between perylene emission and PdTPBP Soret-band absorption may have promoted strong perylene-to-PdTPBP back-transfer. The same overlap also caused greater reabsorption of perylene fluorescence within the PdTPBP films as a result of their drastically increased absorbance between 440 and 475 nm (Figure 2d). This secondary inner filter effect was evident in the PdTPBP films’ disproportionate quenching of the perylene fluorescence peak centered at ∼450 nm (Figure 2c, inset). Because high concentrations of PtOEP had little effect on film absorbance in this wavelength range (Figure 2d), they had little effect on the shape of the perylene emission spectrum. This difference in reabsorption is also evident in the shape of the films’ UC emission spectra in Figure 2bwhile the 0.5 mM PtOEP and 0.5 mM PdTPBP films emitted comparable fluorescence between 480 and 500 nm, reabsorption within the PdTPBP films significantly reduced their UC fluorescence at shorter wavelengths. Characterization of Dual-Layer UC Film Systems. Overlaying 1.0 mM PtOEP and 0.50 mM PdTPBP films produced dual-layer UC film systems that vastly outperformed

Figure 1. (a) Jablonski diagram illustrating the steps in the TTA-UC mechanism: (1) Low-energy photon absorption, (2) intersystem crossing, (3) triplet−triplet energy transfer, (4) triplet−triplet annihilation, and (5) high-energy photon emission. (b) Absorption spectra (solid) and emission spectra (dashed) and (c) chemical structures of the sensitizers PtOEP (orange) and PdTPBP (blue) and the acceptor perylene (yellow).

(1A*) that emit blue fluorescence. As shown in Figure 1b, this fluorescence largely falls within the transparent window between the sensitizers’ Q-band absorption peaks and their higher-energy Soret absorption peaks (S0 → S2, ∼390 and 440 nm), a characteristic absorption feature that minimizes detrimental spectral overlap, making metallated porphyrins particularly favorable as TTA-UC sensitizers. To select optimal sensitizer concentrations for our dual-layer film systems, we characterized the performance of singlesensitizer green-to-blue and red-to-blue UC films under white LED excitation. We fabricated rubbery PU films containing perylene paired with either PtOEP or PdTPBP, encasing the films between thin sheets of glass during fabrication to deter the infiltration of ambient oxygen through the mobile chains of the PU matrix. This PU host has been shown to enable highly efficient TTA-UC and, using the same fabrication method reinforced by epoxy seals around the exposed film edges, to yield films with long-term photostability.13,15 Selecting chromophore concentrations similar to those previously employed in this host,15,18 we fixed the perylene concentration of all films at 18 mM (0.43 wt %) and compared the performance of films containing sensitizer concentrations of 0.50 and 1.0 mM (0.034 and 0.068 wt % for PtOEP and 0.043; 0.087 wt % for PdTPBP). As shown in Figures 2a and S1, these concentrations enabled the films to absorb a large fraction of the incident light within the sensitizers’ Q-bands. We compared the TTA-UC performance of these singlesensitizer films under excitation with a white LED, using an aspheric condenser lens to reduce its divergence and a 525 nm C

DOI: 10.1021/acsami.9b01945 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

Figure 2. (a) Absorption spectra of UC films (solid traces) and normalized emission spectrum of a white LED through a 525 nm LP filter (dashed trace). Each absorption spectrum is the average obtained from at least four films. (b) UC fluorescence of UC films under white LED excitation (525 nm LP, 94.1 mW cm−2) at an angle of ∼45°. Each emission spectrum is the average obtained from nine films. (c) Normalized Stokes fluorescence of UC and perylene-only films under direct excitation of perylene at 340 nm. Each emission spectrum is the average obtained from four scans of at least four different films. Inset: ratio of fluorescence intensity at 451 and 472 nm. (d) Absorption spectra of UC films as compared to perylene-only control films. Each spectrum is the average obtained from at least four films. All films contain 18 mM perylene, and all shaded regions denote one standard deviation.

the single-sensitizer films and dual-layer film systems mirrored their performance under white excitation (Figure 3b). Under red excitation, the PdTPBP−PtOEP system emitted shortwavelength (λ < 475 nm) UC fluorescence 23% weaker than that of a lone PdTPBP film, demonstrating some fluorescence reabsorption within the underlying PtOEP film (Figure 3c, left panel). Under green excitation, however, the PtOEP−PdTPBP system emitted short-wavelength (λ < 475 nm) UC fluorescence 38% weaker than that of a lone PtOEP film (Figure 3c, center panel), demonstrating far stronger fluorescence reabsorption within the underlying PdTPBP film. This confirms that the difference in the emission spectra of the two dual-layer film systems under white LED excitation (Figure 3b) resulted from the difference in the severity of UC fluorescence reabsorption within their PtOEP and PdTPBP films. A more detailed analysis incorporating control films showed that in each dual-layer system, the chromophores present in the bottom film affected the fraction of UC fluorescence that we were able to collect from the top film. While placing a blank, chromophore-free PU film beneath a PtOEP film had little effect on its UC fluorescence under green excitation, replacing this blank film with control films containing only perylene or PdTPBP caused considerable fluorescence reabsorption below 475 nm because of the spectral overlap between perylene emission and perylene/PdTPBP absorption (Figures 1b and S3a). In UC films containing both PdTPBP and perylene, the combined effects of the two chromophores gave rise to even stronger reabsorption (Figure S3a). PtOEP

their individual layers under broadband LED excitation. To best assess the promise of these film systems for incorporation into sub-bandgap photocatalyst devices, in which LED excitation and photocatalyst sensitization occur on opposite sides of the UC films,17 we employed back-face LED excitation, exciting the films from one direction and collecting their fluorescence from the other. As illustrated in Figure 3a, two possible film orientations exist for the dual-layer systems, one in which the PdTPBP film is positioned above the PtOEP film (PdTPBP−PtOEP), closer to the LED excitation, and the other in which it is positioned below (PtOEP−PdTPBP). Throughout this study, the film listed first in each dual-layer system is the top filmthat is, the film closest to the LED excitation source. As shown in Figure 3b, the film order significantly affected the TTA-UC performance under white LED excitation (525 nm LP, 95.2 mW cm−2). Although both dual-layer systems emitted much stronger peak UC fluorescence than their constituent single-sensitizer films, the PtOEP−PdTPBP system emitted markedly weaker fluorescence below 475 nm (Figure 3b) because its film orientation required the UC fluorescence of the PtOEP film to pass through the underlying PdTPBP film prior to collection, thereby giving rise to far greater reabsorption. To directly characterize UC fluorescence reabsorption within the two dual-layer film systems, we used red and green LEDs to approximate the PdTPBP and PtOEP excitation provided by the white LED (Figure S2). As shown in Figure 3c, under simultaneous red and green LED excitation at 31.5 and 30.8 mW cm−2, respectively, the TTA-UC performance of D

DOI: 10.1021/acsami.9b01945 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

Figure 3. (a) Schematic illustrating the two possible dual-layer UC film systems relative to the directions of LED excitation and UC fluorescence collection. In each system, the top film is listed first. (b) UC fluorescence of individual and layered UC films under back-face white LED excitation (525 nm LP, 95.2 mW cm−2) at an angle of ∼45°. Each emission spectrum is the average obtained from four PtOEP films and/or four PdTPBP films, and the dashed trace marks the summed average emission of the PtOEP and PdTPBP films. (c) UC fluorescence of individual and layered UC films under back-face red (31.5 mW cm−2), green (30.8 mW cm−2), and simultaneous red and green LED excitation (525 nm LP). Each emission spectrum is the average obtained from four PtOEP films and/or four PdTPBP films. The dashed traces mark the summed average emission of each dual-layer film system under red and green excitation. All films contain 18 mM perylene with either 1.0 mM PtOEP or 0.50 mM PdTPBP, and all shaded regions denote one standard deviation.

hydroxyl groups to serve as •OH precursors and VOC adsorption centers. Although N-TiO2 (Eg dependent on dopant concentration39,40) has been previously shown to degrade acetaldehyde,41 it showed a very low degradation efficiency under blue LED excitation (Table S1). TiO2 efficiently degrades a variety of VOCs because of its high surface hydroxyl group concentration and the strong oxidative power of its VB.42,43 In N-TiO2, the mixing of N and O p orbitals reduces Eg by generating dopant energy levels with lower potential than the TiO2 VB; while this enables visiblelight absorption, it also gives rise to a decreased degradation efficiency.44−46 We improved upon the poor performance of these existing photocatalysts by modifying tungsten oxide (WO3) to produce a new composite photocatalyst capable of harnessing perylene fluorescence for efficient acetaldehyde degradation. Unlike CdS, CN, and N-TiO2, WO3 has a VB that generates holes with sufficient oxidative power (3.0 VNHE) to degrade acetaldehyde either directly or through •OH generation.46 However, the Eg of pure WO3 restricts its light absorption to λ < 460 nm, preventing it from being efficiently activated by our UC films’ fluorescence. Doping WO3 with certain foreign atoms, most commonly metals (e.g., Sb, Nb, Fe, Mn, and Mo), has been shown to enhance its visible-light absorption through alterations in its electronic structure,47−51 and interstitial doping with alkali metals has shown particular promise.52,53 Employing this strategy, we doped WO3 with low concentrations of potassium (K-WO3, Figure 4a). As confirmed by diffuse reflectance spectroscopy, this extended its visible-light

absorption, in contrast, has little overlap with perylene emission (Figure 1b). As a result, placing a PtOEP-only control film beneath a PdTPBP UC film had little effect on its UC fluorescence under red excitation (Figure S3c), confirming that the perylene in the PtOEP UC films was responsible for their fluorescence reabsorption below 475 nm (Figure 3c, left panel). In addition, as shown in Figures 3c and S3b,d, the top film in each dual-layer system slightly decreased the UC fluorescence of the underlying film. The magnitude of this effect proved independent of the top film’s chromophore composition (Figure S3b,d), suggesting that reflection off of its glass surface (Figure S1) prevented a small fraction of the incident excitation from reaching the bottom film. Photocatalyst Development and Characterization. In pursuit of a photocatalyst compatible with the UC fluorescence of our film systems and suitable for device integration, we first tested the VOC degradation performance of widely used visible-light-active photocatalysts cadmium sulfide (CdS), polymeric carbon nitride (CN), and nitrogen-doped TiO2 (N-TiO2). We measured the efficiency with which each photocatalyst degraded acetaldehyde, a prototypical VOC, within a closed-circulation gas-phase reactor over the course of 30 min of excitation with a blue LED (16 mW cm−2) passed through a 455 nm LP filter to approximate perylene fluorescence (Table S1). Despite their ability to absorb this excitation, CdS (Eg = 2.5 eV37) and C3N4 (Eg = 2.64 eV38) were unable to degrade gaseous acetaldehyde because their valence band (VB) potentials are insufficient for hydroxyl radical (•OH) production (2.8 VNHE) and their surfaces lack E

DOI: 10.1021/acsami.9b01945 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

Figure 4. (a) XPS spectra analyzing the potassium 2p orbitals of bare WO3 and WO3 doped with potassium (K-WO3) and surface-loaded with ND (ND/K-WO3); (b) diffuse reflectance spectra of bare WO3, K-WO3, and ND/K-WO3. Spectra were processed using the Kubelka−Munk transform, F(R∞) = (1 − R∞)2/2R∞, where R∞ denotes diffuse reflectance. Inset: corresponding Tauc plot for the WO3 and ND/K-WO3 spectra. (c) XRD spectra and (d) ATR−FT-IR spectra of WO3, K-WO3, ND/K-WO3, and ND. The potassium and ND concentrations of each photocatalyst were fixed at 0.5 and 8 wt %, respectively, with respect to WO3.

Figure 5. (a) HR-TEM image of ND/K-WO3, corresponding EELS mapping of (b) W, (c) C, and (d) O, and corresponding spot-profile EDS spectra characterizing the elemental composition (provided in tables) of the (e) ND cocatalyst and (f) K-WO3 photocatalyst.

absorption range (Figure 4b), decreasing its Eg from 2.7 to 2.6 eV (Figure 4b, inset). X-ray photoelectron spectroscopy (XPS) analysis showed that the VB edges of WO3 and K-WO3 were equal in potential, indicating that potassium doping shifted the WO3 conduction band (CB) potential without affecting its VB potential (Figure S4). The photocatalysts’ X-ray diffraction (XRD) patterns confirmed that the incorporation of potassium

distorted the WO3 lattice without disrupting its overall monoclinic crystal structure (Figure 4c). As a secondary benefit, the KOH solution used for potassium doping also increased the concentration of hydroxyl groups on the K-WO3 surface, amplifying the intensity of the O−H band (3364 cm−1) in its Fourier-transform infrared (FT-IR) spectrum as compared to that of WO3 (Figure 4d) and promoting VOC F

DOI: 10.1021/acsami.9b01945 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

its limited light absorption. Potassium doping increases the efficiency of photocatalytic acetaldehyde degradation, and ND/K-WO3 containing 0.5 wt % potassium showed the highest degradation efficiency. Beyond a doping concentration of 0.5 wt %, however, the photocatalytic efficiency decreased with increasing potassium concentration, eventually dropping to zero when this concentration reached 10 wt %, at which point the high potassium levels and the high alkalinity required during the photocatalyst synthesis disrupted the WO3 crystal structure (Figure S5).58,59 Moving forward, we fixed the potassium concentration in ND/K-WO3 at 0.5 wt % and fabricated photocatalyst films suitable for device integration. Sub-Bandgap Photocatalyst Devices for Acetaldehyde Degradation. Using sub-bandgap photocatalyst devices that paired dual-layer UC film systems with films of ND/KWO3, we successfully demonstrated the promise of the multilayer UC film strategy in photocatalytic VOC degradation. We assembled a series of photocatalyst devices by using polyamide spacer tape (thickness ≈ 100 μm) to adhere a UC film or film system to each photocatalyst film, thereby enabling UC fluorescence to sensitize ND/K-WO3 and degrade acetaldehyde flowing between the two layers (Figure S6). We evaluated the photocatalytic degradation efficiency of each device using the same reactor setup employed for photocatalyst screening. Each degradation experiment consisted of 30 min of circulation to bring the contents of the reactor to adsorption/desorption equilibrium followed by 90 min of white LED excitation (525 nm LP, 130.1 mW cm−2). Figure 6 shows the time profiles of various sub-bandgap photocatalyst devices’ acetaldehyde degradation and accompanying CO2 generation, which confirm that all devices pairing ND/KWO3 with UC films successfully degraded acetaldehyde to CO2 (CH3CHO + 5/2O2 → 2CO2 + 2H2O). The device containing the PdTPBP−PtOEP dual-layer film system achieved 6 ppmv of acetaldehyde degradation during the 90 min reaction period, which was 2−3 times higher than that of the devices containing single-sensitizer films. Control experiments separately testing the photocatalyst films and UC films confirmed that the devices operated through the interaction between their UC films and the photocatalyst (Figure S7). Devices containing control films doped with only PtOEP, PdTPBP, or perylene exhibited no photocatalytic activity (Figure 6), confirming that UC fluorescence was responsible for sensitizing ND/K-WO3. A device containing WO3 in place

degradation by trapping photogenerated holes to generate surface-bound •OH. Since the CB of WO3 is incapable of single-electron oxygen reduction because of its positive potential (E0 = −0.33 VNHE for O2/O2−•),46,54 WO3 is far more effective when loaded with cocatalysts that facilitate multielectron oxygen reduction (E0 = +0.68 VNHE for O2/H2O2).55 Nanodiamond (ND), a costeffective cocatalyst whose interfacial charge separation abilities rival those of platinum, has recently been shown to enhance the photocatalytic degradation of both aqueous and gaseous pollutants.46,56 We successfully loaded the surface of K-WO3 with 8 wt % ND (ND/K-WO3), which we confirmed through electron energy loss spectroscopy (EELS) mapping and elemental analysis of high-resolution transmission electron microscopy (HR-TEM) images of the photocatalyst (Figure 5). Because ND consists of a diamond (sp3 carbon) core surrounded by a thin layer of graphitic (sp2) carbon, it is transparent across the visible region and thus does not interfere with the light absorption of K-WO3 (Figure 4b). This presents a marked advantage over more common carbon-based cocatalysts such as graphene, which are difficult to deposit in layers thin enough to be optically transparent.57 Table 1 summarizes the acetaldehyde degradation efficiency of ND/K-WO3 doped with a range of potassium concenTable 1. Photocatalytic Degradation of Gaseous Acetaldehyde by ND/K-WO3 Containing Various Potassium Doping Concentrations Photocatalyst WO3 ND/WO3 ND/K-WO3 ND/K-WO3 ND/K-WO3 ND/K-WO3

(0.1 wt %) (0.5 wt %) (1 wt %) (10 wt %)

CH3CHO degradation (ppmv)a

kd (102 min−1)b

0 3.09 7.01 8.73 4.39 0

0 0.80 1.93 2.98 1.09 0

[CH3CHO]0 = 15 ppmv. bCalculated by fitting 30 min of experimental data to a pseudo−first-order model. a

trations under excitation, as above, with a blue LED (455 nm LP, 16 mW cm−2) approximating perylene fluorescence. Bare WO3 showed no photocatalytic activity, and ND/WO3 showed only moderately efficient acetaldehyde degradation because of

Figure 6. (a) Acetaldehyde (CH3CHO) degradation and (b) CO2 production, under white LED excitation (525 nm LP, 130.1 mW cm-2), of subbandgap photocatalyst devices pairing a ND/K-WO3 photocatalyst film (unless otherwise indicated) with a PdTPBP UC film, a PtOEP UC film, a dual-layer UC film system (top film listed first), or a control film containing only perylene, only PtOEP (PtOEP′), or only PdTPBP (PdTPBP′) under white LED excitation. Concentrations were recorded once per minute. [CO2] is presented as the average of ten successive measurements, and error bars denote one standard deviation. Films contain PtOEP, PdTPBP, and perylene concentrations of 1.0, 0.50, and 18 mM, respectively. G

DOI: 10.1021/acsami.9b01945 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

lower than that of its inverse (i.e., PdTPBP−PtOEP) counterpart (Figure 6). In fact, this device barely even outperformed the single-sensitizer PtOEP devicethe disadvantage of its diminished short-wavelength fluorescence nearly offset the advantage of its elevated peak fluorescence (Figure 3b). This underscores the importance of minimizing detrimental inner filter effects when designing layered UC film systems for device integration. Since •OH generation is one of the primary mechanisms by which photocatalysts degrade VOCs, we tested the ability of an optimized PdTPBP−PtOEP sub-bandgap photocatalyst device to produce •OH using chemical trapping methods that rely on DMPO (5,5-dimethyl-1-pyrroline-N-oxide) and coumarin. These probes react with •OH to produce adducts OHDMPO and 7-hydroxycoumarin. To provide better probe solution access, we modified the orientation of the device’s films as shown in Figure S8. After 30 min of white LED excitation (525 nm LP, 130.1 mW cm−2), electron paramagnetic resonance (EPR) spectroscopy detected OH-DMPO within the DMPO solution (Figure 8a) and fluorescence spectroscopy detected 7-hydroxycoumarin within the coumarin solution (Figure 8b). This confirmed •OH production, making this the first successful use of a layered TTA-UC film system to generate •OH under broadband excitation. A control

of ND/K-WO3 likewise exhibited no photocatalytic activity (Figure 6), confirming that bare WO3 was not suitable for use in these sub-bandgap photocatalyst devices. We also fabricated a second set of devices containing ND/K-WO3 coated onto conductive fluorine-doped tin oxide (FTO) instead of glass and monitored their photocurrent generation under white LED excitation (525 nm LP, 130.1 mW cm−2), which directly demonstrates their photocatalyst films’ activation by UC fluorescence absorption. As anticipated, the devices’ relative photocurrent generation was consistent with their relative acetaldehyde degradation (Figures 6 and 7).

Figure 7. (a) Photocurrent generation of sub-bandgap photocatalyst devices pairing a film of ND/K-WO3 coated onto FTO with a PtOEP UC film, PdTPBP UC film, or dual-layer UC film system (top/front film listed first) under periodic white LED excitation (525 nm LP, 130.1 mW cm−2, 30−60 and 90−120 s). UC films contain 18 mM perylene with either 1.0 mM PtOEP or 0.50 mM PdTPBP. Aqueous 0.1 M NaCl solution was employed as the electrolyte, and a bias voltage of +0.5 V was applied. Photograph of this experimental setup (b) without any filter and (c) through a 500 nm shortpass filter.

The inner filter effects discussed above caused the order of the dual-layer film systems’ PdTPBP and PtOEP films to affect their ability to sensitize ND/K-WO3, thereby drastically affecting the performance of their photocatalyst devices (Figures 6 and 7). Although the peak UC fluorescence intensity of the PtOEP−PdTPBP system rivaled that of its inverse dual-layer counterpart, this film system required all fluorescence to pass through the PdTPBP film to reach the photocatalyst, and the resulting reabsorption caused it to emit significantly weaker fluorescence at the wavelengths most strongly absorbed by ND/K-WO3 (Figures 3 and 4b). As a result, the device containing the PtOEP−PdTPBP dual-layer film system degraded only 4 ppmv of acetaldehyde during 90 min of irradiation, exhibiting a degradation efficiency 33%

Figure 8. Photocatalytic hydroxyl radical (•OH) production of WO3 and ND/K-WO3 films and their corresponding sub-band gap photocatalyst devices, rearranged as shown in Figure S8, during 30 min of white LED excitation (525 nm LP, 130.1 mW cm−2). (a) EPR spectra probing the generation of •OH adduct OH-DMPO in an aqueous solution of DMPO (C0 = 10 mM). (b) Fluorescence emission spectra (λEx = 332 nm, λEm ≈ 450 nm) probing the generation of •OH adduct 7-hydroxycoumarin in an aqueous solution of coumarin (C0 = 1 mM). H

DOI: 10.1021/acsami.9b01945 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

Figure 9. (a) Schematic illustrating the two dual-layer UC film systems relative to the directions of LED excitation and UC fluorescence collection. In each system, the top film is listed first. (b) UC fluorescence of individual and layered UC films under front-face white LED excitation (525 nm LP, 97.1 mW cm−2) at an angle of ∼45°. Each emission spectrum is the average obtained from four PtOEP films and/or four PdTPBP films, and the dashed trace marks the summed average emission of the PtOEP and PdTPBP films. (c) UC fluorescence of individual and layered UC films under front-face red (30.1 mW cm−2), green (28.2 mW cm−2), and simultaneous red and green LED excitation (525 nm LP). Each emission spectrum is the average obtained from four PtOEP films and/or four PdTPBP films. The dashed traces mark the summed average emission of each dual-layer film system under red and green excitation. All films contain 18 mM perylene with either 1.0 mM PtOEP or 0.50 mM PdTPBP, and all shaded regions denote one standard deviation.

relevant to photovoltaic applications, we characterized their UC performance under front-face white LED excitation (525 nm LP, 97.1 mW cm−2), exciting them and collecting their emission from the same direction (Figure 9a). This delivered the strongest excitation to the chromophores subject to the least fluorescence reabsorption, thereby minimizing detrimental inner filter effects and causing the UC fluorescence of the dual-layer systems to more closely approach the summed fluorescence of their individual films (Figures 9b and 3b). Since the film closer to the LED reabsorbed the UC fluorescence of the underlying film in this scenario, the PtOEP−PdTPBP dual-layer system emitted the strongest short-wavelength UC fluorescence (Figure 9b), reversing the trend previously shown under back-face excitation (Figure 3b). In further departures from previous trends, this film system significantly surpassed the peak UC fluorescence of its inverse (i.e., PdTPBP−PtOEP) counterpart and emitted stronger fluorescence at λ ≤ ∼490 nm (Figure 9b), confirming that in a device operating under front-face excitation, the PtOEP−PdTPBP film system would provide the greatest performance enhancement. A detailed analysis incorporating single-chromophore and chromophore-free control films, shown in Figure S9, traced the source of these departures to reflections off of the films’ glass faces, which favored the top film in each dual-layer system (see Figure 9a). The top film reflected a small amount of incident excitation, slightly reducing the light absorption of the underlying film and thereby reducing its UC fluorescence across all wavelengths (Figures 9c and S9b,d), analogous to

device containing WO3 in place of ND/K-WO3 exhibited minimal •OH production (Figure 8), consistent with the inability of its counterpart in Figure 6 to degrade acetaldehyde. Under identical excitation, isolated WO3 and ND/K-WO3 films exhibited no •OH production (Figure 8), confirming that UC fluorescence was essential for the operation of the sub-bandgap photocatalyst devices. While these tests were performed in aqueous solutions, we would expect the photocatalysts to exhibit analogous •OH production behavior at their interfaces with air.60 Broader Device Integration Insights. Whereas this study focuses on using layered TTA-UC film systems to improve the light harvesting of photocatalytic devices for air purification analogous film systems could just as easily be used to extend the light absorption range of devices with a variety of different applications. Our sub-bandgap photocatalyst devices positioned the UC film systems above the photocatalyst films they were intended to sensitize, thereby preventing scattering by the photocatalyst films from reducing the intensity of the UC films’ excitation. However, not all devices employ the same excitation geometry. Researchers employing TTA-UC for photovoltaic applications, for instance, typically position UC materials beneath a solar cell, where they harvest transmitted sub-bandgap light and recycle its energy back toward the device in the form of UC fluorescence.61 Since inner filter effects strongly influence the UC performance of layered film systems, as discussed above, it is crucial to factor excitation geometry into the development of UC film systems. To study our film systems’ performance in the excitation scenario I

DOI: 10.1021/acsami.9b01945 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

Figure 10. (a) Absorption spectra of UC films containing 1.0 mM PtOEP or PdOEP with 18 mM perylene. (b) Schematic illustrating the back-face and front-face excitation of dual-layer film systems containing these UC films, marking the directions of LED excitation and UC fluorescence collection. (c) Normalized UC fluorescence of individual and layered UC films under back-face and front-face green LED excitation (525 nm LP, 31.5 mW cm−2) at an angle of ∼45°. In layered film systems, the top film is listed first. Each emission spectrum is the average obtained from three PtOEP films and/or three PdOEP films. The dashed traces mark the summed average emission of the PtOEP and PdOEP films, whose peak is used to normalize the emission spectra, and the shaded regions mark one standard deviation.

at λ < 480 nm (Figure S10b−c). As a result, our sub-bandgap photocatalyst devices would undoubtedly have shown different performance trends under warm white LED excitation. This underscores the importance of tailoring layered UC film systems for device integration to both their devices’ absorption ranges and the emission spectra of the light sources that they are intended to harvest, necessitating many different combinations of sensitizer and acceptor chromophores. Film systems intended to harvest sunlight, for instance, will benefit from incorporating emergent TTA-UC sensitizers capable of harvesting near-infrared light.63,64 In any case, film systems with strong light absorption across broad wavelength ranges will ultimately require sensitizer assemblies with far more spectral overlap than PtOEP and PdTPBP. While the distinct absorption ranges of these two sensitizers were convenient for the purposes of this study because they allowed for excitation of each sensitizer individually, they do not allow us to draw conclusions about the potential implications of sensitizers’ absorption overlap. As a preliminary exploration of these implications, we also characterized the performance of dual-layer UC film systems pairing PtOEP with its palladium analogue, palladium(II) octaethylporphyrin (PdOEP), which has an absorption profile mirroring that of PtOEP but red-shifted by ∼20 nm (Figure 10a). Because neither sensitizer absorbs perylene fluorescence, the order of the PtOEP and PdOEP films had no impact on the secondary inner filter effects within these new dual-layer systems, and the sensitizers’ competition for green excitation became the primary factor determining the optimal film order. Under both back-face and front-face green LED excitation (525 nm LP, 31.5 mW cm−2, Figure 10b), the PdOEP films

trends observed under back-face excitation (Figures 3c and S3b,d). At the same time, however, the underlying film slightly enhanced the UC fluorescence of the top film, presumably improving its light absorption by acting as a beneficial reflector (Figures 9c and S9a,c). The combined effects of these two processes elevated the UC fluorescence of the PtOEP− PdTPBP dual-layer system, in which the film with stronger emission (i.e., PtOEP) held the advantageous top position, above that of its inverse counterpart. Notably, studies that incorporate UC systems into solar cells often employ underlying reflector layers,62 so the measurable effect of the minor reflections within our dual-layer film systems suggests that it will be imperative to factor these reflector layers into future efforts to develop layered UC film systems for photovoltaic applications. To approximate the indoor light that photocatalytic air purification devices would seek to harvest, we have thus far characterized our films’ UC performance under excitation with a single white LED. In reality, of course, photocatalytic and photovoltaic devices harvest light from artificial and natural light sources (i.e., sunlight) with a wide range of different emission spectra, which would affect the relative light absorption of UC films within a layered system. As an example, changing our light source to a warm white LED with weaker green emission and stronger red emission (525 nm LP, Figure S10a) resulted in UC performance trends markedly different from those shown in Figure 3. Under warm white excitation, the single-sensitizer PdTPBP films outperformed their PtOEP counterparts, which offset the effects of their elevated UC fluorescence reabsorption and thereby brought the two dual-layer film systems to equivalent UC fluorescence J

DOI: 10.1021/acsami.9b01945 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

performance of dual-layer TTA-UC film systems and identify factors that will be important to consider during future efforts to develop more complex multilayer film systems for device integration. We superimpose optimized PtOEP and PdTPBP films to produce dual-layer UC film systems with broad light absorption that gives rise to superior performance under white LED excitation. By pairing these film systems with a new composite photocatalyst, we produce the first UC-enhanced devices able to harvest both red and green sub-bandgap photons through TTA-UC. Their enhanced degradation of gaseous acetaldehyde provides the first demonstration of layered UC film systems’ promise in real-world applications. As a whole, this study underscores the benefits of layered UC film systems’ remarkable adaptability, establishing a framework for the development of more broadly absorbing solid-state TTA-UC systems with potential applications in a wide range of photocatalytic and photovoltaic devices. First and foremost, these film systems’ sensitizers should be chosen to best harvest the otherwise-wasted low-energy emission of their intended light source, and their acceptors should be chosen to best activate their intended devices. Our collective findings demonstrate that the spectral overlap between the selected chromophores, in conjunction with the device architecture, the corresponding excitation geometry, and the light source emission, dictates the optimal ordering of the individual layers within these film systems. For instance, we show (1) that in film systems pairing green-to-blue and red-toblue UC sensitizers with distinct absorption profiles, the layers should be ordered to minimize the detrimental secondary inner filter effects that result from the overlap between the absorption of PdTPBP and the emission of its paired acceptor and (2) that the implications of these effects depend on the light source employed for excitation. In film systems pairing green-to-blue UC sensitizers with strongly overlapping absorption, on the other hand, the layers should be ordered to position the film with stronger UC fluorescence closer to the excitation source, thereby minimizing detrimental primary inner filter effects. We are confident that by following and building on this initial framework, future researchers will be able to produce film systems capable of harvesting broad wavelength ranges of otherwise-wasted light for improved device performance.

outperformed their PtOEP analogues, and the two dual-layer film systems outperformed both films. The competing absorption of their sensitizers, however, kept the layered film systems’ UC fluorescence well below the combined fluorescence of their individual PtOEP and PdOEP films (Figure 10c). As previously observed in the PtOEP−PdTPBP systems (Figures 3b and 9b), inner filter effects made this fluorescence disparity greater under back-face excitation (Figure 10c). In a significant departure from the trends shown by the PtOEP−PdTPBP film system (Figures 3b and 9b), the PdOEP−PtOEP film system outperformed its inverse (i.e., PtOEP−PdOEP) counterpart under both back-face and frontface excitation (Figure 10c). The sensitizers’ overlapping absorption introduced strong primary inner filter effects, causing light absorption by the top film in each dual-layer system to drastically reduce the amount of excitation that reached the underlying film. Since the order of its films favored the light absorption of the film with stronger UC fluorescence, the PdOEP−PtOEP system exhibited better UC performance regardless of the direction in which its fluorescence was collected (Figure 10). This suggests that in the absence of sensitizer-induced reabsorption, the relative fluorescence emission of UC films should dictate their order within layered systems. As discussed in detail above, the reflections that occur during front-face excitation simultaneously enhance the light absorption of the top film and handicap the light absorption of the underlying film, so this excitation scenario gave rise to a greater disparity between the UC performances of the two dual-layer systems (Figure 10c). Finally, in light of the paramount importance of inner filter effects within layered UC film systems, we explored the influence of film thickness by layering multiple PtOEP or PdTPBP films. Under back-face white LED excitation (525 nm LP, 95.2 mW cm−2), layering an increasing number of films yielded rapidly diminishing returns in UC fluorescence, and increasing the number of film layers beyond two offered no additional benefit (Figure S11a). This reflects the strong primary inner filter effect of the films’ high sensitizer concentrations, which caused absorption by the sensitizers to rapidly attenuate the intensity of LED excitation as it passed through each film (Figure S1). In addition, each film layer increased the distance that UC fluorescence had to travel within the films, thereby increasing the amount of UC fluorescence reabsorbed by the films’ PdTPBP and/or perylene. This secondary inner filter effect was evident in the decreasing intensity of the films’ short-wavelength fluorescence as well as the red shift in their peak fluorescence wavelength (Figure S11a). Comparing the emission spectra of individual PtOEP and PdTPBP films under back-face and front-face excitation allowed us to also observe these inner filter effects within a single film layer. Front-face excitation exposed the film face closest to the point of fluorescence collection to the strongest LED excitation, which minimized fluorescence reabsorption within the films and thus gave rise to significantly stronger fluorescence below 460 nm (Figure S11b). Collectively, these results suggest that during future development of layered systems for device integration, optimizing the thickness of each individual layer could yield considerably improved performance.

■

EXPERIMENTAL METHODS

Preparation of UC Films. All chromophores were used as received without additional purification. Stock solutions of PtOEP (Frontier Scientific), PdOEP (Frontier Scientific), PdTPBP (Lumtec), and perylene (Aldrich) were prepared in tetrahydrofuran (THF) and stored in the dark. PU precursor Clearflex 50 (CLRFLX), purchased from Smooth-on, Inc., comprised two separately stored components: component A, 4,4′-methylenebis(cyclohexyl isocyanate), and component B, a polyester polyol. Precursor mixtures for chromophore-doped PU films were prepared by adding aliquots of the appropriate stock solutions to a mixture of CLRFLX components A and B (1:2 by volume). Disparities between the volumes of stock solution in each precursor mixture were corrected using THF, thereby making the PU precursor concentration consistent across all mixtures and enabling comparison between the performances of all films within a given set of experiments. The resulting mixtures were vortex mixed and then sonicated for 5 min to ensure thorough mixing. Individual films were fabricated by casting 120 μL of the precursor mixture onto a square microscopy cover glass (22 mm × 22 mm) and then carefully placing a second cover glass on top to yield a flat, level surface. Films were cured overnight at room temperature in the dark and then heated at 60 °C for 2 d to remove residual THF. Films used in VOC

■

CONCLUSIONS In this study, we utilize the intrinsic adaptability of layered film assemblies to investigate the properties that influence the K

DOI: 10.1021/acsami.9b01945 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces degradation experiments were subsequently heated at 70 °C for an additional week to ensure thorough THF removal. Preparation of Photocatalyst Films. CdS was used as purchased (Aldrich), and C3N4 was synthesized in an alumina crucible with a cover by calcining urea (99%, Aldrich) at 550 °C for 2 h with a ramp rate of 5 °C min−1. N-TiO2 was synthesized by annealing commercial P25 TiO2 (Degussa) at 550 °C for 3 h in a tube furnace under ammonia flow (NH3/Ar = 1:4) with a flow rate of 150 mL min−1.65 WO3 was either used as purchased (Aldrich, diameter