Recent advances in the application triplet – triplet annihilation-based photon upconversion systems to solar technologies

. Upconversion (UC) of low-energy photons into high-energy light can in principle increase the efficiency of solar devices by converting photons with energies below the energy absorption threshold into radiation that can be utilized. Among UC mechanisms, the sensitized triplet – triplet annihilation-based upconversion (sTTA-UC) is the most recent that would be applied in solar technologies. sTTA-UC was demonstrated using sunlight in 2006, and due to its high efficiency at low excitation intensities with noncoherent light, it is considered a promising strategy for the recovering of subbandgap photons. Here, we describe briefly the working principle of the sTTA-UC, and we review the most recent advances of its use in solar applications. © The Authors. Published by SPIE under a Creative Commons Attribution 3.0 Unported License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI. [DOI: 10.1117/1.JPE.8.022005]


Introduction
The Shockley-Queisser model for single-junction solar cells (SCs) states that the efficiency of a single-junction device depends on the bandgap of the reactive material. This is a direct consequence of the fact that photons of lower energy than the semiconductor bandgap cannot be absorbed, while charges created after absorption of photons of higher energy than the bandgap undergo thermalization. 1 Since all solar devices feature this intrinsic absorption energy threshold, incident photons with energies below this threshold do not contribute significantly to current generation in the SC. These losses are called transmission or subbandgap losses. 2 The extent of the losses is shown in Fig. 1(a) for different single-junction SC technologies.
Upconversion (UC) of low-energy photons into high-energy light can in principle increase the efficiency of solar devices by converting photons with energies below the energy absorption threshold into radiation that can be utilized. In their theoretical study, more recently confirmed and developed by Atre and Dionne, 3 Trupke et al. 4 showed that a system consisting of a SC with a single bandgap of 2 eV and an ideal upconverter coupled to its rear can achieve a maximum power conversion efficiency of 47.6% under nonconcentrated sunlight. Since (i) this efficiency value is remarkably higher than the Shockley-Queisser limit of 34% for devices with a bandgap of 1.34 eV and (ii) upconverters can be in principle easily implemented in existing technologies, in the last decade many efforts have been made by the scientific community towards the development of efficient upconverters for real-world solar technologies. 5 Among UC mechanisms, the sensitized triplet-triplet annihilation-based upconversion (sTTA-UC) is the most recent photon UC technique to be potentially applied to solar technologies. 6 sTTA-UC has been demonstrated with sunlight by Baluschev et al., 7 and due to its high efficiency at low excitation intensities with noncoherent light, it is indeed considered a very promising strategy for the recovery of subbandgap photons in photovoltaic and *Address all correspondence to: Angelo Monguzzi, E-mail: angelo.monguzzi@unimib.it photocatalytic devices working under the solar irradiance. 2,8 In the following, we describe the working principle of the sTTA-UC and review the most recent advances of its application to solar technologies.

Principles of Sensitized Triplet-Triplet Annihilation-Based Upconversion
UC of low-energy photons from a noncoherent radiation source, such as the sun, is most frequently a multistep process that involves the sensitized population of metastable optically dark electronic states of a given system. 5 These can transit to higher emissive states by receiving additional energy through energy transfer (ET) from other sensitizers (ions or molecules) or by mutual annihilation upon biexcitonic collisions. The radiative transition from the reached higher energy level to the ground state leads to the emission of the desired higher energy photon. There are two main reasons why most frequently an ET process is involved: first, it is not necessary that one sort of ion or molecule provides all needed energy levels; second, an excited ion or molecule does not have to absorb more than one photon, which would be an extremely unlikely event requiring high excitation intensity. Indeed, to realize devices working under standard solar irradiance, traditional UC mechanisms, such as second-harmonic generation or sequential multiphoton absorption by lanthanide ions, are still not suitable because they require coherent and/or high-intensity irradiation to reach high efficiencies. 5,9 Conversely, the sTTA-UC also occurs at excitation intensities as low as a few mW cm −2 and exhibits quantum efficiencies (QY uc ) at low power that are orders of magnitude higher than those of the aforementioned processes. [10][11][12] In this multistep UC route, energy absorbed by a sensitizer is transferred to a second moiety, the emitter, by means of a triplet-triplet ET process. Subsequently, two optically dark emitter triplets (E T ) undergo triplet-triplet annihilation (TTA) to produce one excited singlet state, from which a photon with higher energy than those absorbed is emitted [ Fig. 2(a)]. Since its demonstration in solution with concentrated sunlight and thanks to its surprisingly high efficiency at low powers, materials for sTTA-UC have been intensively studied and developed to achieve higher performances and to cover a larger fraction of the visible (Vis)-near-infrared (NIR) spectrum. 7,12,[30][31][32] By choosing the proper emitter/sensitizer pair, UC quantum efficiencies (QY uc ) above 30% have been obtained at excitation power densities of 10s of mW cm −2 . This remarkable value is not far from the maximum theoretical yield of 50% arising from the energy conservation law. 13,33 Figure 2(b) shows the gradual increment of QY uc achieved in the last 10 years. Notably, despite a slower initial development, a good performance is now assessed even for solid-state materials. Nowadays, the research on sTTA-UC for solar applications is indeed focused on the design and development of solid systems that match the strict requirements for technological applications. The crucial issue for solid sTTA-UC is to keep the proper mixing of sensitizer and emitter moieties while maintaining the exciton diffusivity that maximizes the ET and the TTA steps as in low-viscosity solutions. If these requirements are met, the excitation intensity threshold I th required to obtain the maximum QY uc can be achieved below the solar irradiance, thus allowing for the design and development of UC-assisted devices. 34   several types of SCs, as shown in Fig. 3, which summarizes the most important milestones and advances in the development of sTTA-UC materials and applications in solar technologies.
In the following, we describe the state-of-the-art of UC-coupled SCs that exploit the sTTA-UC mechanism to improve their photon-harvesting ability. The text is organized in several sections, according to the type of semiconductor material employed to harvest the incoming radiation.

Silicon-Based Devices
The Australian research group lead by Schmidt 2,45,46 is a pioneer in the sTTA-UC field not only for the deep study of the fundamental photophysics occurring in the process but also for the development of numerical models and prototype devices that describe and exploit the sTTA-UC. In particular, they focused on the efficiency enhancing of silicon-based SC cells, the most diffuse solar devices commercially available. In 2012, they reported the first integrated a-Si:H/TTA-UC photovoltaic device as a proof-of-principle of a third-generation device, and they demonstrated how to characterize such an upconversion-enhanced solar cell (UC-SC) so that its performance under solar conditions may be evaluated. 47 As shown in Fig. 4, they realized a multilayer device based on the use of a thin bifacial a-Si:H p-i-n cell coupled to a sTTA-UC solution where the tetrakisquinoxalinoporphyrin (PQ 4 Pd) and its derivative PQ 4 PdNA were used as sensitizers and rubrene as an annihilator/emitter moiety for NIR-to-VIS UC. With a 50-nm-thick UC-SC, they observed an overall relative energy conversion efficiency increase of about 0.1% under 48 suns of noncoherent white excitation. A following improvement of the device architecture lead to an enhancement of ∼0.2% under 19 suns, 48 a value further enlarged with the inclusion of a back reflector to improve the harvesting of the upconverted light by the silicon layer. 49

Organic and Hybrid Devices
The electronic properties of the most efficient sTTA-UC pairs allow for the exploitation of light in the NIR and Vis spectral ranges. Therefore, it is not difficult to imagine that these systems can be coupled with a SC with an energy bandgap wider than silicon as dye-sensitized solar cells (DSSC). DSSCs are a photovoltaic technology that has generated significant interest across a broad range of research fields since their first report. 50 Unfortunately, despite the large effort made, the highest reported device efficiencies remain relatively low. However, this is partially compensated for by low projected manufacturing costs and their strong performance in ambient and diffuse lighting conditions. The large driving forces required for effective charge transfer in DSSCs necessitate the use of absorbers with large absorption thresholds. 51,52 This large energy gap, therefore, results in a significant section of the solar spectrum remaining unused by the cell, which can be recovered using a suitable UC system. 53 In 2013, Nattestad et al. 54 demonstrated that a DSSC cell displays Fig. 3 Timeline of the technological advances for the application of sTTA-UC to solar energy devices.
enhanced current under subbandgap illumination at a low light concentration (3 suns) when an SC has been coupled to the a PQ 4 PdNA∕rubrene solution, whose upconverted emitted photons are recaptured by the D149 dye. In 2016, a significant step forward in this research was made by Akindoyo et al., 55 who presented a study where, for the first time, the sTTA-UC enhancement of an SC performance is combined with the ability to embed the chromophores in polymer matrices that can be tailored to films and many different shapes and sizes. The synthetic flexibility and longevity of these sTTA-UC films are of the utmost importance for economically practical applications. The films are indeed robust against oxygen quenching and can be readily tailored to existing SC architectures [ Fig. 5(a)]. The photovoltaic performance of the UC-assisted DSSC device [Figs. 5(b) and 5(c)] was measured under both high-power coherent laser and low-power incoherent light irradiation. 56 The resulting photovoltaic performance is summarized in Fig. 5(c). In a DSSC based on a 5-μm-thick mesoporous TiO 2 layer and a single sTTA-UC film, the short-circuit photocurrent J sc was enhanced by 13.6% and 20.3% upon illumination with the equivalent of 1 sun and 2 suns, respectively. The addition of a rear reflector to utilize more effectively the upconverted photons almost doubles the J sc enhancement observed.
An alternative device architecture for UC-enhanced DSSC was proposed by the group lead by Hanson in 2015, inspired by the concept of intermediate band cells (vide infra, Sec. 2.3). 57 To facilitate the sTTA-UC and the charge separation of the upconverted excitons, they used an inorganic−organic interface based on self-assembled bilayers of a TTA molecular pair on metal oxide surfaces. This self-assembled motif via metal ion linkages is shown in Fig. 6(a) and provides a simple and modular method for multimolecular assembly on a metal oxide surface. 58,59 This strategy is not diffusion limited and offers unprecedented geometric and spatial control of the donor−acceptor interactions at an interface. The bilayer film depicted is prepared by stepwise soaking of nanocrystalline TiO 2 or ZrO 2 films in three separate solutions of 4,4′-(anthracene-9,10-diyl)bis(4,1-phenylene)diphosphonic acid (DPPA), ZnðCH 3 COOÞ 2 , and Pt(II)tetrakis(4carboxyphenyl)porphyrin (PtTCPP). Anthracene and platinum porphyrin derivatives were selected as the TTA-UC pair because they are known to exhibit efficient UC-emission. Figure 6(b) shows the electronic transitions and relative energetics for DPPA and PtTCPP and the conduction band (CB) energies of TiO 2 and ZrO 2 . The authors demonstrated the generation of photocurrent with subbandgap photons by monitoring the peak photocurrent for TiO 2 -DPPA-Zn-PtTCPP with respect to 532-nm excitation. The photocurrent density exhibits a change in slope from 1.2 to 0.61 that is reminiscent of the quadratic to linear behavior typical for sTTA-UC [ Fig. 6(c)]. This change in slope is in contrast to that for TiO 2 -PtTCPP, which exhibits a slope of 0.53 throughout the entire excitation intensity range from ∼300 to 7000 mW cm −2 . Under the same conditions, a similar linear dependence (slope ¼ 0.5) is observed for the common DSSC dye employed N3, which is known to undergo a one photon-to-one electron photocurrent generation mechanism. 60 By combining intensity dependence and incident photon-to-electron conversion efficiency measurements, they further demonstrated that these bilayer films effectively generate a photocurrent by two different mechanisms: (1) direct excitation and electron injection from the acceptor molecule and (2) low-energy light  absorption by the sensitizer molecule followed by sTTA-UC and electron injection from the upconverted state. Notably, the power conversion efficiency from the upconverted photons is the highest yet reported for an SC integrated with sTTA-UC. 61 Further developing this strategy, to overcome the intrinsic limitation of porphyrin-like sensitizers, i.e., their narrow absorption bandwidth that limits the system absorptance, Dilbeck et al. 62 recently introduced a trilayer architecture as a new system for incorporating multiple sensitizers into sTTA-UC film-based DSSC (Fig. 7). By means of interlayer ET from both sensitizers in the bilayer, and in energy cascade for the trilayer, it is possible to efficiently absorb low-energy light and generate acceptor triplet-excited states. A photocurrent enhancement is observed in the multisensitizer films due to cooperative broadband absorption of employed light harvesters. The efficiency of the trilayer device (1.2 × 10 −3 %), with high sensitizer density and directional ET that cascades toward the charge separation interface, marks another record among the performance of integrated sTTA-UC-SCs currently reported. Importantly, intensity dependence measurements demonstrate that high triplet densities and peak UC efficiency can be achieved at subsolar irradiance using dual sensitized self-assembled multilayer films. Even though the device efficiency is still low, these results suggest that an integrated sTTA-UC DSSC offers a viable means of enhancing solar energy harvesting efficiencies upon several improvements, including optimizing overlap between the sensitizer(s) and the solar spectrum, increasing injection yield from the singlet-excited state of the acceptor molecule, maximizing the regeneration rate, and reducing the triplet quenching losses.

Intermediate Band Solar Cells
Intermediate band solar cells (IBSCs) are a third-generation SC technology where, as the name suggests, the inclusion of an absorbing band in the energy gap of the "main" semiconducting material allows for the increase in the efficiency, potentially exceeding the Shockley-Queisser limit. The new band opens a second absorption channel; therefore, the photoexcitation of carriers into the CB can occur from direct absorption from the valence band (VB) or after consequent absorption of lower energy photons from the VB to the intermediate band (IB) and then from the IB to the CB.
The fabrication of IBSCs has always been limited by strict constraints of the intermediate absorption band, namely (i) a generally low-absorption cross section, (ii) the need to use spectrally selective absorbers, i.e., to avoid the excitation of carriers to the IB via a higher energy photon, and (iii) the rapid nonradiative relaxation from the IB to the VB. Organic IBSCs based on sTTA-UC allow overcoming all the above constraints by exploiting the unique electronic structures of the employed components. In detail, the extremely high molar extinction coefficient of metalated porphyrins allows for high absorption of light, while the narrow absorption peak selects only the wavelengths that will be used to activate the TTA process. Simultaneously, the E T that are populated via TTA are metastable dark states with an extremely low oscillator They demonstrated that the most efficient electronic structure for an sTTA-UC IBSC features a symmetrical band structure arrangement, with the IB lying at around midgap. The power conversion efficiency has been calculated to be up to 40.6% for 1.9-eV excitation light under 1-sun illumination, exceeding the Shockley-Queisser limit. However, no significant IBSC has been fabricated until very recently. 2 The first TTA-UC-based IBSC was demonstrated by Simpson et al. 59 in 2015 by staining a screen-printed commercial TiO 2 paste with an upconverting solution composed of Pt(II)-tetrabenzotetraphenylporphyrin (PtTBTPP, the sensitizer) and 4,4′-(anthracene-9,10-diylbis(ethyne-2,1-diyl))dibenzoic acid (BDCA, the emitter). The energy-level schematic of the prototype devices is shown in Fig. 8. This IBSC is shown to generate photocurrent when illuminated with incoherent light whose energy is well below the bandgap of TiO 2 (560 nm < λ < 640 nm), while a control device with no BDCA showed no photocurrent under the same conditions.
A different approach to sTTA-UC-based IBSCs has been recently given by Lin et al. 65 They fabricated an all-solid-state organic cell with the active layer composed of a blend of PtTPTBP and α-sexithiophene (α-6T) as the sensitizer and emitter, respectively. The acceptor layer for the IBSC was diindenoperylene and was compared with C60, which has a charge transfer state with a much lower energy than the triplet states of α-6T. Therefore, the cell with C60 has a direct funneling of charges from α-6T triplets to C60, bypassing TTA and serving as a model, control system. With their IBSC, the authors were able to reach a 12% increase in photocurrent with respect to the control system under subbandgap illumination intensity lower than 0.1 mW cm −2 . The examples reported above are interesting and promising steps forward to the realization of efficient IBSC based on sTTA-UC due to its photophysical properties that overcome the major limitations of this SC technology.

Sensitized Triplet-Triplet Annihilation-Based Upconversion in Photochemical Systems
Beyond improving SC efficiencies, the sTTA-UC can also be used to enable photochemical reactions, many of whose require high-energy photons, generally blue or UV, which are scarce in the solar spectrum. Taking into account that the spectral range where sTTA-UC is more efficient is from the Vis to the near-UV, the expected response to the coupling with an upconverter is expected to be greater for these systems than that of SCs. In 2006, Islangulov and Castellano 66 demonstrated that bimolecular cycloaddition between two anthracene molecules was possible even when excited with light that anthracene itself does not absorb due to the exploitation of a proper sensitizer. In this study, there was no need for the emitter to radiate a photon, as it is the singlet-excited acceptor produced in the TTA step that reacts further. This was the first example representing a larger group of synthetic applications where reactions requiring high-energy UV-blue photons can be promoted by UC mechanisms, either where the produced excited singlet acceptor is the reacting species, as above, or where the emitted photon from the excited singlet acceptor in turn excites the reacting specimen.
In addition to the traditional photochemistry, the sTTA-UC is especially suited to be coupled with water-splitting cells that exploit wide-bandgap photocatalysts, since currently such catalysts cannot utilize about 95% of the available solar photons. 67 sTTA-UC-enhanced photo-catalytic water splitting (PCWS) was demonstrated in 2012 under noncoherent irradiation in a proof-ofprinciple photoelectrochemical device (Fig. 9). 68 The blue photons emitted from a quartz cuvette containing 9,10-diphenylanthracene (DPA) and palladium(II)-octaethylporphyrin (PdOEP) upconverting solution were used to activate the water-splitting reaction driven by a tungsten trioxide (WO 3 ) photoanode under low-energy excitation (λ > 500 nm), well below the lowenergy absorption tail of WO 3 (450 nm). Kim et al. developed this approach successively with a Pt-loaded WO 3 (Pt∕WO 3 ) photocatalyst. Once irradiated with photons of energy above its bandgap (λ < 440 nm), it produces hydroxyl radicals (· OH). 27 The sTTA-UC enabled the photoassisted production using subbandgap green photons (532 nm). In this case, the photon UC was achieved directly in the aqueous phase by encapsulating the solvent phase containing the DPA/platinum(II)-octaethyl-porphyrin (PtOEP) benchmark sensitizer/emitter pair in a rigid polymer shell in the form of aqueous dispersible microcapsules. A mixture of hexadecane and polyisobutylene was used as the inner solvent phase. Importantly, this eliminated the need for the deoxygenation step that is essential for preserving the triplets involved in the sTTA-UC process from the quenching by molecular oxygen, enabling a stable UC signal to be detected after a month of exposure to the ambient environment. 69 From this proof-of-concept work, the "encapsulated" sTTA-UC system has been developed significantly, resulting in the fabrication of complex and robust silica-coated capsules. The silica shell not only protects the included liquid upconverter from oxygen but also provides sites for further surface modification, which allows, when combined with its nanoscale dimension and structural rigidity, this sTTA-UC system to acquire various useful functionalities (Fig. 10). By decorating the upconverters silica shell with cadmium sulphide nanocrystals, as photocatalyst, and graphene oxide nanodisks, as a cocatalyst, Kim et al. demonstrated the photoassisted production of · OH and H 2 O 2 using subbandgap photons. Considering that solar-powered H 2 O 2 is not only a useful solar fuel but also a green oxidant and disinfectant for water purification, they suddenly spread the ensemble field of sTTA-UC applications, introducing new environment-related problems. 70,71 In parallel, the development of UC-assisted PCWS devices did not stop. In particular, our group focused on the realization of multilayer devices with improved optical coupling between the PCWS cell and the UC system, to maximize the enhancement in the photocurrent and hydrogen production upon absorption of Vis photons. To realize a solid upconverter that can be easily implemented in the PCWS cell architecture, we propose the use of single-phase elastomers with low glass transitions temperature (<0°C) doped with proper dye pairs (Fig. 11). 21 Due to the residual molecular diffusion provided by the soft host and to the reduced quenching of the involved metastable electronic excited states in a solid environment compared with a liquid one, we obtained efficient sTTA-UC at the solid state. Notably, due to the adhesive properties of the material, we were able to position the upconverter on the photoanode backside, where the subbandgap photons can be recovered properly and the upconverted photons  harvested. By further developing this approach, we recently obtained the record enhancement of photocurrent generated in a standard PCWS cell under solar irradiance exploiting a sTTA-UC polymer boosted with highly fluorescent semiconductor nanocrystals, thus showing how the sTTA-UC materials can be successfully implemented in technologically relevant devices while matching the strict requirements of clean-energy production. 72 Börjesson et al. 73 reported another example of a photochemical reaction enhanced by sTTA-UC, where the photoisomerization of a fulvalene diruthenium derivative (FvRu 2 ) was driven by blue photons upconverted from truncated white light. In the study, they set up a microfluidic system (Fig. 12) to enable the flowing of both the FvRu 2 containing liquid and the sTTA-UC fluid, consisting of DPA and PdOEP in toluene. They achieved a 130% increase in conversion of the photoisomer when coupled to the upconverter.

Conclusions
The aim of this review is to point out how the sTTA-UC began to be implemented in technologically appealing material and devices after its demonstration under solar light about a decade ago. Several strategies have been proposed to exploit the upconverted radiation. Therefore, several device architectures have been designed to implement the sTTA-UC mechanism in photovoltaic, photocatalytic, and photochemical systems, resulting in the realization of prototype devices that demonstrate the applicability of sTTA-UC. The results obtained indicate that the application of this UC mechanism to real-world devices demands the realization of more stable materials that are easy to integrate, for example, by embedding the upconverters in transparent matrices. It is important to point out that, considering the photophysics of the biexcitonic processes involved, the sTTA-UC should profit from the realization of structured and preorganized systems to enhance excitonic diffusivity and, therefore, promote TTA efficiency at ultralow powers. This will enable reaching threshold intensities well below the solar irradiance, making sTTA-UC particularly suitable for application in real-life technologies.
However, an extra effort is still required to design and synthesize sTTA-UC materials. Table 1 summarizes some of the dye pairs currently known that are suitable for sTTA-UC and their spectral working range. It is easy to see that, while there is plenty of alternative pairs for Vis-to-Vis photon UC, the family of available materials for Vis-to-UV and NIR-to-Vis working ranges is significantly limited. Despite enormous progresses having been made on the design and architecture of sTTA-UC devices, this fundamental lack of materials is particularly detrimental for the evaluation of a reliable road-map evaluation for their commercial production. The use of semiconductor nanocrystals as sensitizers in substitution for the commonly employed metallated porphyrins was recently proposed, 23,78 opening the possibility of shifting the sTTA-UC working range toward the IR. However, the application of nanocrystals to sTTA-UC for solar technologies still has to be evaluated. Since nanocrystals lack transparency windows in the UC-emission spectral range, their reabsorption limits in principle their applicability to solar devices, where high optical densities are required. While the use of semiconductor nanocrystals may open the possibility to upconvert IR light, a striking strategy to produce UV upconverted photons still has to be found. The realization of efficient Vis-to-UV or, ideally, NIR-to-UV upconverters will indeed have a huge potential for photocatalytic and photochemical applications, but it would also greatly benefit other fields, such as photolithography and water sterilization. 83,84 Disclosures The authors declare no competing financial interests.