An integrated device, incorporating a dye-sensitized solar cell and triplet-triplet annihilation up-conversion unit was produced, affording enhanced light harvesting, from a wider section of the solar spectrum. Under modest irradiation levels a significantly enhanced response to low energy photons was demonstrated, yielding a record figure of merit for dye-sensitized solar cells.
The poor response of dye-sensitized solar cells (DSCs) to red and infrared light is a significant impediment to the realization of higher photocurrents and hence higher efficiencies. Photon up-conversion by way of triplet-triplet annihilation (TTA-UC) is an attractive technique for using these otherwise wasted low energy photons to produce photocurrent, while not interfering with the photoanodic performance in a deleterious manner. Further to this, TTA-UC has a number of features, distinct from other reported photon up-conversion technologies, which renders it particularly suitable for coupling with DSC technology. In this work, a proven high performance TTA-UC system, comprising a palladium porphyrin sensitizer and rubrene emitter, is combined with a high performance DSC (utilizing the organic dye D149) in an integrated device. The device shows an enhanced response to sub-bandgap light over the absorption range of the TTA-UC sub-unit resulting in the highest figure of merit for up-conversion assisted DSC performance to date.
Dye-sensitized solar cells (DSCs) have been proclaimed as a promising concept in affordable solar energy collection1-3. In spite of this enthusiasm, widespread commercialization has yet to occur. A number of reasons have been put forward for this, with one pressing issue being the relatively high energy of the absorption onset, limiting the achievable light harvesting efficiency of these devices4. Although this can be overcome, lowering the absorption onset is typically accompanied by a drop in open circuit voltage, which disproportionately erodes any gains in current density5, 6.
The general operation of DSCs involves electron transfer from a photoexcited dye to a semiconductor (typically TiO2), followed by the regeneration of the oxidized dye by a redox mediator. Both these processes appear to require substantial driving forces (potential) in order to proceed with high efficiency7. With such significant inherent losses, it becomes obvious that the optimal absorption onset for these devices is reasonably high in energy. Similar problems exist for organic photovoltaics (OPV), due once again to the large chemical driving forces required for effective charge separation. Accordingly, predictions of upper solar-to-electric conversion efficiency limits to single junction devices based on both of these technologies involve absorbers with wide (effective) band gaps4.
In order to overcome the light harvesting issue raised above, a number of approaches have been taken. This includes the ‘third generation’8 approaches of tandem structures9, 10 and photon upconversion11-14.
Recently11 we reported an integrated device composed of a DSC working and counter electrode, with a triplet-triplet annihilation based up-conversion (TTA-UC) system incorporated into the structure. This TTA-UC element was able to harvest red light transmitted through the active layer and chemically convert it (as described in detail below) to higher energy photons which could be absorbed by the active layer of the DSC and generate photocurrent. There are two important points to note about this system. Firstly, TTA-UC has many prospective advantages over other photon upconversion systems11; secondly it demonstrates a feasible architecture (proof-of-principle) for the incorporation of TTA-UC, which had been lacking from the TTA-UC literature up to that point.
The process of TTA-UC15-24 involves the excitation of ‘sensitizer’ molecules, in this case Pd porphyrins, by light with energy below the device onset energy. The singlet-excited sensitizers undergo rapid intersystem crossing to the lowest-energy triplet state. From there, they can transfer energy to a ground-state triplet-accepting ‘emitter’ species such as rubrene, as long as the transfer is allowed by free energy25. The first triplet state of rubrene (T1) is greater than half the energy of its first excited singlet state (S1) but less than half the energy of T2, meaning that an encounter complex of two triplet-excited rubrenes can annihilate to give one singlet excited emitter molecule (and the other in the ground state) with a fairly high probability. Other states, statistically predicted, are most likely energetically inaccessible for rubrene26. The singlet excited rubrene molecule can then emit a photon (as per fluorescence) with energy sufficient to excite the dye on the working electrode of the DSC. This process is shown in Animation 1.
TTA-UC offers a number of advantages compared to other UC systems, such as a broad absorption range and incoherent nature27, 28, making it an attractive option for coupling with DSC (as well as OPV). TTA-UC has been demonstrated operating at relatively low light intensities and in diffuse lighting conditions. Both DSC and OPV are most efficient in the low light intensity regime. Solar concentration is expensive and only justifiable for high efficiency, high cost devices. The relatively high performance of TTA-UC systems in low intensity lighting conditions is attributable to the process involving sensitizer chromophores with strong, broad absorption bands in concert with long-lived triplet states which are capable of diffusing in order to come into contact with interacting species. In addition, TTA-UC has been found to have high intrinsic efficiency from a kinetic study26.
Although TTA-UC operates at low light intensity, there is still a quadratic relationship between incident light intensity and emitted light (at least at low light intensities). This is due to the bimolecular nature of the process. To account for this and the varied experimental conditions (particularly light intensity) reported by different groups, a figure of merit (FoM) system should be employed to meter the performance enhancement offered by upconversion. This FoM has been defined as ΔJSC/ʘ, where ΔJSC is the increase in short circuit current (usually determined by integration of the Incident Photon to Charge Carrier Efficiency, IPCE, with and without the upconversion effect) and ʘ is the effective solar concentration (based on the photon flux in the relevant region, that is the Q-band absorption of the sensitizer)229.
Herein, a protocol for producing and correctly characterizing an integrated DSC-TTA-UC device is reported, paying special attention to potential pitfalls in device testing. It is hoped that this will serve as a basis for further work in this field.
1. DSC Fabrication
1.1. Working Electrode Preparation
1.2. Counter Electrode Preparation
1.3. Reflector
1.4. Device Assembly
1.5. Filling Cavities
2. Measurement
2.1. Electrical Contacts
2.2. IPCE Measurement Setup
2.3. Pump Source Characterization
3. Data processing
3.1. Interpolate All Data to 1 nm Increments.
3.2. IPCE Determination
3.3. Model Fitting and Figures of Merit Determination
Figures 3A – D display enhancement responses measured under different measurement conditions, with the effects discussed in more detail below. From the raw current density enhancements it should be clear that the results in Figure 4A and 4B are attributable to upconversion, with the peak current enhancement and IPCE enhancement matching well with the absorption spectrum of the sensitizer, attenuated by transmission through the active layer of the DSC.
In order to avoid measurement artifacts introduced by laser biasing the pump beam has been adjusted to arrive at the UC layer at a greater angle to the probe beam, shown schematically in Figure 2. Figure 4A shows enhancement without significant biasing effect, whereas both Figures 4C and 4D are affected by this problem. The consequence of correct alignment on measurements is shown in Figure 4A where the difference in JSC reflects the absorption property of the sensitizer which has an absorption peak at 675 nm. Apart from the absorption region of the sensitizer and the transparent region of the device, the difference in JSC is embedded in noise.
A significant relative IPCE enhancement of the integrated device in the red end of visible spectrum can be observed in Figure 4C. However, the insert of Figure 4C which shows the difference between aligned and misaligned JSC measurements, does not reflect the spectral property of the sensitizer. The alignment of pump and probe seem to enhance the cell performance across the entire visible spectrum and suggests that the enhancement comes from trap-filling which enhances the overall performance of the device, due to laser biasing30.
In order to verify the suspicion, the integrated device was replaced by an analogous device except that the UC chamber was left empty (Figure 4D). Under the identical experimental condition, enhancement has been found across the visible spectrum. It confirms the previous enhancement effect comes from laser-biasing instead of TTA-UC. In the case of the device without TTA-UC solution, since the majority of the laser is scattered back to the device, the biasing effect is even more significant.
Figure 5 expands upon the results shown in Figures 4A and 4B. In this case, the light intensity of the pump beam was adjusted from 6 to 27 ʘ. ΔJSC is seen to scale with the square of light intensity, as per expectation (power law fit 2.02). As such, the FoM is seen to be light intensity independent, suggesting that the TTA-UC system is limited by bimolecular processes.
Animation 1: Schematic operation of triplet-triplet annihilation photon up-conversion with PQ4PdNA sensitizer and rubrene emitter, resulting in illumination of D149 dye and subsequent electron injection into TiO2. Please click here to view a larger version of this figure.
Figure 1. Device configuration, prior to introduction of liquid components. Layers are placed together and sealed by application of heat to soften the gasket layers. Please click here to view a larger version of this figure.
Figure 2. Setup for the enhancement measurement. The integrated device is irradiated by modulated incoherent monochromatic light from a white light source (laser-driven lamp) passed through a monochromator, and achromatically focused onto the sample by an off-axis parabolic mirror. The probe light is split with a glass filter (beam slitter) and the reflected probe light is detected by a photodiode attached to a power meter. The TTA-UC layer of the integrated device is continuously excited by a 670 nm continuous wave laser (pump) to generate background triplets to allow the TTA-UC enhancement effect to be probed with the weak monochromatic beam. The output current from the device is fed through a current amplifier and measured by lock-in amplification. Please click here to view a larger version of this figure.
Figure 3. Representative data showing (A) the relative IPCE enhancement (col(Aligned)/col(Misaligned) and response difference (col(Aligned)-col(Misaligned) averaged from six aligned and six misaligned measurements. The response difference confirms the spectral shape of IPCE enhancement is from sub-bandgap light harvested by the sensitizer of the up-convertor, as the enhancement spectral shape matches to the Q-band absorption of the sensitizer and (B) the relative enhancement model fitted (described previously31) onto an experimental IPCE enhancement curve by least-squares fitting. The model includes cell transmittance, the original cell IPCE (no pump) and the sensitizer absorption cross section corresponding to probe and pump source. The modeled enhancement curve is then used for calculating additional short-circuit current generated from TTA-UC and thus FoM. Please click here to view a larger version of this figure.
Figure 4. IPCE enhancement (under 27 ʘ) traces for (A) an integrated device with correct misalignment measurement (inset showing the gain in raw response), (B) modeled relative IPCE enhancement trace for data in (A) with inset showing the raw current response curves of the device with pump and probe beam aligned and misaligned (C) the same device as in (A) excepting that the pump and probe being aligned at the same site on the active electrode, resulting in a measurement artifact, described in the text (D) an identical device with an empty UC chamber, measured as per (C), further highlighting this measurement problem, with inset showing the gain in raw response. Please click here to view a larger version of this figure.
Figure 5. FoM dependence of the integrated device on solar concentration factor. Inset shows the dependence of calculated current gain (ΔJSC) from TTA-UC with both axes on a logarithmic scale.
This protocol provides a means to achieve photon up-conversion enhanced DSC and detail on how to correctly measure such a device. The FoM allows for the simple calculation of anticipated ΔJSC improvements to be expected at different light intensities, including at 1 sun. The values shown here are invariant with light intensity (inset of Figure 4), as per expectation when the system is below its saturation threshold33. With the FoM, we can standardize the enhancement effect of TTA-UC or other non linear UC processes to allow easy comparison.
Although the FoM values obtained in this study are the highest among the reported FoMs for DSCs, they are still far from commercial interest (~1 mA∙cm-2ʘ-2). In addition to this, enhancements of this scale can be problematic to measure. In this report (specifically in Figures 3C and 3D) the perils of incorrect measurement technique are shown, where the pump beam causes a (somewhat) unexpected problem. This issue may be unique to DSCs, however if there is any uncertainty it is crucial that control experiments (such as shown in Figure 3D) are undertaken and conditions modified accordingly.
There are a few limiting factors that restrict the performance of TTA-UC. First of all is the triplet decay rate of the emitter, rubrene (~8,000 sec-134), which is much faster than the excitation rate of the sensitizer under 1 ʘ illumination (6.8 sec-1), while the TTA rate of rubrene triplets is only ~1 × 108 M-1sec-1, three orders of magnitude below the diffusion limit of rubrene in common organic solvents35. The consequence of this is the majority of triplet rubrene decays to the ground state before performing TTA.
In order to reduce the amount of rubrene triplets undergoing unimolecular decay before TTA one can attempt to increase the triplet concentration, by increasing the sensitizer concentration. Unfortunately, porphyrins in solution tend to aggregate at high concentrations, and sensitizer-sensitizer TTA may take place. A potential solution overcome these issues is to attach sensitizers onto inorganic nanoparticle surfaces36. As such, high concentrations of (relatively) immobilized sensitizer can be accommodated with reduced self-quenching, and may increase the local concentration of triplets available for efficient TTA.
The sensitizer used in this study is not ideal for the coupled DSC, as the Q-band absorption of the porphyrin overlaps with the DSC’s absorption onset (600 – 700 nm). Thus there are losses in transmitted light available for TTA-UC, the efficiency of which depends on triplet concentration and thus photon flux. We expect to measure a more significant enhancement with a sensitizer that absorbs deeper into the near infrared with similar intersystem crossing efficiency to the one used in this study. The FoM offers a convenient metric of comparison, if and when such a system is characterized.
The dye used here, D149, is among the best performing organic dyes available for DSC, however others, such as N719 or “black dye” have further red-shifted absorption onsets3. In order for TTA-UC to enhance these devices, appropriate porphyrins with Q-band absorptions at wavelengths greater than 900 nm need to created. On the other hand, the highest reported DSC efficiency to date has an absorption onset of ~730 nm37, only marginally beyond the onset for the dye used here.
The authors have nothing to disclose.
A.N. acknowledges contributions from the Australian Renewable Energy Agency (ARENA) and the Australian National Fabrication Facility (ANFF). This research project is funded by the Australian Solar Institute (6-F020 and A-023), with contributions from The New South Wales Government and the University of Sydney. Aspects of this research were supported under Australian Research Council’s Discovery Projects funding scheme (DP110103300). Equipment was purchased with support from the Australian Research Council (LE0668257).
Name of Material/ Equipment | Company | Catalog Number | Comments/Description |
(tetrakis(3,5-di-tert-butylphenyl)-6’-amino-7’-nitro-tetrakisquinoxalino[2,3-b'7,8-b''12,13-b'''17,18-b''''-porphyrinato) palladium(II)) | in house | in house | Chem. Commun., 4851–4853 (2007) |
1,2-dimethyl-3-propylimidazolium iodide | Solaronix | 33150 | Material warning: Irritant |
405 nm longpass filter | Semrock | BLP01-405R-25 | – |
670 nm laser | Thorlabs | LDS5 + CPS198 | – |
Acetone | Chemsupply | AA008-20L-P | Material warning: Flammable |
Acetonitrile | Sigma | 271004 | Material warning: Flammable |
Alumina | Alfa Aesar | 12733 | – |
Alumina | Leeco | 810-782 | – |
Back filling chamber | Sistema | 1303 | Kilip it round, modified |
Benzene | Scharlau | BE0033 | Material warning: Toxic |
BNC cable | Jaycar | RG- 59U | – |
Cerasolzer | MBR | CS186 | – |
Chopper wheel | Thorlabs | MC1000A | – |
Control software | in house | in house | Written in LabVIEW |
Current Amplifier | Standford Research | SR 570 | – |
D149 dye | 1m | OSO149 | – |
Dental burr | Priority dental supplies | 835.104.008 | – |
Detergent | Palmolive | Original | – |
Diamond wheel | Frameco | 14220 | – |
Drill | Dremmel | 220 | – |
Dynamic dignal acquisition device | National Instruments | USB-4431 | Analog to Digital |
Ethanol | Univar | 214 | Material warning: Flammable |
F:SnO2 glass | Hartford | TEC8 | 2.3mm, < 8 Ω/□ |
Glovebox | IT systems | – | – |
H2PtCl6 | Sigma | 334472 | Material warning: corrosive |
Hot melt adhesive gasket | Solaronix | Meltronic 1170-25 | Surlyn |
Hot melt adhesive gasket | Solaronix | Meltronix 1170-60 | Surlyn |
Hotplate | Harry Gestigkeit | PR 5 3T / PZ28-3T | – |
Hotplate | IKA | RCT basic | – |
Image analysis software | National Institutes for Health | Image-J | – |
Iodine | Sigma | 326143 | Material warning: corrosive |
Laser engraver | Universal Laser Systems | PLS6WM | – |
Liquid Nitrogen | Air Liquide | – | |
Lithium Iodide | Aldrich | 518018 | Material warning: toxic |
Methoxypropionitrile | Sigma | 65290 | Material warning: Flammable |
Mirror | Thorlabs | PF10-03-P01 | – |
Mirror mount | Thorlabs | KM100 | – |
Monochromator | Spectral Products | CM110 | – |
Neutral density filters | Edmund Industrial Optics | 64-352 | – |
Parabolic mirror | Newport | 50329AL, 50338AL | – |
Photodiode | Newport | 918D-UV-OD3 | – |
Power meter | Newport | 1936-C | – |
Rubrene | Sigma | 551112 | – |
Semi-automatic screen printer | Keywell | KY-500FH | – |
Spray pyrolyser | Glaskeller | – | – |
Tape | 3M | Magic Tape | – |
Terminal block | Jaycar | HM3194 | – |
tert-Butanol | Sigma | 360538 | Material warning: Flammable |
TiCl4 | Sigma | 89545 | Material warning: corrosive |
Tile | Johnson tiles | – | – |
Tile cutter | DTA | DTA-310 | – |
TiO2 paste | Dyesol | NR18-T | – |
Titanium diisopropoxide bis(acetylacetonate) (75% in isopropanol) | Aldrich | 325252 | Material warning: Flammable |
Ultrasonic soldering iron | MBR | USS-9200 | – |
UV cure epoxy | Dymax | 425 | Material warning: Irritant |
UV cure system | Dymax | BlueWave 50 | – |
UV Visible Spectrophotometer | Varian Cary | 1E | – |
Vacuum cuvette | Custom made | Custom made | – |
Vacuum pump | N/A | Rotary backed diffusion pump | – |
Wipes | Kimtech | 34120KC | Kimwipes |
Xe lamp | Energetiq | LDLSTM EQ-1500 | White light source |