- Email: [email protected]
Multi-dimensional titanium dioxide with desirable structural qualities for enhanced performance in quantum-dot sensitized solar cells
Journal of Power Sources 282 (2015) 202e210
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
Multi-dimensional titanium dioxide with desirable structural qualities for enhanced performance in quantum-dot sensitized solar cells Dapeng Wu a, b, Jinjin He a, Shuo Zhang a, Kun Cao a, Zhiyong Gao a, b, Fang Xu a, Kai Jiang a, b, * a
School of Chemistry and Chemical Engineering, Henan Normal University, Xinxiang, Henan, 453007, PR China Engineering Technology Research Center of Motive Power and Key Materials of Henan Province, Henan Key Laboratory of Photovoltaic Materials, Henan Normal University, Xinxiang, Henan, 453007, PR China
b
h i g h l i g h t s A bridge-linking mechanism was proposed to illustrate the formation of MD-THS. The wide pore size distribution facilitates the anchoring process of quantum dots. The oriented aligned primary nanocrystals inhibit undesired charge recombination. Cell-MD-THS demonstrates much higher PCE of 4.15% than Cell-Nanocrystal (3.03%).
a r t i c l e i n f o
a b s t r a c t
Article history: Received 8 October 2014 Received in revised form 8 January 2015 Accepted 10 February 2015 Available online 11 February 2015
Multi-dimensional TiO2 hierarchal structures (MD-THS) assembled by mesoporous nanoribbons consisted of oriented aligned nanocrystals are prepared via thermal decomposing Ti-contained gelatin-like precursor. A unique bridge linking mechanism is proposed to illustrate the formation process of the precursor. Moreover, the as-prepared MD-THS possesses high surface area of ~106 cm2 g1, broad pore size distribution from several nanometers to ~100 nm and oriented assembled primary nanocrystals, which gives rise to high CdS/CdSe quantum dots loading amount and inhibits the carries recombination in the photoanode. Thanks to these structural advantages, the cell derived from MD-THS demonstrates a power conversion efficiency (PCE) of 4.15%, representing ~36% improvement compared with that of the nanocrystal based cell, which permits the promising application of MD-THS as photoanode material in quantum-dot sensitized solar cells. © 2015 Elsevier B.V. All rights reserved.
Keywords: Titanium dioxide Hierarchical structure Growth mechanism Quantum-dot sensitized solar cells
1. Introduction Quantum-dot sensitized solar cell (QDSSC) represents a group of new solar energy conversion devices. Although it shares many design configuration and working principle with dye-sensitized solar cell (DSSC), the Quantum dots (QDs) sensitizers, generally the metal chalcogenides, have many inherit advantages compared with the costly Ru based dyes, such as the flexible solar spectrum matching capability, high extinction ratio and the unique multiple excitation generation (MEG) resulting from the impact ionization effect [1e4]. Because of the MEG effect, a high theoretical PCE of
* Corresponding author. School of Chemistry and Chemical Engineering, Henan Normal University, Xinxiang, Henan, 453007, PR China. E-mail address: [email protected] (K. Jiang). http://dx.doi.org/10.1016/j.jpowsour.2015.02.062 0378-7753/© 2015 Elsevier B.V. All rights reserved.
~44% could be achieved, which is much higher than the traditional Shockley and Queisser limit of ~31% [5]. CdS/CdSe co-sensitized strategy is commonly used to achieve high PCE in QDSSC [6e9]. In addition, other sensitizers such as CdTe, PbS and PbSe were also developed to enhance the PCE by broadening the absorption spectrum or improving the injection efficiency of the photoelectrons [10e12]. Most recently, QDSSCs based on low toxicity sensitizers (Ag2S, Sb2S3 or CuInS2), solid-state or flexible configurations were also designed and prepared [13e16]. However, the highest PCE of these devices merely reaches to ~7%, which still lags far behind compared with DSSC (~13%) [17e20]. The low PCE generally results from the massive recombination in the unsophisticated cell configurations, indicating the photoanode, electrolyte and the counter electrode are not satisfactorily optimized. [21e24]. Photoanode is one of the key components in QDSSCs, which
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
serves as both the scaffold to anchor QDs and diffusing matrix for the injected photoelectrons, many groups have made various efforts to optimize the TiO2 structures so as to enhance the light harvesting capability and electron transport performance [25e29]. Several design rules should be considered as priorities. First of all, the QDs have much greater size than the monolayer adsorbed dye molecules, thus a photoanode film with large pore size distribution is favorable for the QDs loading. Especially in the pre-synthesis strategies, in which the QDs are prepared before loaded on the photoanode, the small pore size inevitably jeopardizes the QDs penetration and results in QDs aggregation inside the film structures [30,31]. Secondly, the high surface area could increase the initial nucleation population of the QDs on the photoanode structures, which can improve the QDs coverage and retard undesired interface recombination [27,32e34]. Particularly, for the in situ growth strategies, in which the QDs were loaded on the photoanode film by means of successive ionic layer adsorption and reaction (SILAR) or chemical bath deposition (CBD), photoanode film with high surface area is of same significance as the pore size distribution. Thirdly, although the continuous QDs coating layer may serve as additional electron transport path, the majority of the generated photoelectrons are collected through the TiO2 matrix. Therefore, just like the DSSC cases, hierarchical structures consisted of one-dimensional subunits or orderly aligned primary units can greatly improve the electron collection efficiency by minimizing the population of the grain boundaries [35]. However, the previously reported hierarchical structures can hardly balance the required structural qualities of high surface area, broad pore size distribution and oriented aligned subunits. Especially for the TiO2 hierarchical structures with small pore size distribution, the QDs aggregation inevitably takes place in the pore channels, which results in increased charge recombination and leads to PCE plummet. Therefore, the TiO2 hierarchical structures should be further optimized for the particular demands in QDSSCs. Herein, we designed and prepared a unique MD-THS which is
203
assembled by mesoporous nanoribbons consisted of oriented aligned nanocrystals. Time dependent trails reveal an interesting growth and assembling process of the hierarchal gelatin precursor and the formation is driven by the energy minimization on the coordinating modes of carboxyl groups. In addition, when used as photoanode material in CdS/CdSe co-sensitized cells, the MD-THS could integrate the advantages of different dimensional materials: the three dimensional (3D) hierarchical structure provides a suitable outer diameter of ~1 mm and a wide pore size distribution from several nanometers to ~100 nm, which improves the harvesting of incident light. The one dimensional (1D) nanoribbon with oriented assembled primary nanocrystals reduces the grain boundaries and charge recombination. Moreover, the zero dimensional (0D) nanocrystals with diameter of ~15 nm leads to high surface area of ~106 cm2 g1, which favorites the loading process of QDs sensitizers. Therefore, the cell derived from MD-THS demonstrates a high PCE of 4.15%, which indicates ~36% improvement compared with nanocrystal based cell (3.03%). 2. Experimental section 2.1. Materials synthesis The MD-THSs were prepared in a solvothermal method. 1 mL of titanium n-butoxide (TBT) was added drop wise to 20 mL of acetic acid with rigorous stirring. Afterwards, the mixture was transferred into 25 mL Teflon-lined stainless steel reactor and heated at 200 C for 2 h. After cooling, the white product were collected by centrifugation and washed with distilled water and ethanol several times, then dried at 60 C for 3 h. Finally, the samples were heated to 450 C at a rate of 5 C min1 and calcinated for 3 h to obtain the final product. The nanocrystals were prepared based on M. Gratzel et al.'s work [36]. Briefly, 7.5 mL titanium isopropoxide was added into 45 mL 0.1 M HNO3 aqueous solution. The suspension was placed in
Fig. 1. FESEM images of (a) hierarchical gelatin precursor obtained by solvothermal treatment; (b) MD-THSs after calcination; (c) the magnified image located at the black dot square in (b); (d) and (e) are the corresponding TEM and HRTEM images of the mesoporous ribbons.
204
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
80 C water bath and stirred for 8 h. The concentrated suspension was then transferred into 50 mL autoclave and heated at 230 C for 12 h. The final product was collected by centrifugation and stocked in wet for later use. The as-prepared nanoparticles have average diameter of ~15 nm and BET surface area of ~110 m2 g1.
2.2. Fabrication of the QDSSCs The cleaned fluorine-doped tin oxide (FTO 15 U square1, NSG, Japan) glass was immersed in TiCl4 aqueous solution (0.15 M) in a beaker and then kept in an oil bath at 70 C for 1 h to coat a thin layer of TiO2 on the surface for the sake of forming a barrier layer. The treated FTO was calcinated at 450 C for 20 min. The paste was prepared according to our previous literature [37]. The prepared paste was coated onto the FTO layer using a traditional doctor-blade method with adhesive tape to be rough in the middle area as frame and spacer. Then, the coated FTO glass flake was calcinated at 500 C for 30 min to remove organic reagent. A modified chemical bath deposition method [25] was applied to load QDs on TiO2 photoanodes. An aqueous solution, with the composition of 0.02 mM CdCl2, 0.14 mM thiourea, 0.07 mM NH4Cl and 0.23 mM ammonia with final pH of ~9, was used to deposite CdS QDs. The reaction was carried out in 10 C water bath for 1 h. After rinsed with water, the TiO2/CdS films were immersed in an aqueous solution containing 0.026 mM CdSO4, 0.04 mM N(CH2COONa)3 and 0.026 mM Na2SeO3 for 4 h in the same way. Both the QDs (CdS and CdSe) depositions films were fully washed with water. For the ZnS passivation, the sensitized films were deposited by twice dipping alternatively into 0.1 M Zn(CH3COO)2 and 0.1 M Na2S solutions for 1 min per dip, rinsed with distilled water between dips. Cu2S in situ deposited on brass was used as a counter electrode. The QDs sensitized photoanode and counter electrode were sandwiched to form the cell. The electrolyte composed of 1 M Na2S and 1 M S in deionized water was injected into the cell via capillarity. A metal mask with a window of ~0.2 cm2 was covered on the back side of photoanode to control the active area.
3. Results and discussion Fig. 1a shows FESEM images of the as-prepared gelatin precursor which is assembled from 1D nanoribbons and has an average diameter of ~1.5 mm. As depicted in Fig. 1b, the average diameter of the peony-like structures decrease to ~1 mm after calcination, indicating a ~30% shrinkage due to the decomposition of the gelatin precursor [38]. The corresponding TGA measurement shows the product experiences ~40% weight loss during the calcination (Fig. S1). Moreover, the black box area was magnified and displayed in Fig. 1c, which shows the ribbons are 10e20 nm in thickness and assembly by numerous primary nanocrystals. The TEM image in Fig. 1d indicates that the nanocrystals are ~15 nm in diameter and closely linked together. Moreover, the HRTEM image in Fig. 1d depicts the lattice fringes of two adjacent nanoparticles, which indicates the primary nanocrystals are linked with same orientation. The spacing lattices are well-defined and measured to be 0.35 nm, suggesting they are highly crystallized and with anatase structure. Fig. 2 a shows the XRD pattern of the precursor and MD-THSs. The precursor exhibits unknown diffraction peaks, which is in accord with the previously reported results [38]. After calcination, the diffraction peaks of MD-THSs could be well indexed to the tetragonal anatase phase (JCPDS cards No.01-0562), suggesting the precursor with unknown crystal phase is completely converted into anatase TiO2. The N2 adsorptionedesorption isotherms and the pore size distribution of the MD-THSs are shown in Fig. 2b. The
2.3. Material characterization The samples were studied by field-emission scanning electron microscope (FESEM, HITACHI, S4800), transmission electron microscopy (TEM, FEI, Tecnai F30), X-ray powder diffraction (XRD, Rigaku D/max-2500 diffractometer with Cu Ka radiation, l ¼ 0.1542 nm, 40 kV, 100 mA) and BrunauereEmmetteTeller (BET, Micrometrics ASAP 2010). The optical diffusion reflection spectra and absorption spectra were measured using a PerkineElmer Lambda 35 UVeVis spectrometer. The thickness of films was measured on a profilometer (DEKTAK 150, Veeco Instruments Inc.). Moreover, X-ray energy dispersive spectroscopy (EDX, Oxford, INCA MICSFþ) was also used to perform the element mapping of photoanodes loaded with QDs. Fourier transform infrared spectroscopy was characterized by (FTIR, FTS-40).
2.4. Photovoltage measurements Photocurrentevoltage (IeV) test were measured using a Keithley 2400 sourcemeter, and the output power density of the xenon lamp solar simulator (Newport) was adjusted to 100 mW cm2 (AM1.5). Incident photon to current conversion efficiency (IPCE) was also measured on the 7-scspec system (Saifan Beijing) with a 1/ 4 m monochromator.
Fig. 2. (a) XRD pattern of the precursor sample and the MD-THS (b) Nitrogen adsorptionedesorption isotherms of the MD-THS, the inset in (b) is the corresponding pore size distribution curve.
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
205
Fig. 3. SEM images of the products harvested at different time intervals through solvothermal reaction: (a) 0 min, (b) 30 min, (c) 1 h, (d) 2 h, (e) 3 h and (f) 6 h.
isotherm exhibits typical type IV curve with H1-type hysteresis at high relative pressures, which is usually ascribed to the predominance of mesopores and the existence of pore connectivity. In addition, a broad pore size is observed in the distribution curve (the inset of Fig. 2b), which ranges from several nanometers to ~100 nm. The broad pore size distribution may favorite both the loading of the quantum dots and the electrolyte diffusion in the film structure. Moreover, the BET measurement indicates that the sample possesses a high surface area of ~106 m2 g1. In order to reveal the formation mechanism of MD-THS, the solvothermal process was monitored by time dependent experiments. When the reaction temperature reaches to 200 C, the reactor was immediately fetched out and cooled in water and the sample was denoted as 0 min. At this initial stage (Fig. 3a), the firstly formed gelatin is soluble both in water and alcohol and no particular morphology was observed. Although the corresponding XRD pattern (Fig. 4a) cannot match with any of the existed crystal structure data, it is obviously different from the next generated gelatin structures. After 30 min, 1D ribbons with interlaced structure are formed in the sample (Fig. 3b). When the reaction proceeds to 1 h, the ribbons are inclined to assemble into 3D structures. At this stage, individual ribbons could be founded in the sample, suggesting the assembling process is not finished (Fig. 3c). After 2 h, the ribbons are fully assembled into 3D spherical hierarchical structures. During this process, the XRD patterns of the gelatins were not changed, indicating that the gelatin simply experiences an
assembling process without structural variation (Fig. 4a). As it was demonstrated previously, the gelatin-like intermediate products could act as a reservoir to gradually release soluble titaniumcontaining species for the nucleation and growth of TiO2 spindles. As displayed in Fig. 3d, TiO2 spindles are found coexisted with the gelatin structures after 3 h reaction, which could be confirmed by XRD investigation. When the time reaches to 6 h, the gelatin precursor is fully converted into TiO2 spindles (Figs. 3f and 4a). Moreover, the growth process was also monitored by the FTIR spectra, as displayed in Fig. 4b. The peaks centered at 1719, 1525, 1463, 1417 and 1256 cm1 can be ascribed to the different vibration modes of the carboxylic groups coordinated to Ti [39]. The two distinct vibration signals centered at around 1719 and 1256 cm1 indicate the existence of monodentate coordination. The peaks at the frequencies of 1463 and 1417 cm1 demonstrate approximate frequency separations of Dn ¼ 62 and Dn ¼ 108 cm1 between the signals around 1525 cm1, which respectively represents the bidentate and the bridging modes of the acetates coordinated to the titanium atoms [37,40]. In addition, the peak intensities experience gradual receding as the reaction prolonged, implying the monodentate and bidentate ligands were removed from the products. As the reaction time increases to 6 h, TieOeTi rigid frameworks with high stability are formed, which could be indicated from the broad and strong absorption around 600 cm1 assigning to the TieOeTi moiety of anatase TiO2. Base on the control experiments, a possible growth mechanism
206
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
of MD-THS was proposed and illustrated in Figs. 5 and 6. The formation process of the Ti-contained gelatin is shown in Fig. 5a. At the very beginning, the coordinating carboxyl groups would bond to the titanium atom along with the ester groups to generate a hexa-coordinated titanium central atom, which favorites the formation of anatase TiO2 (Step 1). During this process, except the chelating coordination, monodentate coordination mode may also form. As the reaction proceeded, the adjacent Ti atoms could be linked together by converting the acetate ligand from chelating coordination to bridging mode (Step 2). The ester groups of the asformed two-atom system would be then substituted by the abundant acetate groups (Step 3). Subsequently, the bridging effect of the acetate could further link the Ti atoms into a four-atom system (Step 4). Thereafter, the repeating linkage forms the gelatin structure and induces the later assembling process (Step 5). Moreover, the acetate molecules with chelating and bridging structural modes are compared after the energy minimization (Fig. 5b). It is found that in the chelating mode, the average bond length of CeO and OeTi bonds are 1.486 Å and 1.984 Å. Meanwhile, in the bridging mode, the values are 1.433 Å and 1.975 Å. The decreased bond length indicates the bridging coordination is more energy favorable
Fig. 4. (a) XRD and (b) FTIR investigations of the intermediate samples collected at 0 min, 30 min, 1 h, 3 h, and 6 h through solvothermal reaction.
during the formation of the hierarchical gelatin. The assembling process of the MD-THS is illustrated in Fig. 6. At the initial stage of the solvothermal reaction, only gelatin with small size and low molecular weight is formed, which makes them easily dissolved in water and ethanol. The high reaction temperature causes high rate collision of the gelatin particles, which leads to the linkage of the gelatin by the bridging coordination of acetate. Therefore, 1D ribbon-like structures are formed afterwards (Step 1). Moreover, the cross-link of ribbons could further reduce the system free energy by forming interlaced ribbons and the intermediate hierarchical structures (Step 2). Considering the sphere-like pattern could greatly reduce the surface tension at the solid and liquid interface, 3D spherical gelatin structures with radially aligned 1D ribbons are finally formed (Step 3). After calcination, the removal of organic composites finally generates the peony-like MD-THS assembled by interlaced mesoporous ribbons. Considering the unique structural qualities of the as-prepared MD-THSs, they were adopted as the photoanode material for QDSSC. In order to optimize photoanode structure, the deposition times and the film thickness of the mesoporous film were controlled. Fig. S2 and Table S1 show the current densityevoltage (IeV) characteristics of the cells prepared with different depositing times. The absorption intensity of the photoanode continuously improves with more reaction time (Fig. S3), indicating the loading amount of the sensitizers increases. However, the overall PCE starts to drop after 4 h, which probably results from that the increased thickness of the quantum-dots layer reduces the electron injection efficiency and brings forth more recombination. Moreover, the thickness of photoanode film was also varied to investigate its influence over the PCE. The film thicknesses were respectively controlled at ~8, ~12, ~15, ~17 and ~20 mm. In addition, a semitransparent film consisted of ~15 mm nanocrystals was also prepared as reference. The IeV curves of the as-prepared cells with different film thicknesses are presented in Fig. 7 and the corresponding photovoltaic characteristics are summarized in Table 1. For the MD-THS photoanodes with different film thicknesses, the Jsc and Voc improve from 8.20 to 14.05 mA cm2 and from 526 to 571 mV, as the thickness increases from ~8 to ~15 mm. It is reasonable to anticipate that the thicker film leads to greater quantum dots loading amount, which generates higher photoelectron concentration in the TiO2 conduction band and then leads to the Jsc augment. In addition, the increased free electron population could also elevate the quasi-Fermi level and result in higher Voc. However, the Jsc and Voc decrease, if the film thickness further increases to ~17 and ~20 mm. Different from the fluctuation of Jsc and Voc, the filling factor (FF) of the cells experiences monotonically decrease as increasing the film thickness. The FF largely depends on the series resistance of the film, and thicker film with higher inner resistance would inevitably decrease the FF [41]. Therefore, the PCE of the Cells with film thicknesses of ~17 and ~20 mm respectively drop to 3.77% and 2.59%. The best performance is achieved by CellMD-THS with film thickness of ~15 mm, which demonstrates Jsc of 14.05 mA cm2, Voc of 571 mV, FF of 51.7%, and PCE of 4.15%. Moreover, the Cell-Nanocrystal with photoanode film thickness of ~15 mm was also investigated, which leads to a PCE of 3.03%. Compared with Cell-Nanocrystal, it is obvious that the high PCE of Cell-MD-THS is attributed to the synergetic improvement on Jsc, Voc and FF. The cross-sectional SEM image for the QDs sensitized MD-HTS photoanode is shown in Fig. 8a. The MD-THS is evenly deposited on the FTO substrate with thickness of ~15 mm. Moreover, the uniform coverage of QDs on TiO2 throughout the film thickness is confirmed by elemental mapping via EDX analysis. As shown in Fig. 8bed, the elemental mapping on the rectangular area demonstrates uniform distribution of Ti, Cd, and Se atoms throughout
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
207
Fig. 5. (a) Possible formation mechanism of the gelatin: continuous bridge-linking effect of acetate ligand among the Ti atoms; (b) the structural comparison of acetate molecules with chelating and bridging coordination modes.
Fig. 6. Schematic illustration of the growth mechanism of MD-THS.
Fig. 7. IeV curves of the CdS/CdSe/ZnS QDSSCs made from hierarchical TiO2 with different thicknesses and the as-prepared nanoparticles TiO2 film (under one sun illumination (AM 1.5, 100 mw/cm2).
208
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
Table 1 Photovoltaic characteristics of the as-prepared QDSSCs (the active area is ~0.2 cm2). Cells
MD-THSs
Thickness (mm) Voc (mV) Jsc (mA cm2) FF (%) PCE (%)
~8 526 8.20 60.7 2.62
~12 548 11.33 53.0 3.29
Nanoparticles ~15 571 14.05 51.7 4.15
~17 567 12.91 51.5 3.77
~20 509 11.98 42.5 2.59
~15 550 10.77 51.2 3.03
the film depth. Atomic percentages from EDX elemental analysis on the film cross section were found to be 22.08, 3.83 and 3.06% for Ti, Cd, and Se respectively. Moreover, a similar test was performed in nanoparticles based film, the atomic percentages were 24.80, 3.89 and 2.86% for Ti, Cd, and Se. (Table S2) Based on the EDX test, the Cd/Ti ratio in the MD-THS film could be determined as ~0.173, which is greater than that in the nanocrystal film of ~0.157 (Fig. S4). The uniform distribution of the QDs as well as the high Cd/Ti ratio in MD-THS film offer solid support to the viewpoint that although MD-THS has similar surface area with nanocrystals, the wide pore size distribution could facilitate the QDs penetration throughout the film depth, which leads to higher QDs loading amount [42]. Fig. 9a plots the optical absorption spectra of the nanoparticle and HM-THS photoanode films after CdS/CdSe sensitization (~15 mm). Film-MD-THS has a higher absorbance in the whole
region, indicating the higher QDs loading amount and light harvest efficiency. In addition, as shown in Fig. 9b, the optical bandgap for the CdSe QDs deposited in difference film structures could be evaluated using the equation for the near-edge absorption and the extrapolation of the linear region of the plot of (Ahn)2 against photon energy (hn) [39,43]. The direct forbidden bandgap of ~1.95 eV and ~1.99 eV could be determined and the redshift can be attributed to the quantum-size effect, suggesting larger CdSe QDs are formed inside the MD-THS film structure. Moreover, as displayed in Fig. 9c, Film-MD-THS exhibits a prominent scattering effect at the wavelength of 400e800 nm, which could further improve the harvesting of the visible and near-infrared light. It can be concluded that the large pore size distribution of MD-THS not only increases the QDs loading amount but also leads to the formation of QDs with greater diameters, which improves the absorption of incident light and stretches the absorption spectrum to longer wave range. Moreover, the micrometer sized hierarchical structure could induce multiple-scattering of the incident light to further improve the light harvesting, which finally leads to the high Jsc. Therefore, as depicted in Fig. 9d, Cell-MD-THS demonstrates a higher IPCE value compared with Cell-Nanoparticles. Moreover, as it is shown in Eqn. (1) [44]
V∞
RT AI h i ln ¼ bF n0 k0 Sn þ n0 Kr Dþ
!
(1)
Fig. 8. (a) Cross-sectional SEM image of ~15 mm thick QDs sensitized MD-HTS film; (bed) Elemental mapping of titanium (b), cadmium (c), and selenium (d) by EDX spectroscopy for the rectangular area indicated by the black box in panel a showing the uniform distribution of the QDs sensitizer throughout the film thickness.
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
209
Fig. 9. (a) The absorption spectra of nanoparticles and HM-THS films (~15 mm in thickness) with CdS/CdSe sensitization; (b) the corresponding (Ahn)2 vs hn spectra; (c) diffusereflectance spectra of photoanode film without sensitization; (d) the IPCE spectra of the corresponding cells.
where R is the molar gas constant, T the temperature, b the reaction order of S n and electrons, F the Faraday constant, A the electrode surface area for light absorption, I the incident photon flux, n0 the accessible electronic state concentration in the conduction band, kb the backward reaction kinetic constant of the injected electrons with polysulfide electrolyte and kr the recombination kinetic constant of the electrons with oxidized QDs (Dþ), the Voc of the device is largely determined by the recombination of the photoelectron with the oxides species in the electrolytes. In order to investigate the charge recombination process, the Voc decay test was carried out and displayed in Fig. 10. At the open circle conditions, as the illumination is removed, the photogenerated electrons and holes would recombine via the trap states in the photoanode, which leads to the exponential Voc decay [45,46]. The dots shown in Fig. 10a represent the measured Voc data points after the illumination is removed. After fitted with exponential decay functions, the solid lines and the inset normalized decay curves prove that Cell-Nanocrystal exhibits an evident faster Voc decay, indicating a more rapid recombination process. Moreover, Based on Eqn. (2) [47,48]
k T tn ¼ B e
dVoc dt
1 (2)
the electron lifetime (tn) in function of Voc could be obtained by differentiating the decay curves and then taking an inversing, which are displayed in Fig. 10b. It is obvious that Cell-MD-THS has greater tn than Cell-Nanocrystal in the whole voltage range, indicating the recombination kinetic constants of kb and kr are reduced
in Cell-MD-THS. Furthermore, based on Eqn. (1) the higher A of Cell-MD-THS could improve the light harvesting and the free electron population in the conduction band of TiO2, which finally results in the elevation of quasi-Fermi level and leads to the higher Voc.
4. Conclusions Multi-dimensional TiO2 hierarchal structures composited of 1D nanoribbions and 0D primary nanocrystals were deliberately designed and prepared for the QDSSC application. It was found the continuous bridging effect of the acetate plays a crucial role in forming the MD-THS. When used as photoanode material in CdS/ CdSe co-sensitized solar cells, a high PCE of 4.15% was demonstrated, indicating ~36% improvement compared with nanocrystal based cell. The structural advantages of MD-THS could be summarized as: (a) the high surface area and wide pore size distribution could increase the QDs loading amount and broaden the absorption spectrum, which enhances the light absorption of the photoanode. (b) The good scattering effect could further improve the harvesting of visible and near-infrared light, which leads to the high Jsc. In addition, (c) the oriented assembled primary nanocrystals provide direct diffusion path for electron collection, which reduces the recombination opportunities of photogenerated carriers and prolongs the electron lifetime, so as to enhance the Voc. Therefore, the unique MD-THS could well integrate the desirable structural qualities of an ideal photoanode material, which enables it a promising candidate for the high performance QDSSCs with different sensitizers.
210
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
Fig. 10. The Voc decay tests of Cell-Nanocrystal and Cell-MD-THS: (a) the Voc decay curves after cut off the illumination, the scattered dots are the experimental data points and the solid lines represent the fitted curves (the inset is the normalized Voc decay curves); (b) electron recombination lifetimes in function of the Voc.
Acknowledgments This work is supported by NSFC (Grant Nos. 61204078, 61176004 and U1304505), Program for Changjiang Scholars and Innovative Research Team in University (PCSIRT). Program for Innovative Research Team (in Science and Technology) in University of Henan Province (No. 13IRTSTHN026), Basic and Frontier Research Programs of Henan Province(No. 142300410420) and the Key Project of Science and Technology of Henan Province (No.13A150517 and 14A150002). Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2015.02.062. References [1] J.J. Tian, R. Gao, Q.F. Zhang, S.G. Zhang, Y.W. Li, J.L. Lan, X.H. Qu, G.Z. Cao, J. Phys. Chem. C 116 (2012) 18655e18662. [2] X.Y. Yu, B.X. Lei, D.B. Kuang, C.Y. Su, J. Mater. Chem. 22 (2012) 12058e12063. [3] X.H. Song, M.Q. Wang, T.Y. Xing, J.P. Deng, J.J. Ding, Z. Yang, X.Y. Zhang,
J. Power Sources 253 (2014) 17e26. [4] X.M. Huang, S.Q. Huang, Q.X. Zhang, X.Z. Guo, D.M. Li, Y.H. Luo, Q. Shen, T. Toyoda, Q.B. Meng, Chem. Commun. 47 (2011) 2664e2666. [5] L.B. Yu, Z. Li, Y.B. Liu, F. Cheng, S.Q. Sun, J. Power Sources 270 (2014) 42e52. [6] R. Zhou, Q. Zhang, E. Uchaker, J. Lan, M. Yin, G. Cao, J. Mater. Chem. A 2 (2014) 2517e2525. [7] J.L. Xu, K. Li, W.Y. Shi, R.J. Li, T.Y. Peng, J. Power Sources 260 (2014) 233e242. [8] J. Tian, E. Uchaker, Q. Zhang, G.Z. Cao, ACS Appl. Mater. Interfaces 6 (2014) 4466e4472. [9] Z. Ali, I. Shakira, D.J. Kang, J. Mater. Chem. A 2 (2014) 6474e6479. [10] C.J. Raj, S.N. Karthick, S. Park, K.V. Hemalatha, S.K. Kim, K. Prabakar, H.J. Kim, J. Power Sources 248 (2014) 439e446. [11] H.P. Shen, X.J. Jiao, D. Oron, J.B. Li, H. Lin, J. Power Sources 240 (2013) 8e13. [12] J. Wang, I.M. Sero, Z.X. Pan, K. Zhao, H. Zhang, Y.Y. Feng, G. Yang, X.H. Zhong, J. Bisquert, J. Am. Chem. Soc. 135 (2013) 15913e15922. [13] T.P. Brennan, O. Trejo, K.E. Roelofs, J. Xu, F.B. Prinzbc, S.F. Bent, J. Mater. Chem. A 1 (2013) 7566e7571. [14] Z.X. Pan, I.M. Sero, Q. Shen, H. Zhang, Y. Li, K. Zhao, J. Wang, X.H. Zhong, J. Bisquert, J. Am. Chem. Soc. 136 (2014) 9203e9210. [15] T.T. Ngo, S. Chavhan, I. Kosta, O. Miguel, H. Grande, R. Tena-Zaera, Appl. Mater. Interfaces 6 (2014) 2836e2841. [16] M.L. Que, W.X. Guo, X.J. Zhang, X.Y. Li, Q.L. Hua, L. Dong, C.F. Pan, J. Mater. Chem. A 2 (2014) 13661e13666. [17] J.H. Im, C.R. Lee, J.W. Lee, S.W. Park, N.G. Park, Nanoscale 3 (2011) 4088e4093. [18] P.K. Santra, P.V. Kamat, J. Am. Chem. Soc. 134 (2012) 2508e2511. [19] S.H. Im, C.S. Lim, J.A. Chang, Y.H. Lee, N. Maiti, H.J. Kim, M. Nazeeruddin, K.M. Gratzel, S.S. Toward, Nano. Lett. 11 (2011) 4789e4793. [20] A. Yella, H.W. Lee, H.N. Tsao, C.Y. Yi, A.K. Chandiran, M.K. Nazeeruddin, E.W.G. Diau, C.Y. Yeh, S.M. Zakeeruddin, M. Gr€ atzel, Science 334 (2011) 629e633. [21] X. Song, M. Wang, J. Deng, Y. Ju, T. Xing, J. Ding, Z. Yang, J. Shao, J. Power Sources 269 (2014) 661e670. [22] F. Wang, H. Dong, J. Pan, J. Li, Q. Li, D.S. Xu, J. Phys. Chem. C 118 (2014) 19590, 1958919598. [23] H. Kim, S. Kim, C. Gopi, S. Kim, S. Rao, M. Jeong, J. Power Sources 268 (2014) 163e170. [24] I. Mora-Sero, S. Gimenez, F. Fabregat-Santiago, R. Gomez, Q. Shen, T. Toyoda, J. Bisquert, Acc. Chem. Res. 42 (2009) 1848e1857. [25] Q.X. Zhang, X.Z. Guo, X.M. Huang, S.Q. Huang, D.M. Li, Y.H. Luo, Q. Shen, ́ T. Toyoda, Q.B. Meng, Phys. Chem. Chem. Phys. 13 (2011) 4659e4667. [26] Y.K. Lai, Z.Q. Lin, D.J. Zheng, L.F. Chi, R.G. Du, C.J. Lin, Electrochim. Acta 79 (2012) 175e181. [27] M. Samadpour, S. Gimenez, A.I. Zad, N. Taghavinia, I.M. Sero, Phys. Chem. ́ Chem. Phys. 14 (2012) 522e528. [28] E.H. Kong, Y.J. Chang, Y.C. Park, Y.H. Yoon, H.J. Park, H.M. Jang, Phys. Chem. Chem. Phys. 14 (2012) 4620e4625. [29] H. Dong, D.P. Wu, F. Zhu, F.F. Wang, S.F. Zhu, Q. Li, D.S. Xu, Sol. Energy Mat. Sol. C 133 (2015) 201e207. [30] A. Salant, M. Shalom, Z. Tachan, S. Buhbut, A. Zaban, U. Banin, Nano. Lett. 12 (2012) 2095e2100. [31] A. Salant, M. Shalom, I. Hod, A. Faust, A. Zaban, U. Banin, J. Am. Chem. Soc. 4 (2010) 5962e5968. [32] O. Niitsoo, S.K. Sarkar, C. Pejoux, S. Ruhle, D. Cahen, G. Hodes, J. Photochem. Photobiol. A Chem. 181 (2006) 306e313. [33] H.J. Lee, J. Bang, J. Park, S. Kim, S.M. Park, Chem. Mater. 22 (2010) 5636e5643. [34] X. Song, M. Wang, Y. Shi, J. Deng, Z. Yang, X. Yao, Electrochim. Acta 81 (2012) 260e267. [35] T. Toyoda, Q. Shen, J. Phys. Chem. Lett. 3 (2012) 1885e1893. [36] C.J. Barbe, F. Arendse, P. Comte, M. Jirousek, F. Lenzmann, V. Shklover, M. Gratzel, J. Am. Ceram. Soc. 80 (1997) 3157e3171. [37] D.P. Wu, F. Zhu, J.M. Li, H. Dong, Q. Li, K. Jiang, D.S. Xu, J. Mater. Chem. 22 (2012) 11665e11671. [38] J.F. Ye, W. Liu, J.G. Cai, S. Chen, X.W. Zhao, H.H. Zhou, L.M. Qi, J. Am. Chem. Soc. 133 (2011) 933e940. [39] D. Jiang, Y. Xu, B. Hou, D. Wu, Y.H. Sun, J. Inorg. Chem. 8 (2008) 1236e1240. [40] L. Li, C.Y. Liu, CrystEngComm 12 (2010) 2073e2078. [41] G. Schlichthorl, N.G. Park, A.J. Frank, J. Phys. Chem. B 103 (1999) 782e791. [42] D.P. Wu, X.J. Shi, H. Dong, F. Zhu, K. Jiang, D.S. Xu, X.C. Ai, J.P. Zhang, J. Mater. Chem. A 2 (2014) 16276e16284. [43] A. Hagfeldt, M. Graetzel, Chem. Rev. 95 (1995) 49e68. [44] F. Zhu, H. Dong, Y. Wang, D.P. Wu, J.M. Li, J.L. Pan, Q. Li, X.C. Ai, J.P. Zhang, D.S. Xu, Phys. Chem. Chem. Phys. 15 (2013) 17798e17803. [45] D.P. Wu, Y. Wang, H. Dong, F. Zhu, K. Jiang, L.M. Fu, J.P. Zhang, D.S. Xu, Nanoscale 5 (2013) 324e330. [46] X. Song, M. Wang, H. Zhang, J. Deng, Z. Yang, C. Ran, X. Yao, Electrochim. Acta 108 (2013) 449e457. [47] A. Zaban, M. Greenshtein, J. Bisquert, ChemPhysChem 4 (2003) 859e864. [48] J. Bisquert, A. Zaban, M. Greenshtein, I. Mora-Sero, J. Am. Chem. Soc. 126 (2004), 13550e13359.
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
Multi-dimensional titanium dioxide with desirable structural qualities for enhanced performance in quantum-dot sensitized solar cells Dapeng Wu a, b, Jinjin He a, Shuo Zhang a, Kun Cao a, Zhiyong Gao a, b, Fang Xu a, Kai Jiang a, b, * a
School of Chemistry and Chemical Engineering, Henan Normal University, Xinxiang, Henan, 453007, PR China Engineering Technology Research Center of Motive Power and Key Materials of Henan Province, Henan Key Laboratory of Photovoltaic Materials, Henan Normal University, Xinxiang, Henan, 453007, PR China
b
h i g h l i g h t s A bridge-linking mechanism was proposed to illustrate the formation of MD-THS. The wide pore size distribution facilitates the anchoring process of quantum dots. The oriented aligned primary nanocrystals inhibit undesired charge recombination. Cell-MD-THS demonstrates much higher PCE of 4.15% than Cell-Nanocrystal (3.03%).
a r t i c l e i n f o
a b s t r a c t
Article history: Received 8 October 2014 Received in revised form 8 January 2015 Accepted 10 February 2015 Available online 11 February 2015
Multi-dimensional TiO2 hierarchal structures (MD-THS) assembled by mesoporous nanoribbons consisted of oriented aligned nanocrystals are prepared via thermal decomposing Ti-contained gelatin-like precursor. A unique bridge linking mechanism is proposed to illustrate the formation process of the precursor. Moreover, the as-prepared MD-THS possesses high surface area of ~106 cm2 g1, broad pore size distribution from several nanometers to ~100 nm and oriented assembled primary nanocrystals, which gives rise to high CdS/CdSe quantum dots loading amount and inhibits the carries recombination in the photoanode. Thanks to these structural advantages, the cell derived from MD-THS demonstrates a power conversion efficiency (PCE) of 4.15%, representing ~36% improvement compared with that of the nanocrystal based cell, which permits the promising application of MD-THS as photoanode material in quantum-dot sensitized solar cells. © 2015 Elsevier B.V. All rights reserved.
Keywords: Titanium dioxide Hierarchical structure Growth mechanism Quantum-dot sensitized solar cells
1. Introduction Quantum-dot sensitized solar cell (QDSSC) represents a group of new solar energy conversion devices. Although it shares many design configuration and working principle with dye-sensitized solar cell (DSSC), the Quantum dots (QDs) sensitizers, generally the metal chalcogenides, have many inherit advantages compared with the costly Ru based dyes, such as the flexible solar spectrum matching capability, high extinction ratio and the unique multiple excitation generation (MEG) resulting from the impact ionization effect [1e4]. Because of the MEG effect, a high theoretical PCE of
* Corresponding author. School of Chemistry and Chemical Engineering, Henan Normal University, Xinxiang, Henan, 453007, PR China. E-mail address: [email protected] (K. Jiang). http://dx.doi.org/10.1016/j.jpowsour.2015.02.062 0378-7753/© 2015 Elsevier B.V. All rights reserved.
~44% could be achieved, which is much higher than the traditional Shockley and Queisser limit of ~31% [5]. CdS/CdSe co-sensitized strategy is commonly used to achieve high PCE in QDSSC [6e9]. In addition, other sensitizers such as CdTe, PbS and PbSe were also developed to enhance the PCE by broadening the absorption spectrum or improving the injection efficiency of the photoelectrons [10e12]. Most recently, QDSSCs based on low toxicity sensitizers (Ag2S, Sb2S3 or CuInS2), solid-state or flexible configurations were also designed and prepared [13e16]. However, the highest PCE of these devices merely reaches to ~7%, which still lags far behind compared with DSSC (~13%) [17e20]. The low PCE generally results from the massive recombination in the unsophisticated cell configurations, indicating the photoanode, electrolyte and the counter electrode are not satisfactorily optimized. [21e24]. Photoanode is one of the key components in QDSSCs, which
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
serves as both the scaffold to anchor QDs and diffusing matrix for the injected photoelectrons, many groups have made various efforts to optimize the TiO2 structures so as to enhance the light harvesting capability and electron transport performance [25e29]. Several design rules should be considered as priorities. First of all, the QDs have much greater size than the monolayer adsorbed dye molecules, thus a photoanode film with large pore size distribution is favorable for the QDs loading. Especially in the pre-synthesis strategies, in which the QDs are prepared before loaded on the photoanode, the small pore size inevitably jeopardizes the QDs penetration and results in QDs aggregation inside the film structures [30,31]. Secondly, the high surface area could increase the initial nucleation population of the QDs on the photoanode structures, which can improve the QDs coverage and retard undesired interface recombination [27,32e34]. Particularly, for the in situ growth strategies, in which the QDs were loaded on the photoanode film by means of successive ionic layer adsorption and reaction (SILAR) or chemical bath deposition (CBD), photoanode film with high surface area is of same significance as the pore size distribution. Thirdly, although the continuous QDs coating layer may serve as additional electron transport path, the majority of the generated photoelectrons are collected through the TiO2 matrix. Therefore, just like the DSSC cases, hierarchical structures consisted of one-dimensional subunits or orderly aligned primary units can greatly improve the electron collection efficiency by minimizing the population of the grain boundaries [35]. However, the previously reported hierarchical structures can hardly balance the required structural qualities of high surface area, broad pore size distribution and oriented aligned subunits. Especially for the TiO2 hierarchical structures with small pore size distribution, the QDs aggregation inevitably takes place in the pore channels, which results in increased charge recombination and leads to PCE plummet. Therefore, the TiO2 hierarchical structures should be further optimized for the particular demands in QDSSCs. Herein, we designed and prepared a unique MD-THS which is
203
assembled by mesoporous nanoribbons consisted of oriented aligned nanocrystals. Time dependent trails reveal an interesting growth and assembling process of the hierarchal gelatin precursor and the formation is driven by the energy minimization on the coordinating modes of carboxyl groups. In addition, when used as photoanode material in CdS/CdSe co-sensitized cells, the MD-THS could integrate the advantages of different dimensional materials: the three dimensional (3D) hierarchical structure provides a suitable outer diameter of ~1 mm and a wide pore size distribution from several nanometers to ~100 nm, which improves the harvesting of incident light. The one dimensional (1D) nanoribbon with oriented assembled primary nanocrystals reduces the grain boundaries and charge recombination. Moreover, the zero dimensional (0D) nanocrystals with diameter of ~15 nm leads to high surface area of ~106 cm2 g1, which favorites the loading process of QDs sensitizers. Therefore, the cell derived from MD-THS demonstrates a high PCE of 4.15%, which indicates ~36% improvement compared with nanocrystal based cell (3.03%). 2. Experimental section 2.1. Materials synthesis The MD-THSs were prepared in a solvothermal method. 1 mL of titanium n-butoxide (TBT) was added drop wise to 20 mL of acetic acid with rigorous stirring. Afterwards, the mixture was transferred into 25 mL Teflon-lined stainless steel reactor and heated at 200 C for 2 h. After cooling, the white product were collected by centrifugation and washed with distilled water and ethanol several times, then dried at 60 C for 3 h. Finally, the samples were heated to 450 C at a rate of 5 C min1 and calcinated for 3 h to obtain the final product. The nanocrystals were prepared based on M. Gratzel et al.'s work [36]. Briefly, 7.5 mL titanium isopropoxide was added into 45 mL 0.1 M HNO3 aqueous solution. The suspension was placed in
Fig. 1. FESEM images of (a) hierarchical gelatin precursor obtained by solvothermal treatment; (b) MD-THSs after calcination; (c) the magnified image located at the black dot square in (b); (d) and (e) are the corresponding TEM and HRTEM images of the mesoporous ribbons.
204
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
80 C water bath and stirred for 8 h. The concentrated suspension was then transferred into 50 mL autoclave and heated at 230 C for 12 h. The final product was collected by centrifugation and stocked in wet for later use. The as-prepared nanoparticles have average diameter of ~15 nm and BET surface area of ~110 m2 g1.
2.2. Fabrication of the QDSSCs The cleaned fluorine-doped tin oxide (FTO 15 U square1, NSG, Japan) glass was immersed in TiCl4 aqueous solution (0.15 M) in a beaker and then kept in an oil bath at 70 C for 1 h to coat a thin layer of TiO2 on the surface for the sake of forming a barrier layer. The treated FTO was calcinated at 450 C for 20 min. The paste was prepared according to our previous literature [37]. The prepared paste was coated onto the FTO layer using a traditional doctor-blade method with adhesive tape to be rough in the middle area as frame and spacer. Then, the coated FTO glass flake was calcinated at 500 C for 30 min to remove organic reagent. A modified chemical bath deposition method [25] was applied to load QDs on TiO2 photoanodes. An aqueous solution, with the composition of 0.02 mM CdCl2, 0.14 mM thiourea, 0.07 mM NH4Cl and 0.23 mM ammonia with final pH of ~9, was used to deposite CdS QDs. The reaction was carried out in 10 C water bath for 1 h. After rinsed with water, the TiO2/CdS films were immersed in an aqueous solution containing 0.026 mM CdSO4, 0.04 mM N(CH2COONa)3 and 0.026 mM Na2SeO3 for 4 h in the same way. Both the QDs (CdS and CdSe) depositions films were fully washed with water. For the ZnS passivation, the sensitized films were deposited by twice dipping alternatively into 0.1 M Zn(CH3COO)2 and 0.1 M Na2S solutions for 1 min per dip, rinsed with distilled water between dips. Cu2S in situ deposited on brass was used as a counter electrode. The QDs sensitized photoanode and counter electrode were sandwiched to form the cell. The electrolyte composed of 1 M Na2S and 1 M S in deionized water was injected into the cell via capillarity. A metal mask with a window of ~0.2 cm2 was covered on the back side of photoanode to control the active area.
3. Results and discussion Fig. 1a shows FESEM images of the as-prepared gelatin precursor which is assembled from 1D nanoribbons and has an average diameter of ~1.5 mm. As depicted in Fig. 1b, the average diameter of the peony-like structures decrease to ~1 mm after calcination, indicating a ~30% shrinkage due to the decomposition of the gelatin precursor [38]. The corresponding TGA measurement shows the product experiences ~40% weight loss during the calcination (Fig. S1). Moreover, the black box area was magnified and displayed in Fig. 1c, which shows the ribbons are 10e20 nm in thickness and assembly by numerous primary nanocrystals. The TEM image in Fig. 1d indicates that the nanocrystals are ~15 nm in diameter and closely linked together. Moreover, the HRTEM image in Fig. 1d depicts the lattice fringes of two adjacent nanoparticles, which indicates the primary nanocrystals are linked with same orientation. The spacing lattices are well-defined and measured to be 0.35 nm, suggesting they are highly crystallized and with anatase structure. Fig. 2 a shows the XRD pattern of the precursor and MD-THSs. The precursor exhibits unknown diffraction peaks, which is in accord with the previously reported results [38]. After calcination, the diffraction peaks of MD-THSs could be well indexed to the tetragonal anatase phase (JCPDS cards No.01-0562), suggesting the precursor with unknown crystal phase is completely converted into anatase TiO2. The N2 adsorptionedesorption isotherms and the pore size distribution of the MD-THSs are shown in Fig. 2b. The
2.3. Material characterization The samples were studied by field-emission scanning electron microscope (FESEM, HITACHI, S4800), transmission electron microscopy (TEM, FEI, Tecnai F30), X-ray powder diffraction (XRD, Rigaku D/max-2500 diffractometer with Cu Ka radiation, l ¼ 0.1542 nm, 40 kV, 100 mA) and BrunauereEmmetteTeller (BET, Micrometrics ASAP 2010). The optical diffusion reflection spectra and absorption spectra were measured using a PerkineElmer Lambda 35 UVeVis spectrometer. The thickness of films was measured on a profilometer (DEKTAK 150, Veeco Instruments Inc.). Moreover, X-ray energy dispersive spectroscopy (EDX, Oxford, INCA MICSFþ) was also used to perform the element mapping of photoanodes loaded with QDs. Fourier transform infrared spectroscopy was characterized by (FTIR, FTS-40).
2.4. Photovoltage measurements Photocurrentevoltage (IeV) test were measured using a Keithley 2400 sourcemeter, and the output power density of the xenon lamp solar simulator (Newport) was adjusted to 100 mW cm2 (AM1.5). Incident photon to current conversion efficiency (IPCE) was also measured on the 7-scspec system (Saifan Beijing) with a 1/ 4 m monochromator.
Fig. 2. (a) XRD pattern of the precursor sample and the MD-THS (b) Nitrogen adsorptionedesorption isotherms of the MD-THS, the inset in (b) is the corresponding pore size distribution curve.
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
205
Fig. 3. SEM images of the products harvested at different time intervals through solvothermal reaction: (a) 0 min, (b) 30 min, (c) 1 h, (d) 2 h, (e) 3 h and (f) 6 h.
isotherm exhibits typical type IV curve with H1-type hysteresis at high relative pressures, which is usually ascribed to the predominance of mesopores and the existence of pore connectivity. In addition, a broad pore size is observed in the distribution curve (the inset of Fig. 2b), which ranges from several nanometers to ~100 nm. The broad pore size distribution may favorite both the loading of the quantum dots and the electrolyte diffusion in the film structure. Moreover, the BET measurement indicates that the sample possesses a high surface area of ~106 m2 g1. In order to reveal the formation mechanism of MD-THS, the solvothermal process was monitored by time dependent experiments. When the reaction temperature reaches to 200 C, the reactor was immediately fetched out and cooled in water and the sample was denoted as 0 min. At this initial stage (Fig. 3a), the firstly formed gelatin is soluble both in water and alcohol and no particular morphology was observed. Although the corresponding XRD pattern (Fig. 4a) cannot match with any of the existed crystal structure data, it is obviously different from the next generated gelatin structures. After 30 min, 1D ribbons with interlaced structure are formed in the sample (Fig. 3b). When the reaction proceeds to 1 h, the ribbons are inclined to assemble into 3D structures. At this stage, individual ribbons could be founded in the sample, suggesting the assembling process is not finished (Fig. 3c). After 2 h, the ribbons are fully assembled into 3D spherical hierarchical structures. During this process, the XRD patterns of the gelatins were not changed, indicating that the gelatin simply experiences an
assembling process without structural variation (Fig. 4a). As it was demonstrated previously, the gelatin-like intermediate products could act as a reservoir to gradually release soluble titaniumcontaining species for the nucleation and growth of TiO2 spindles. As displayed in Fig. 3d, TiO2 spindles are found coexisted with the gelatin structures after 3 h reaction, which could be confirmed by XRD investigation. When the time reaches to 6 h, the gelatin precursor is fully converted into TiO2 spindles (Figs. 3f and 4a). Moreover, the growth process was also monitored by the FTIR spectra, as displayed in Fig. 4b. The peaks centered at 1719, 1525, 1463, 1417 and 1256 cm1 can be ascribed to the different vibration modes of the carboxylic groups coordinated to Ti [39]. The two distinct vibration signals centered at around 1719 and 1256 cm1 indicate the existence of monodentate coordination. The peaks at the frequencies of 1463 and 1417 cm1 demonstrate approximate frequency separations of Dn ¼ 62 and Dn ¼ 108 cm1 between the signals around 1525 cm1, which respectively represents the bidentate and the bridging modes of the acetates coordinated to the titanium atoms [37,40]. In addition, the peak intensities experience gradual receding as the reaction prolonged, implying the monodentate and bidentate ligands were removed from the products. As the reaction time increases to 6 h, TieOeTi rigid frameworks with high stability are formed, which could be indicated from the broad and strong absorption around 600 cm1 assigning to the TieOeTi moiety of anatase TiO2. Base on the control experiments, a possible growth mechanism
206
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
of MD-THS was proposed and illustrated in Figs. 5 and 6. The formation process of the Ti-contained gelatin is shown in Fig. 5a. At the very beginning, the coordinating carboxyl groups would bond to the titanium atom along with the ester groups to generate a hexa-coordinated titanium central atom, which favorites the formation of anatase TiO2 (Step 1). During this process, except the chelating coordination, monodentate coordination mode may also form. As the reaction proceeded, the adjacent Ti atoms could be linked together by converting the acetate ligand from chelating coordination to bridging mode (Step 2). The ester groups of the asformed two-atom system would be then substituted by the abundant acetate groups (Step 3). Subsequently, the bridging effect of the acetate could further link the Ti atoms into a four-atom system (Step 4). Thereafter, the repeating linkage forms the gelatin structure and induces the later assembling process (Step 5). Moreover, the acetate molecules with chelating and bridging structural modes are compared after the energy minimization (Fig. 5b). It is found that in the chelating mode, the average bond length of CeO and OeTi bonds are 1.486 Å and 1.984 Å. Meanwhile, in the bridging mode, the values are 1.433 Å and 1.975 Å. The decreased bond length indicates the bridging coordination is more energy favorable
Fig. 4. (a) XRD and (b) FTIR investigations of the intermediate samples collected at 0 min, 30 min, 1 h, 3 h, and 6 h through solvothermal reaction.
during the formation of the hierarchical gelatin. The assembling process of the MD-THS is illustrated in Fig. 6. At the initial stage of the solvothermal reaction, only gelatin with small size and low molecular weight is formed, which makes them easily dissolved in water and ethanol. The high reaction temperature causes high rate collision of the gelatin particles, which leads to the linkage of the gelatin by the bridging coordination of acetate. Therefore, 1D ribbon-like structures are formed afterwards (Step 1). Moreover, the cross-link of ribbons could further reduce the system free energy by forming interlaced ribbons and the intermediate hierarchical structures (Step 2). Considering the sphere-like pattern could greatly reduce the surface tension at the solid and liquid interface, 3D spherical gelatin structures with radially aligned 1D ribbons are finally formed (Step 3). After calcination, the removal of organic composites finally generates the peony-like MD-THS assembled by interlaced mesoporous ribbons. Considering the unique structural qualities of the as-prepared MD-THSs, they were adopted as the photoanode material for QDSSC. In order to optimize photoanode structure, the deposition times and the film thickness of the mesoporous film were controlled. Fig. S2 and Table S1 show the current densityevoltage (IeV) characteristics of the cells prepared with different depositing times. The absorption intensity of the photoanode continuously improves with more reaction time (Fig. S3), indicating the loading amount of the sensitizers increases. However, the overall PCE starts to drop after 4 h, which probably results from that the increased thickness of the quantum-dots layer reduces the electron injection efficiency and brings forth more recombination. Moreover, the thickness of photoanode film was also varied to investigate its influence over the PCE. The film thicknesses were respectively controlled at ~8, ~12, ~15, ~17 and ~20 mm. In addition, a semitransparent film consisted of ~15 mm nanocrystals was also prepared as reference. The IeV curves of the as-prepared cells with different film thicknesses are presented in Fig. 7 and the corresponding photovoltaic characteristics are summarized in Table 1. For the MD-THS photoanodes with different film thicknesses, the Jsc and Voc improve from 8.20 to 14.05 mA cm2 and from 526 to 571 mV, as the thickness increases from ~8 to ~15 mm. It is reasonable to anticipate that the thicker film leads to greater quantum dots loading amount, which generates higher photoelectron concentration in the TiO2 conduction band and then leads to the Jsc augment. In addition, the increased free electron population could also elevate the quasi-Fermi level and result in higher Voc. However, the Jsc and Voc decrease, if the film thickness further increases to ~17 and ~20 mm. Different from the fluctuation of Jsc and Voc, the filling factor (FF) of the cells experiences monotonically decrease as increasing the film thickness. The FF largely depends on the series resistance of the film, and thicker film with higher inner resistance would inevitably decrease the FF [41]. Therefore, the PCE of the Cells with film thicknesses of ~17 and ~20 mm respectively drop to 3.77% and 2.59%. The best performance is achieved by CellMD-THS with film thickness of ~15 mm, which demonstrates Jsc of 14.05 mA cm2, Voc of 571 mV, FF of 51.7%, and PCE of 4.15%. Moreover, the Cell-Nanocrystal with photoanode film thickness of ~15 mm was also investigated, which leads to a PCE of 3.03%. Compared with Cell-Nanocrystal, it is obvious that the high PCE of Cell-MD-THS is attributed to the synergetic improvement on Jsc, Voc and FF. The cross-sectional SEM image for the QDs sensitized MD-HTS photoanode is shown in Fig. 8a. The MD-THS is evenly deposited on the FTO substrate with thickness of ~15 mm. Moreover, the uniform coverage of QDs on TiO2 throughout the film thickness is confirmed by elemental mapping via EDX analysis. As shown in Fig. 8bed, the elemental mapping on the rectangular area demonstrates uniform distribution of Ti, Cd, and Se atoms throughout
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
207
Fig. 5. (a) Possible formation mechanism of the gelatin: continuous bridge-linking effect of acetate ligand among the Ti atoms; (b) the structural comparison of acetate molecules with chelating and bridging coordination modes.
Fig. 6. Schematic illustration of the growth mechanism of MD-THS.
Fig. 7. IeV curves of the CdS/CdSe/ZnS QDSSCs made from hierarchical TiO2 with different thicknesses and the as-prepared nanoparticles TiO2 film (under one sun illumination (AM 1.5, 100 mw/cm2).
208
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
Table 1 Photovoltaic characteristics of the as-prepared QDSSCs (the active area is ~0.2 cm2). Cells
MD-THSs
Thickness (mm) Voc (mV) Jsc (mA cm2) FF (%) PCE (%)
~8 526 8.20 60.7 2.62
~12 548 11.33 53.0 3.29
Nanoparticles ~15 571 14.05 51.7 4.15
~17 567 12.91 51.5 3.77
~20 509 11.98 42.5 2.59
~15 550 10.77 51.2 3.03
the film depth. Atomic percentages from EDX elemental analysis on the film cross section were found to be 22.08, 3.83 and 3.06% for Ti, Cd, and Se respectively. Moreover, a similar test was performed in nanoparticles based film, the atomic percentages were 24.80, 3.89 and 2.86% for Ti, Cd, and Se. (Table S2) Based on the EDX test, the Cd/Ti ratio in the MD-THS film could be determined as ~0.173, which is greater than that in the nanocrystal film of ~0.157 (Fig. S4). The uniform distribution of the QDs as well as the high Cd/Ti ratio in MD-THS film offer solid support to the viewpoint that although MD-THS has similar surface area with nanocrystals, the wide pore size distribution could facilitate the QDs penetration throughout the film depth, which leads to higher QDs loading amount [42]. Fig. 9a plots the optical absorption spectra of the nanoparticle and HM-THS photoanode films after CdS/CdSe sensitization (~15 mm). Film-MD-THS has a higher absorbance in the whole
region, indicating the higher QDs loading amount and light harvest efficiency. In addition, as shown in Fig. 9b, the optical bandgap for the CdSe QDs deposited in difference film structures could be evaluated using the equation for the near-edge absorption and the extrapolation of the linear region of the plot of (Ahn)2 against photon energy (hn) [39,43]. The direct forbidden bandgap of ~1.95 eV and ~1.99 eV could be determined and the redshift can be attributed to the quantum-size effect, suggesting larger CdSe QDs are formed inside the MD-THS film structure. Moreover, as displayed in Fig. 9c, Film-MD-THS exhibits a prominent scattering effect at the wavelength of 400e800 nm, which could further improve the harvesting of the visible and near-infrared light. It can be concluded that the large pore size distribution of MD-THS not only increases the QDs loading amount but also leads to the formation of QDs with greater diameters, which improves the absorption of incident light and stretches the absorption spectrum to longer wave range. Moreover, the micrometer sized hierarchical structure could induce multiple-scattering of the incident light to further improve the light harvesting, which finally leads to the high Jsc. Therefore, as depicted in Fig. 9d, Cell-MD-THS demonstrates a higher IPCE value compared with Cell-Nanoparticles. Moreover, as it is shown in Eqn. (1) [44]
V∞
RT AI h i ln ¼ bF n0 k0 Sn þ n0 Kr Dþ
!
(1)
Fig. 8. (a) Cross-sectional SEM image of ~15 mm thick QDs sensitized MD-HTS film; (bed) Elemental mapping of titanium (b), cadmium (c), and selenium (d) by EDX spectroscopy for the rectangular area indicated by the black box in panel a showing the uniform distribution of the QDs sensitizer throughout the film thickness.
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
209
Fig. 9. (a) The absorption spectra of nanoparticles and HM-THS films (~15 mm in thickness) with CdS/CdSe sensitization; (b) the corresponding (Ahn)2 vs hn spectra; (c) diffusereflectance spectra of photoanode film without sensitization; (d) the IPCE spectra of the corresponding cells.
where R is the molar gas constant, T the temperature, b the reaction order of S n and electrons, F the Faraday constant, A the electrode surface area for light absorption, I the incident photon flux, n0 the accessible electronic state concentration in the conduction band, kb the backward reaction kinetic constant of the injected electrons with polysulfide electrolyte and kr the recombination kinetic constant of the electrons with oxidized QDs (Dþ), the Voc of the device is largely determined by the recombination of the photoelectron with the oxides species in the electrolytes. In order to investigate the charge recombination process, the Voc decay test was carried out and displayed in Fig. 10. At the open circle conditions, as the illumination is removed, the photogenerated electrons and holes would recombine via the trap states in the photoanode, which leads to the exponential Voc decay [45,46]. The dots shown in Fig. 10a represent the measured Voc data points after the illumination is removed. After fitted with exponential decay functions, the solid lines and the inset normalized decay curves prove that Cell-Nanocrystal exhibits an evident faster Voc decay, indicating a more rapid recombination process. Moreover, Based on Eqn. (2) [47,48]
k T tn ¼ B e
dVoc dt
1 (2)
the electron lifetime (tn) in function of Voc could be obtained by differentiating the decay curves and then taking an inversing, which are displayed in Fig. 10b. It is obvious that Cell-MD-THS has greater tn than Cell-Nanocrystal in the whole voltage range, indicating the recombination kinetic constants of kb and kr are reduced
in Cell-MD-THS. Furthermore, based on Eqn. (1) the higher A of Cell-MD-THS could improve the light harvesting and the free electron population in the conduction band of TiO2, which finally results in the elevation of quasi-Fermi level and leads to the higher Voc.
4. Conclusions Multi-dimensional TiO2 hierarchal structures composited of 1D nanoribbions and 0D primary nanocrystals were deliberately designed and prepared for the QDSSC application. It was found the continuous bridging effect of the acetate plays a crucial role in forming the MD-THS. When used as photoanode material in CdS/ CdSe co-sensitized solar cells, a high PCE of 4.15% was demonstrated, indicating ~36% improvement compared with nanocrystal based cell. The structural advantages of MD-THS could be summarized as: (a) the high surface area and wide pore size distribution could increase the QDs loading amount and broaden the absorption spectrum, which enhances the light absorption of the photoanode. (b) The good scattering effect could further improve the harvesting of visible and near-infrared light, which leads to the high Jsc. In addition, (c) the oriented assembled primary nanocrystals provide direct diffusion path for electron collection, which reduces the recombination opportunities of photogenerated carriers and prolongs the electron lifetime, so as to enhance the Voc. Therefore, the unique MD-THS could well integrate the desirable structural qualities of an ideal photoanode material, which enables it a promising candidate for the high performance QDSSCs with different sensitizers.
210
D. Wu et al. / Journal of Power Sources 282 (2015) 202e210
Fig. 10. The Voc decay tests of Cell-Nanocrystal and Cell-MD-THS: (a) the Voc decay curves after cut off the illumination, the scattered dots are the experimental data points and the solid lines represent the fitted curves (the inset is the normalized Voc decay curves); (b) electron recombination lifetimes in function of the Voc.
Acknowledgments This work is supported by NSFC (Grant Nos. 61204078, 61176004 and U1304505), Program for Changjiang Scholars and Innovative Research Team in University (PCSIRT). Program for Innovative Research Team (in Science and Technology) in University of Henan Province (No. 13IRTSTHN026), Basic and Frontier Research Programs of Henan Province(No. 142300410420) and the Key Project of Science and Technology of Henan Province (No.13A150517 and 14A150002). Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2015.02.062. References [1] J.J. Tian, R. Gao, Q.F. Zhang, S.G. Zhang, Y.W. Li, J.L. Lan, X.H. Qu, G.Z. Cao, J. Phys. Chem. C 116 (2012) 18655e18662. [2] X.Y. Yu, B.X. Lei, D.B. Kuang, C.Y. Su, J. Mater. Chem. 22 (2012) 12058e12063. [3] X.H. Song, M.Q. Wang, T.Y. Xing, J.P. Deng, J.J. Ding, Z. Yang, X.Y. Zhang,
J. Power Sources 253 (2014) 17e26. [4] X.M. Huang, S.Q. Huang, Q.X. Zhang, X.Z. Guo, D.M. Li, Y.H. Luo, Q. Shen, T. Toyoda, Q.B. Meng, Chem. Commun. 47 (2011) 2664e2666. [5] L.B. Yu, Z. Li, Y.B. Liu, F. Cheng, S.Q. Sun, J. Power Sources 270 (2014) 42e52. [6] R. Zhou, Q. Zhang, E. Uchaker, J. Lan, M. Yin, G. Cao, J. Mater. Chem. A 2 (2014) 2517e2525. [7] J.L. Xu, K. Li, W.Y. Shi, R.J. Li, T.Y. Peng, J. Power Sources 260 (2014) 233e242. [8] J. Tian, E. Uchaker, Q. Zhang, G.Z. Cao, ACS Appl. Mater. Interfaces 6 (2014) 4466e4472. [9] Z. Ali, I. Shakira, D.J. Kang, J. Mater. Chem. A 2 (2014) 6474e6479. [10] C.J. Raj, S.N. Karthick, S. Park, K.V. Hemalatha, S.K. Kim, K. Prabakar, H.J. Kim, J. Power Sources 248 (2014) 439e446. [11] H.P. Shen, X.J. Jiao, D. Oron, J.B. Li, H. Lin, J. Power Sources 240 (2013) 8e13. [12] J. Wang, I.M. Sero, Z.X. Pan, K. Zhao, H. Zhang, Y.Y. Feng, G. Yang, X.H. Zhong, J. Bisquert, J. Am. Chem. Soc. 135 (2013) 15913e15922. [13] T.P. Brennan, O. Trejo, K.E. Roelofs, J. Xu, F.B. Prinzbc, S.F. Bent, J. Mater. Chem. A 1 (2013) 7566e7571. [14] Z.X. Pan, I.M. Sero, Q. Shen, H. Zhang, Y. Li, K. Zhao, J. Wang, X.H. Zhong, J. Bisquert, J. Am. Chem. Soc. 136 (2014) 9203e9210. [15] T.T. Ngo, S. Chavhan, I. Kosta, O. Miguel, H. Grande, R. Tena-Zaera, Appl. Mater. Interfaces 6 (2014) 2836e2841. [16] M.L. Que, W.X. Guo, X.J. Zhang, X.Y. Li, Q.L. Hua, L. Dong, C.F. Pan, J. Mater. Chem. A 2 (2014) 13661e13666. [17] J.H. Im, C.R. Lee, J.W. Lee, S.W. Park, N.G. Park, Nanoscale 3 (2011) 4088e4093. [18] P.K. Santra, P.V. Kamat, J. Am. Chem. Soc. 134 (2012) 2508e2511. [19] S.H. Im, C.S. Lim, J.A. Chang, Y.H. Lee, N. Maiti, H.J. Kim, M. Nazeeruddin, K.M. Gratzel, S.S. Toward, Nano. Lett. 11 (2011) 4789e4793. [20] A. Yella, H.W. Lee, H.N. Tsao, C.Y. Yi, A.K. Chandiran, M.K. Nazeeruddin, E.W.G. Diau, C.Y. Yeh, S.M. Zakeeruddin, M. Gr€ atzel, Science 334 (2011) 629e633. [21] X. Song, M. Wang, J. Deng, Y. Ju, T. Xing, J. Ding, Z. Yang, J. Shao, J. Power Sources 269 (2014) 661e670. [22] F. Wang, H. Dong, J. Pan, J. Li, Q. Li, D.S. Xu, J. Phys. Chem. C 118 (2014) 19590, 1958919598. [23] H. Kim, S. Kim, C. Gopi, S. Kim, S. Rao, M. Jeong, J. Power Sources 268 (2014) 163e170. [24] I. Mora-Sero, S. Gimenez, F. Fabregat-Santiago, R. Gomez, Q. Shen, T. Toyoda, J. Bisquert, Acc. Chem. Res. 42 (2009) 1848e1857. [25] Q.X. Zhang, X.Z. Guo, X.M. Huang, S.Q. Huang, D.M. Li, Y.H. Luo, Q. Shen, ́ T. Toyoda, Q.B. Meng, Phys. Chem. Chem. Phys. 13 (2011) 4659e4667. [26] Y.K. Lai, Z.Q. Lin, D.J. Zheng, L.F. Chi, R.G. Du, C.J. Lin, Electrochim. Acta 79 (2012) 175e181. [27] M. Samadpour, S. Gimenez, A.I. Zad, N. Taghavinia, I.M. Sero, Phys. Chem. ́ Chem. Phys. 14 (2012) 522e528. [28] E.H. Kong, Y.J. Chang, Y.C. Park, Y.H. Yoon, H.J. Park, H.M. Jang, Phys. Chem. Chem. Phys. 14 (2012) 4620e4625. [29] H. Dong, D.P. Wu, F. Zhu, F.F. Wang, S.F. Zhu, Q. Li, D.S. Xu, Sol. Energy Mat. Sol. C 133 (2015) 201e207. [30] A. Salant, M. Shalom, Z. Tachan, S. Buhbut, A. Zaban, U. Banin, Nano. Lett. 12 (2012) 2095e2100. [31] A. Salant, M. Shalom, I. Hod, A. Faust, A. Zaban, U. Banin, J. Am. Chem. Soc. 4 (2010) 5962e5968. [32] O. Niitsoo, S.K. Sarkar, C. Pejoux, S. Ruhle, D. Cahen, G. Hodes, J. Photochem. Photobiol. A Chem. 181 (2006) 306e313. [33] H.J. Lee, J. Bang, J. Park, S. Kim, S.M. Park, Chem. Mater. 22 (2010) 5636e5643. [34] X. Song, M. Wang, Y. Shi, J. Deng, Z. Yang, X. Yao, Electrochim. Acta 81 (2012) 260e267. [35] T. Toyoda, Q. Shen, J. Phys. Chem. Lett. 3 (2012) 1885e1893. [36] C.J. Barbe, F. Arendse, P. Comte, M. Jirousek, F. Lenzmann, V. Shklover, M. Gratzel, J. Am. Ceram. Soc. 80 (1997) 3157e3171. [37] D.P. Wu, F. Zhu, J.M. Li, H. Dong, Q. Li, K. Jiang, D.S. Xu, J. Mater. Chem. 22 (2012) 11665e11671. [38] J.F. Ye, W. Liu, J.G. Cai, S. Chen, X.W. Zhao, H.H. Zhou, L.M. Qi, J. Am. Chem. Soc. 133 (2011) 933e940. [39] D. Jiang, Y. Xu, B. Hou, D. Wu, Y.H. Sun, J. Inorg. Chem. 8 (2008) 1236e1240. [40] L. Li, C.Y. Liu, CrystEngComm 12 (2010) 2073e2078. [41] G. Schlichthorl, N.G. Park, A.J. Frank, J. Phys. Chem. B 103 (1999) 782e791. [42] D.P. Wu, X.J. Shi, H. Dong, F. Zhu, K. Jiang, D.S. Xu, X.C. Ai, J.P. Zhang, J. Mater. Chem. A 2 (2014) 16276e16284. [43] A. Hagfeldt, M. Graetzel, Chem. Rev. 95 (1995) 49e68. [44] F. Zhu, H. Dong, Y. Wang, D.P. Wu, J.M. Li, J.L. Pan, Q. Li, X.C. Ai, J.P. Zhang, D.S. Xu, Phys. Chem. Chem. Phys. 15 (2013) 17798e17803. [45] D.P. Wu, Y. Wang, H. Dong, F. Zhu, K. Jiang, L.M. Fu, J.P. Zhang, D.S. Xu, Nanoscale 5 (2013) 324e330. [46] X. Song, M. Wang, H. Zhang, J. Deng, Z. Yang, C. Ran, X. Yao, Electrochim. Acta 108 (2013) 449e457. [47] A. Zaban, M. Greenshtein, J. Bisquert, ChemPhysChem 4 (2003) 859e864. [48] J. Bisquert, A. Zaban, M. Greenshtein, I. Mora-Sero, J. Am. Chem. Soc. 126 (2004), 13550e13359.