Towards implementing hierarchical porous zeolitic imidazolate frameworks in dye-sensitized solar cells

A one-pot method for encapsulation of dye, which can be applied for dye-sensitized solar cells (DSSCs), and synthesis of hierarchical porous zeolitic imidazolate frameworks (ZIF-8), is reported. The size of the encapsulated dye tunes the mesoporosity and surface area of ZIF-8. The mesopore size, Langmuir surface area and pore volume are 15 nm, 960–1500 m2 · g−1 and 0.36–0.61 cm3 · g−1, respectively. After encapsulation into ZIF-8, the dyes show longer emission lifetimes (greater than 4–8-fold) as compared to the corresponding non-encapsulated dyes, due to suppression of aggregation, and torsional motions.


Introduction
Solar energy is a promising alternative energy source to replace traditional fossil fuels [1][2][3]. Dye-sensitized solar cells (DSSCs, or Grätzel cells) represent a third generation solar cell technology, characterized by the potential for low production costs, green technologies and high conversion efficiency [1][2][3][4]. In DSSCs, the dye chromophore has the central task of absorbing light to generate electrons that are transferred to the sensitized semiconductor substrate typically on an fs-ns time scale or alternatively are lost through dissipation via radiative relaxation or internal conversion (IC) [5][6][7][8][9]. The structure-property & 2019 The Authors. Published by the Royal Society under the terms of the Creative relationship [10,11] for a sensitizing dye depends on a series of properties, such as anchoring group [12] and fluorescence lifetime [13]. The aggregation and torsional motions of the dyes decrease the efficiency of the measured cell. Encapsulating the dyes into porous materials may reduce both aggregation and torsional motions of the dye.
Herein, a method for encapsulation of dyes into ZIF-8 crystals is reported for application in DSSCs. The materials were synthesized using a one-pot method, and resulted in hierarchical porous structures of ZIF-8 [39]. The materials were characterized using X-ray diffraction (XRD), transmission electron microscopy (TEM), high angle annular dark field (HAADF), electron energy loss spectroscopy (EELS), scanning electron microscopy (SEM), N 2 adsorption -desorption isotherms, fluorescence spectroscopy and emission lifetime measurements. The lifetime measurements reveal an increase on the dye's excited state upon encapsulation due to suppression of both aggregation and torsional motions. The materials were also tested for DSSCs.
ZIF-8 and dye@ZIF-8 were synthesized in a scintillation vial using triethylamine-assisted method with modification [39]. The dye is insoluble in water; thus, it was dissolved in methanol. Triethylamine (TEA, 0.7 mmol (0.1 ml), or 7 mmol (1 ml)) was added to a solution of Zn(NO 3 ) 2 . 6H 2 O (0.7 mmol, 0.8 ml). A solution of the dye (L1, LIFc or L1Fc 2 , 1 ml) was added, followed by the addition of 8 ml of the Hmim solution. The reaction volume was completed to 17 ml using deionized water. The reaction mixture was stirred for 30 min. The particles were collected using centrifugation (13 500 r.p.m., 30 min). The materials were washed using water and ethanol (2 Â 25 ml). The yields of the prepared materials were 80 -98% (electronic supplementary material, table S1).
Preparation of dye@ZIF-8-coated TiO 2 electrodes, instruments used for characterization, electrochemical impedance spectroscopy, evaluation of DSSCs parameters [40] and time correlated single photon counting (TCSPC) experiments are described in the electronic supplementary material.

Results and discussion
The dyes used in the solar cells (L1, L1Fc, L1Fc 2 , figure 1a) were encapsulated into ZIF-8 using a one-pot method [39]. The presence of methanol in the reaction mixture increases the growth rate of the ZIF-8 royalsocietypublishing.org/journal/rsos R. Soc. open sci. 6: 190723 crystals as compared to that using only water [39,41]. XRD patterns confirm the formation of pure ZIF-8 phases (figure 1b-d) [42][43][44]. FT-IR spectra show peaks at 420 cm 21 corresponding to Zn-N (electronic supplementary material, figure S1). The spectra show that the vibration and rotation modes (less than 1600 cm 21 ) of the free dye are significantly suppressed due to the encapsulation. The difference between encapsulated and adsorbed dye can be illustrated using FT-IR spectra (electronic supplementary material, figure S2). Conventional microporous ZIF-8 has a small pore size (the largest cavity diameter and pore limiting diameter are 11.4 Å and 3.4 Å , respectively) compared to the dye, thus the dye cannot be located into the pore via adsorption. The band of O -H (3000 cm 21 ) of carboxylic groups in the L1 dye is broad in the case of adsorbed dye compared to the encapsulated dye (electronic supplementary material, figure S2). The adsorbed dye displays a band at 1840 cm 21 corresponding to cyano group C;N, which is extremely weak in the encapsulated dye. There is also a shift in the wavenumber of the carbonyl group at 1580 cm 21 , and 1595 cm 21 for encapsulated and adsorbed dye, respectively (electronic supplementary material, figure S2). These observations are due to association of the dye with the surface of ZIF-8. The band at 420 cm 21 corresponding to Zn-N is strong in the case of encapsulation compared to L1 adsorbed (electronic supplementary material, figure S2). Based on sulfur atom percentage, as determined from elemental analysis, the encapsulated dye concentrations were determined to be 1.3%, 1.8% and 2.5% for L1@ZIF-8, L1Fc@ZIF-8 and L1Fc 2 @ZIF-8, respectively.
TEM images of the prepared materials show that the crystal size of ZIF-8 ranges from 20 to 100 nm (figure 2), and decreases with the increase of dye size (L1Fc 2 . L1Fc . L1 in size due to ferrocene moieties, figure 1a). As previously reported, the addition of TEA to a solution of Zn(NO 3 ) 2 results in the formation of ZnO nanocrystals [39]. Addition of the dye to the formed ZnO leads to adsorption prior to conversion to ZIF-8 crystals after addition of Hmim. The large dyes such as L1Fc 2 may improve the dispersion of the formed ZnO nanocrystals and increase the nucleation rate over the growth rate of ZIF-8 crystal, resulting in small particles of ZIF-8. In addition, inter-particle mesopore structures are formed in L1Fc 2 @ZIF-8 ( figure 2). SEM images show that the particle size of dye@ZIF-8 for L1, L1Fc and L1Fc 2 ranges from 20 to 100 nm (electronic supplementary material, figure S3), which agrees with TEM images (figure 2). The colour of the synthesized materials confirms the presence of the dye in the final products, as shown in electronic supplementary material, figure S4. Results from EDX analyses reveal the presence of elements C, N, O, Zn and more importantly S from the dyes in ZIF-8 (figure 3f ). Although a strong signal from Fe is also observed, it cannot be used as evidence for the presence of the dye because this peak could appear due to stray X-rays in TEM instruments. By contrast, the peak from S confirms the presence of the dyes in the ZIF-8 agglomerate. We thus consider it evidence in favour of the presence of the dye in the same region as the ZIF-8 crystals ( figure 3). The spatial distribution of C, N, O and Zn was figured out using EELS spectrum imaging, and maps from this investigation are presented in figure 3b-e. These maps were generated by first applying weighted principal component analysis to the data cube and selecting the five components of highest variance for reconstruction. The maps from the signals over an energy window of 30 eV are presented in figure 3b-e. Of note is the presence of what appears to be a series of pores in the O, Zn and C maps, which we interpret to be the ZIF-8 mesoporous network. The distribution of O and Zn is strongly correlated within the MOF structure (correlation coefficient þ0.88). Oxygen is not expected to be present in the crystal structure of ZIF-8 or its pore under high vacuum, while the dye has oxygen, which we interpret to mean that the dye has been incorporated directly into the ZIF-8 framework.
The porosity of the prepared material was estimated using N 2 adsorption -desorption isotherms ( figure 4a,b). The surface areas extracted using the methods of Brunauer-Emmett-Teller and Langmuir are tabulated in electronic supplementary material, table S1 [45]. The surface area, pore volume and pore size increase with the increase of dye size (L1Fc 2 . L1Fc . L1, figure 4; electronic supplementary material, table S1), as determined using the Barrett -Joyner -Halenda method (BJH, figure 4b) and non-local density functional theory (electronic supplementary material, figure S5). The results reveal the presence of mesopores, i.e. hierarchical porous ZIF-8, and the pore size ranges from 5 to 20 nm with the majority at approximately 15 nm. The creation of mesopore is due to the dye template molecules.

Emission lifetime for free and encapsulated dyes
The free and encapsulated dyes show similar profiles for emission and excitation spectra (electronic supplementary material, figure S6). However, the excitation and emission signals for solid free dyes L1, L1Fc and L1Fc 2 are much noisier than those for the encapsulated ones due to aggregation and ferrocene moieties [6].
The emission lifetime measurements for the encapsulated dyes (L1, L1Fc and L1Fc 2 , figure 5) into ZIF-8 were performed using TCSPC. All the emission intensities for the solid powders versus time  showed multi-exponential components, due to the heterogeneity present in the solid phase. Thus, the average lifetimes were calculated to facilitate the comparison. All the average lifetimes for the solid forms of the dyes are shorter as compared to the corresponding dyes in solution, due to aggregation and twisting motion present in such dyes [6]. However, upon encapsulation of these dyes in ZIF-8, the corresponding emission lifetimes increase as follows: L1 (approx. 0.35 ns), L1@ZIF-8 (approx. 3.1 ns), L1Fc@ZIF-8 (approx. 2.6 ns) and L1Fc 2 @ZIF-8 (1.75 ns). The lifetimes for solid powders of L1Fc and L1Fc 2 are shorter than the laser pulse width (IRF % 70 ps) and could not be measured. The shorter lifetimes for encapsulated L1Fc and L1Fc 2 dyes are attributed to the charge transfer character present from Fc moiety to the L1 unit [6]. The base concentrations (0.7-7 mmol) have a negligible effect on the emission lifetime values for the encapsulated dyes. An increase of lifetime (greater than 4-8-fold) is observed for L1 and L1@ZIF-8, which is attributed to the suppression of both solid aggregations and twisting motions of the encapsulated dye.
HAADF survey image carbon nitrogen zinc oxygen    figure S9). The DSSC parameters from devices based on L1@ZIF-8 materials in terms of V OC (mV), short circuit current density (Jsc, mA . cm 22 ), fill factor (FF) and efficiencies (h) are tabulated in table 1. The typical V OC of standard TiO 2 -based DSSCs for I 2 -based system are known to be between 700 and 800 mV [30]. However, the working electrodes loaded with L1@ZIF-8 show a lower V OC (table 1). The highest V OC value was observed for cells based on a dye@ZIF-8 deposited using a layer-by-layer method (see experimental section), followed by a direct deposition method. The V OC value increases with the increase of the reaction time. This observation is in agreement with the previous study in [30]. Results show that the efficiency depends on the deposition method: layer-by-layer . direct deposition . in-situ growth. This can be linked to the higher loading of L1@ZIF-8 obtained by using the layer-by-layer method as compared to the other methods (electronic supplementary material, figure S7). The incident photon-to-electron conversion efficiency (IPCE) spectra for L1Fc, and L1Fc@ZIF-8 are shown in electronic supplementary material, figure S10. The IPCE spectrum of L1Fc shows a strong photoelectric response to ultraviolet light in the range from 400 to 600 nm, with the highest IPCE at 465 nm. A shift toward a shorter wavelength was observed for L1Fc@ZIF-8. The change of the shape of the IPCE spectrum for L1Fc@ZIF-8 may point to morphological alterations in the light absorbing dye due to encapsulation. It is not clear if the low performance of the systems in this study is caused by the low  dye loading or due to the absence of direct electron injection mechanisms from the dye to the metal oxide substrate [46,47]. The Nyquist plot of L1Fc@ZIF-8 shows a very high resistance for the encapsulated dye as compared to the free dye L1Fc (287 V) [6], due to the low conductivity of ZIF-8 (electronic supplementary material, figure S11). Encapsulation of the dye molecules limits their ability to uniformly bind to the semiconductor surface, affecting the electron transfer process at the photoanode. Further planned studies may explain the low efficiency and could improve the material performance.

Conclusion
Triethylamine-assisted synthesis of hierarchical porous ZIF-8 and a one-pot encapsulation of solar cell dyes were successfully achieved. The pore structure of the synthesized materials is tuneable and depends on the size of the encapsulated dye molecules. The encapsulated dyes show longer emission lifetimes compared to the corresponding non-encapsulated dyes, however, they tend to have lower DSSCs efficiencies. The encapsulation of solar cell dyes into MOFs may open further exploration for high DSSCs efficiencies.
Data accessibility. This article does not contain any additional data. All data are present in figures 1-5 in the main text and electronic supplementary material figures S1-S11.