Synthesis of NBCAnPbI₃ encapsulated CH₃NH₃PbI₃

First precursor solution was prepared by mixing of CH₃NH₃I (MAI, 0.196 g, 1.23 mmol), 4-carboxyanilinium iodide (CAnI, 0.326 g, 1.23 mmol), and PbI₂ (1.134 g, 2.46 mmol) in 2 ml of γ-butyrolactone at 60 °C for 12 h. 1-Amino-3-butyne (12.3 mmol) and the coupling agent 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC, 5 mg, 0.026 mmol) were added to the above solution for full conversion of −COOH to −CO-NH− group. The solution was stirred at room temperature for 18 h for 3 days, with periodic monitoring. The precipitated solid was collected and heated at 85 °C for 20 h to remove the residual solvent. A gradual shift of color from yellow to brown was observed during the heating and the dry powder was used for further characterization. A similar procedure was adopted to make a control perovskite sample, NBCAnPbI₃ for comparison of properties.

Synthesis of 0.5 NBCAnPbI₃{1 MAPbI₃}

This mixed perovskite was prepared by mixing CAnI and MAI in a ratio of 0.5:1 with stoichiometric amounts of PbI₂, followed by EDC-assisted alkyne functionalization, as described above.

Synthesis of MAPbI₃, CAnPbI₃ {MAPbI₃}, and TPbI₃{MAPbI₃}

To prepare a control sample of pure MAPbI₃, a reported literature procedure was followed.^[Reference Kim, Lee, Im, Lee, Moehl, Marchioro, Moon, Humphry-Baker, Yum, Moser, Grätzel and Park^16] CAnPbI₃{CH₃NH₃PbI₃} was prepared by heating the stoichiometric mixture of CAnI, MAI, and PbI₂ dissolved in γ-butyrolactone (2 mL) at 60 °C for 12 h at 80 °C till drying. p-Toluidinium lead (II) iodide (TPbI₃)-encapsulated MAPbI₃ (TPbI₃{MAPbI₃}) was prepared by mixing equimolar amounts of 4-toluidinium iodide (TI) and MAI along with the stoichiometric amounts of PbI₂ dissolved in γ-butyrolactone (2 mL) at 60 °C for 12 h. The precipitated solid was collected and heated at 80 °C to remove the excess solvent. The dry samples were used for further characterization.

Fabrication of photoanode

Fabrication of photoanode includes following steps: p25-TiO₂ paste (5 g, Degussa p25 powder) was prepared using the previously mentioned method.^[Reference Ito, Chen, Comte, Nazeeruddin, Liska, Pechy and Grätzel^17] A thin film of TiO₂ was cast over a cleaned ITO substrate (0.24 cm²), followed by calcination at 450 °C for 30 min. A slurry of NBCAnPbI₃{MAPbI₃} in γ-butyrolactone was spread over p25-TiO₂ film and heated at 85 °C for 20 h. In the third step, 0.1 g p25-TiO₂ paste (ethanol as solvent) was coated followed by heating at 85 °C for 60 min over NBCAnPbI₃{MAPbI₃}.

Functionalization and proof of encapsulation

Successful alkyne functionalization via amide linkage between CAnI and 1-amino-3-butyne was demonstrated using Fourier-transform infrared spectroscopy (FTIR) spectra [Fig. 1(b)]. Amide-NH (~3352 cm⁻¹) and amide >C=O (~1620 cm⁻¹) along with terminal C≡C–H (~3253 cm⁻¹), C≡C (~2115 cm⁻¹) stretching frequencies in the spectrum of the product after functionalization confirms the alkyne group incorporation through acid–amine coupling reaction. The absence of broad −OH vibrations and shift of >C=O peak (from 1696 to 1620 cm⁻¹⁾ in the IR spectra of the samples before and after functionalization showed the transformation of acid to amide groups in the final product. To prove the chemical formula of alkyne functionalized 4-carboxyanilinium lead iodide, same acid–amine coupling reaction was performed without adding MAI (see the Experimental section) and check with x-ray photoelectron spectroscopy (XPS). After successful alkyne functionalization [Fig. 1(c)], XPS survey scan performed at different regions of the sample revealed the presence of lead (Pb) and iodine (I) with an atomic ratio of 1:3 (Fig. S1). This 1:3 ratio of Pb:I along with successful alkyne functionalization indicates a possible molecular formula of our functionalized material as NBCAnPbI₃. The x-ray diffraction (XRD) pattern of NBCAnPbI₃ was indexed using TREOR software, which indicates an orthorhombic (Pmmn) crystal structure with lattice parameters, a = 19.6442, b = 9.244, and c = 9.1901. As we used MAI with a stoichiometric amount of PbI₂ along with the precursors of NBCAnPbI₃, the formation of a hybrid phase material NBCAnPbI₃{MAPbI₃} is investigated using XRD. The powder XRD data [Fig. 1(d)] of the target material, NBCAnPbI₃{MAPbI₃}, were compared with homogeneous pure MAPbI₃ and NBCAnPbI₃ to identify the presence of both components in the final product. The presence of multiple diffraction peaks of MAPbI₃ in NBCAnPbI₃{MAPbI₃} shown by a dotted line and * marks indicate the presence of MAPbI₃ in our target material NBCAnPbI₃{MAPbI₃}. Although some of the diffraction peaks of MAPbI₃ overlapped with the diffraction peaks of NBCAnPbI₃, the presence of two major diffraction peaks 110 and 220 (* marked) of MAPbI₃ in NBCAnPbI₃{MAPbI₃} indicates the presence of tetragonal phase of MAPbI₃^[Reference Oku^18] in our target material NBCAnPbI₃{MAPbI₃}. The diffraction peaks of the most intense peaks (marked as +) corresponding to the (200) and (110) planes of NBCAnPbI₃ indicate the presence of NBCAnPbI₃ phases inside the hybrid material NBCAnPbI₃{MAPbI₃}. In addition, the hybrid materials prepared with a reduced molar ratio 1:0.5 for MAI:CAnI (see the Experimental section), followed by functionalization with 1-amino-3-butyne, led to the formation of both phases in the sample with reduced water stability. Moreover, UV–vis spectra of pure MAPbI₃ (Fig. S2, i) and NBCAnPbI₃ (Fig. S2, ii) were compared with NBCAnPbI₃{MAPbI₃}(Fig. S2, iii). The broad and strong absorption of NBCAnPbI₃{MAPbI₃} (Fig. S2, iii) in the longer wavelength region indicate the presence of both MAPbI₃ and NBCAnPbI₃ in the material.

The formation of NBCAnPbI₃ shell over MAPbI₃ was investigated by transmission electron microscopy [TEM, Fig. 2(a)], which showed the presence of a well-connected network structure, instead of discrete particles. The presence of network structure implies an encapsulated architecture for the final materials, rather than the formation of a physical mixture of two different materials. Higher magnification image of the materials [Fig. 2(b)] indicates the presence of materials encapsulating discrete darker particles. High-resolution TEM micrograph of a dark spot [inset of Fig. 2(b)] shows lattice fringes with a d-value of ~0.278 nm, which matches with the (310) plane of tetragonal MAPbI₃^[Reference Oku^18] and brighter spot having a d-value of ~0.228 nm, which corresponds to the (104) plane of orthorhombic NBCAnPbI₃. No such encapsulating structure and the presence of discrete dark particles with 0.278 nm d-spacing were observed in the TEM micrographs of pure MAPbI₃ (Fig. S3). The control samples prepared using TI and CAnI using the same procedure (see the Experimental section) were also fully characterized and properties were compared. TEM micrographs of both CAnPbI₃{MAPbI₃} [Fig. 2(c)] and TPbI₃{MAPbI₃} [Fig. 2(d)] showed similar encapsulated structures with less extended architectures than the previously discussed NBCAnPbI₃{MAPbI₃}. This discrepancy in network structure formation could be attributed to different diffusion-limited aggregation (DLA) growth^[Reference Maxim, Ferreira, Vilarinho and Reaney^19] in the presence of phenylammonium cations. The microstructure of the in situ generated bilayer is suggestive of classical DLA of CAnPbI₃ over MAPbI₃ [Fig. 2(c)] as well as TPbI₃ over MAPbI₃ [Fig. 2(d)]. Such a growth mechanism is prevalent in systems that have anisotropic diffusion coefficient of molecular species (in this case CAnPbI₃ and TPbI₃) due to structural heterogeneity (e.g., diffusion coefficient at a step-“edge” can be less than the diffusion coefficient on a “planar” step). This random walk of molecules and irreversible sticking to the perimeter of growing aggregates is already discussed in a previous literature.^[Reference Brune, Romainczyk, Roder and Kern^20] The variation of sticking coefficient (K _stick) and change in cluster shape also explained earlier.^[Reference Ranguelov, Goranova, Tonchev and Yakimova^21] Also, lowering the value of K _stick affects the capture activity of branches of growing aggregates and channelize the diffusion toward center^[Reference Ranguelov, Goranova, Tonchev and Yakimova^21] to produce a network structure of NBCAnPbI₃{MAPbI₃} shown in Figs. 2(a) and 2(b). To identify the presence of different phases in the control samples, we performed powder XRD analyses (Fig. S4) of CAnPbI₃{MAPbI₃} and TPbI₃{MAPbI₃} perovskites. The peaks marked with a star in Fig. S4 indicate the presence of MAPbI₃ in both the materials. As expected, comparison of powder XRD data of pure CAnPbI₃ or TPbI₃ with mixed samples CAnPbI₃{MAPbI₃} and TPbI₃{MAPbI₃} showed the presence of phases corresponding to all components in the final samples (Fig. S4).

Figure 2.

TEM micrographs of (a) network of NBCAnPbI₃{MAPbI₃} and (b) magnified image of the network. The inset in (b) shows high-resolution micrograph where fringes of the dark spot with a d-spacing ~0.278 nm corresponding to the tetragonal MAPbI₃ and brighter region with a d-spacing ~0.228 nm corresponds to the shell of orthorhombic NBCAnPbI₃ (c) CAnPbI₃{MAPbI₃} and (d) TPbI₃{MAPbI₃} with their magnified images are shown in the inset.

Investigation of water stability

Extended network formation observed for NBCAnPbI₃{MAPbI₃} in TEM micrographs motivated us to investigate the water stability and compare with the established MAPbI₃ material. The black color of the material turns to yellow as soon as MAPbI₃ was immersed in water, which can also be monitored by UV–vis spectroscopy [Fig. 3(a)]. At the same time, no significant changes in colors were observed in the case of NBCAnPbI₃{MAPbI₃} after immersing in water for 30 min [Fig. 3(b)], which indicates high water stability of the target hybrid material. The instability of conventional MAPbI₃ was further checked by dipping the film in water just for 30 s and monitored by XRD. The disappearance of all the characteristic diffraction peaks and appearance of completely new diffraction pattern shown in Fig. 3(c) proved that even 30 s of water immersion is sufficient to break tetragonal packing of MAPbI₃ led to the formation of PbI₂,^[Reference Yang, Siempelkamp, Liu and Kelly^4] whereas minimum differences in powder XRD data before and after immersion in water for as long as 30 min [Fig. 3(d)] confirm the stability of NBCAnPbI₃{MAPbI₃}. To check the moisture stability of the NBCAnPbI₃{MAPbI₃}, the sample was kept inside a humidity chamber under 95% humidity for several days. Systematic XRD study of the test material exposed to moisture indicated no changes in the crystal structure (Fig. S5) within the first 10 days, which points to the high stability against moisture. The higher stability of the hybrid NBCAnPbI₃{MAPbI₃} perovskite toward humidity and water is due to the hydrophobicity of the shell, which is confirmed by contact angle measurements. NBCAnPbI₃{MAPbI₃} showed a higher contact angle [~104°, inset in Fig. 3(b)] as compared with hydrophilic MAPbI₃ (contact angle decreases quickly from 51^ο to 33^ο and stabilizes at 27^ο).^[Reference Yang, Wang, Liu, Cheng, Zhao and Yang^8] The instabilities of control samples, CAnPbI₃{MAPbI₃} and TPbI₃{MAPbI₃}, in water indicate the importance of incorporation of hydrophobic alkyne functionalization.

Figure 3.

UV–vis spectra of MAPbI₃ (a) and NBCAnPbI₃{MAPbI₃} (b) before (i) and after (ii) immersion in water. The inset shows the contact angle which was measured by placing a drop of water on top of the sample and recording the image using CCD camera. XRD spectra of MAPbI₃ (c) and NBCAnPbI₃{MAPbI₃} (d) before (i) and after (ii) immersion in water. Dotted lines represent the (110) plane of conventional MAPbI₃ indicating the presence or absence of MAPbI₃ in the system.

Proof of concept demonstration of water splitting with high applied bias photon-to-current efficiency (ABPE)

Since NBCAnPbI₃{MAPbI₃} is stable in water, the conduction band edge position of the shell material, NBCAnPbI₃, needs to be measured before performing an electrochemical reaction using the new mixed perovskite material. The standard electrochemical methods based on impedance spectroscopy and Mott–Schottky equation were used to calculate the flat band potential (see the Experimental section). NBCAnPbI₃-coated ITO was used as working electrode, Pt wire mesh, and Ag/AgCl electrode as a counter and reference electrodes, respectively. K₃[Fe(CN)₆] (7 × 10⁻⁴ M) dissolved in 1 M KCl solution was used as electrolyte to record impedance spectra. Mott–Schottky plot was prepared using a previously published procedure.^[Reference Gelderman, Lee and Donne^22] The point of intersection is independent of frequencies applied, and the flat band potential of NBCAnPbI₃ calculated from intersection point is ~−0.6 V versus Ag/AgCl [−0.403 V versus normal hydrogen electrode (NHE)] [Fig. 4(a)]. The potential value conversion from NHE to vacuum was done by using conventional methods described in the literature.^[Reference Grätzel^23] The calculated conduction band of NBCAnPbI₃ is −4.1 eV close to the conduction band of TiO₂ and lower than that of MAPbI₃,^[Reference Aharon, Dymshits, Rotem and Edgar^24] which enables the photogenerated electrons of MAPbI₃ to pass through the shell and available for electrochemical processes.

Figure 4.

(a) Mott–Schottky plot for NBCAnPbI₃ measured at different frequencies (i) 1526 Hz, (ii) 954 Hz, (iii) 595 Hz, which indicates the flat band position at ~−0.6 V versus Ag/AgCl. (b) Two electrode analysis for photoelectrochemical water splitting using NBCAnPbI₃{MAPbI₃} as a sensitizer (i = current under solar radiation, iii = dark current) and only with p25-TiO₂ (ii = current under solar radiation, iv = dark current). (c) Applied bias photo-to-current efficiency of p25-TiO₂ photoanode in the presence of NBCAnPbI₃{MAPbI₃} (i), p25-TiO₂ only (ii), and in the presence of NBCAnPbI₃ (iii).

Favorable band position and water stability of NBCAnPbI₃{MAPbI₃} imply the hybrid material to act as a sensitizer in the PEC application. As a proof of concept, the water-splitting reaction using the hybrid perovskite material as a sensitizer is demonstrated below. Details of photoanode fabrication are discussed in the Experimental section. A thin layer of commercial p25-TiO₂ was coated over NBCAnPbI₃{MAPbI₃} to facilitate oxygen evolution reaction and to regenerate the NBCAnPbI₃{MAPbI₃} by photogenerated electrons. The performance of fabricated heterostructures has been evaluated using a standard two probe method.^[Reference Hisatomi, Kubota and Domen^25] Under dark conditions, the thermodynamic requirement for water splitting is 1.23 V and a higher potential is required for effective water splitting due to overpotentials. However, in the presence of light, a good photosensitizer generates electrochemical bias from solar light such that the sum of the generated and applied electrochemical bias is >1.23 V. It is known that p25-TiO₂ is a mediocre PEC catalyst due to its high band gap,^[Reference Daghrir, Drogui and Rober^26] which is unable to show major changes in current in the presence and absence of light. On the contrary, a significant increase in photocurrent in the presence of NBCAnPbI₃{MAPbI₃}-coated p25-TiO₂ was observed, which implies the sensitizing capability of mixed perovskite [Fig. 4(b)]. To further quantitatively demonstrate the improved performance of photoanode, applied bias photon-to-current efficiency (ABPE) was evaluated using standard methods.^[Reference Chen, Jaramillo, Deutsch, Kleiman-Shwarsctein, Forman, Gaillard, Garland, Takanabe, Heske, Sunkara, McFarland, Domen, Miller, Turner and Dinh^27] The ABPE efficiency was calculated to evaluate PEC cell performance under external potential using Eq. (1) given below.

(1)$${\rm ABPE} = \left[ {\displaystyle{{ \vert j_{{\rm ph\;}} ({\rm mA\; cm}^{ - 2} ) \vert \times (1.23 - \vert V_{\rm b} \vert )\; ({\rm V})} \over {P_{{\rm total}} ({\rm mWcm}^{ - 2} {\rm \;} )}}} \right]_{{\rm AM 1}.{\rm 5G}} $$

where j _ph is the photocurrent density, V _b is the applied bias between working and counter electrodes, and P _total is the incident illumination power density (AM1.5G, 100 mW/cm²). Since photocurrent achieved at low applied bias values play a significant role to understand the efficiency of a photoactive material, we have estimated the activity of the photoactive assembly by measuring applied bias between the counter and working electrodes below the thermodynamic potential of water (1.23 V), at which the maximum photon-to-current conversion is achieved.^[Reference Upadhyay, Behara, Sharma, Gyanprakash, Pala and Sivakumar^28] ABPE of 0.64% was achieved by the photoanode in the presence of NBCAnPbI₃{MAPbI₃} [Fig. 4(c, i)], whereas ABPE value drops to 0.09% in the absence of NBCAnPbI₃{MAPbI₃} coating [Fig. 4(c, ii)]. A drastic improvement (~seven times) of ABPE in the presence of the hybrid perovskite indicates the important role of NBCAnPbI₃{MAPbI₃} as a sensitizing material. Moreover, no such photoresponsive behavior was achieved without the presence of MAPbI₃ (presence of only shell NBCAnPbI₃), which is clearly represented by 0% ABPE shown in Fig. 4(c, iii), which unerringly proved the sensitizing role of MAPbI₃ to achieve high ABPE. We attribute this enhancement to much higher photon absorption and diffusion length of the MAPbI₃^[Reference Yang, Siempelkamp, Liu and Kelly^4] core present in NBCAnPbI₃{MAPbI₃}. Facile e⁻–h⁺ pair separation is possible due to the favorable band position of NBCAnPbI₃, calculated using Mott–Schottky plot [Fig. 4(a)] with respect to the band position of MAPbI₃ and TiO₂.^[Reference Aharon, Dymshits, Rotem and Edgar^24] The calculated conduction band of NBCAnPbI₃ is −4.1 eV close to the conduction band of TiO₂ and lower than that of MAPbI₃,^[Reference Aharon, Dymshits, Rotem and Edgar^24] which enable photogenerated electrons of MAPbI₃ to pass swiftly to TiO₂ electrode and allow MAPbI₃ to be regenerated by photogenerated electrons from p25-TiO₂ (Fig. S6). Both these effects are responsible for facile e⁻–h⁺ pair separation for PEC water splitting with ~seven times increment of ABPE value [Fig. 4(c)]. Gas evolved from anode during the electrochemical reaction was identified as oxygen and bubbles generated at Pt cathode as H₂, confirming the water-splitting reaction instead of any other reaction inside the cell^[Reference Gao, Ding, Liu, Wang, Lu, Li and Sun^29] (Fig. S7). It is also true that limited regeneration of NBCAnPbI₃{MAPbI₃} by p25-TiO₂ is responsible for the slow oxygen evolution.^[Reference Wang, Zhong, Qin and Zhang^30] Such drawback can limit the photocurrent and reduces the life of the cell, as observed by the reduction in current density at a constant potential of 0.5 V (Fig. S8). This is corrected by adding methylene blue (MB) as a sacrificial agent.^[Reference Ji, Jang, Cho, Myung, Kim, Kim and Park^31] Chronoamperometry measurements in the presence of MB showed a constant current density over a period of time as compared with the data collected without MB (Fig. S8). We are currently optimizing the performance of the materials and further investigation involving the fabrication of cells with good oxygen evolution properties.