Spatial confinement growth of perovskite nanocrystals for ultra-flexible solar cells

Abstract

State-of-the-art flexible perovskite solar cells (PSCs) are generally built on conductive plastic substrates, but they are limited to bending strain effects. We present here the spatial confinement growth of (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ nanocrystals from a solvothermal-processed precusor for ultra-flexible PSCs made in a three-dimensional gel framework. Our focus is placed on systematic studies of photovoltaic behavior at arbitrary deformations. The optimized PSC yields a photoelectric conversion efficiency of 0.88% in the undeformed state, and it increases to 2.51% at a bending angle of 120°, to 3.04% at an elongation of 180%, to 2.35% at a twist angle of 360°, and to 1.79% at a compression ratio of 30%. The PSC demonstrates enhanced photovoltaic performance when suffering repeated deformations and remains 82.1% efficient when exposed in 70%-humidity ambient air over 120 h.

---

1 Introduction

Recent advances in energy research call for the rational design of perovskite solar cells (PSCs). One of the leading components of this type of solar cell is perovskite-structured methylammonium lead halide CH₃NH₃PbX₃ (X = I, Br, Cl) crystals with a unique morphology for excitingly new and interesting charge-transport properties.^1–3 Numerous studies on PSCs have been carried out on advanced cell architectures^4–8 or simplified solution fabrication processes,^9,10 with rising power conversion efficiency comparable to silicon solar cells. Compared with the PSCs made on glass substrates, flexible solar cells have triggered tremendous attention from academia and industry as they offer a convenient alternative energy source for indoor and outdoor applications.^11–13 Flexible PSC panels can be integrated with infrastructures of various shapes and sizes. Nevertheless, our ability to fabricate flexible PSCs is limited on conductive plastics¹⁴ or metal wires,¹⁵ only creating dependence of photovoltaic performance on bendability. Simple flexibility (bending) allows roll-to-roll production of solar cells but does not allow conformal bonding of these devices to non-planar substrates other than cylinders and cones. The need to impart elasticity in response to bending, tensile, twisting, and compressive strains must be considered before flexible PSCs can step in practical life. The combination of ultra-flexibility such as bendability, stretchability, twistability, and compression in a single PSC device has elevated great scientific opportunities in electronics.

Since the birth of PSCs, the pursuit of highly stable perovskite crystals at humidified atmosphere has been a persistent objective.¹⁶ Arguably one of the arising routes to fabricate perovskite layer is to utilize sequential⁹ or vapor deposition.¹ Here we report, for the first time, the spatial confinement synthesis of 6-aminocaproic acid assisted CH₃NH₃PbI₃ [(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃] nanocrystals by filling the solvothermal-processed precusor in a three-dimensional (3D) poly(acrylic acid)–poly(ethylene glycol) (PAA–PEG) gel framework. Due to the ultra-flexibility of the gel matrix, the resultant PSCs display arbitrary bendability, stretachability, twistability, and compression. Moreover, the amphiphilic nature of PAA–PEG framework can protect the perovskite crystals from being attacked by moisture.

2 Experimental

2.1 Synthesis of PANi

2.96 mL of aniline was dissolved in 100 mL of 1 M HCl aqueous solution to obtain a homogeneous solution. 20 mL of 0.225 g L⁻¹ ammounium peroxydisulfate in 1 M HCl aqueous solution was dipped in 20 mL of the above mixture within 10 min. The polymerization reaction was carried out at 0 °C for 3 h. The resultant reactant was rinsed by 1 M HCl aqueous solution, followed by deionized water to pH of ca. 6.5. Finally, the resultant PANi was filtrated, vacuum dried at 50 °C and dissolved in deionized water under ultrasonic dispersion to prepare a saturated PANi aqueous solution due to doping with protonic acid. After filtration, the concentration of the resultant PANi solution was around 0.5 mg mL⁻¹.

2.2 Synthesis of CH₃NH₃I

A mixture consisting of 21.9 mL of hydroiodic acid (45 wt%) and 10 mL of methylamine (40 wt%) having a molar ratio of 1 : 1 was made and transferred into a round-bottom flask and vigorously agitated in the ice bath. After 2 h, the remaining reagent was evaporated in a rotavapor at 70 °C. The precipitate was thoroughly rinsed with diethyl ether and vacuum dried at 60 °C.

2.3 Synthesis of HOOC(CH₂)₅NH₃I (6-ACAI)

Hydroiodic acid and 6-ACA with a molar ratio of 1 : 1 were added into round-bottom flask and vigorously agitated in an ice bath for 2 h. After that the remaining reagent was evaporated in a rotavapor at 70 °C.

2.4 Solvothermal synthesis of (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ precusor

CH₃NH₃I, 6-ACAI, and PbI₂ were dissolved in 4 mL of γ-butyrolactone, subsequently the mixture was agitated at 120 °C for 10 min. Finally, the reagent was transferred into a Teflon-lined autoclave. After the reaction at 120 °C for 12 h, the resultant (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ precusor solution was stored in a sealed condition. The molar ratio of CH₃NH₃I to 6-ACAI was determined at 25 : 1, while the masses of PbI₂ and CH₃NH₃I were 2.292 and 0.79 g, respectively.

2.5 Preparation of PAA–PEG/Ti/PANi/Ti architecture

16 mL of acrylic acid was dissolved in 20 mL of deionized water, while 8.8 g of PEG (M_w = 20 000) was dissolved in 20 mL of 0.5 mg mL⁻¹ PANi aqueous solution. By mixing the two solutions, 2 mL of 0.008 g L⁻¹ N,N′-methylene bisacrylamide and 2 mL of 0.225 g L⁻¹ ammounium peroxydisulfate aqueous solutions were dropped into the reagents. The polymerization reaction was kept at 80 °C in a water bath. The reagent was transferred into a Petri dish at a viscosity of ca. 160 mPa s⁻¹ for PAA–PEG/PANi. Subsequently, two Ti grids with size of 10 mm × 30 mm were embedded in the gel during consolidation process. The distance between bottom Ti grid and bottom surface of PSC as well as the distance between top Ti grid and top surface of PSC are both controlled at around 2 mm. The samples were vacuumly dried at 60 °C. Subsequently, the PAA–PEG/Ti/PANi/Ti was further immersed in an aqueous solution consisting of 2.96 mL of aniline and 10 mL of 0.33 M HCl for 2 days. Later, the PAA–PEG/Ti/PANi-anline/Ti was immersed in 0.225 g L⁻¹ ammounium peroxydisulfate aqueous solution for 2.5 h, and subsequently freeze-dried under vacuum at −60 °C for 72 h to obtain the final PAA–PEG/Ti/PANi/Ti architecture.

2.6 Assembly of PSC devices

The feasibility of assembling ultra-flexible PSC devices was confirmed by the following experimental procedures: the freeze-dried PAA–PEG/Ti/PANi/Ti was immersed in solvothermal-processed (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ precusor solution for one day and vacuumly dried at 60 °C, followed by immersing in 2 wt% phenyl C61-butyric acid methyl ester (PCBM) chlorobenzene solution for 30 min (only the upper layer of PAA–PEG/Ti PANi:(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/Ti architecture was immersed in PCBM solution by controlling the immersion of upper Ti grid in PCBM solution). After vacuum dry at 60 °C, the PAA–PEG/Ti/PANi:(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/PCBM:(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/Ti PSCs were obtained. Unless noted, complex PSC will be presented by 3-layered matrix of even P3LM.

2.7 Photovoltaic measurements

The characteristic photocurrent density–voltage (J–V) curves of the PSCs with an active area of 0.25 cm² were carried out by measuring on a CHI660E electrochemical workstation equipped with 100 W xenon arc lamp (XQ-500 W) and an AM1.5 G filter. When scanned from +1 to 0 V under irradiation of simulated solar light at a light intensity of 100 mW cm⁻² (the light intensity was calibrated using a standard silicon solar cell, Oriel-91150), the J–V curves were recorded by a linear sweep mode at a scanning rate of 100 mV s⁻¹. A black mask with an aperture area of around 0.25 cm² was applied on the surface of cell device to avoid stray light. Each J–V curve measured at an active area of 0.25 cm² was repeated measured at least 10 times to control the experimental errors within ±5%. Notably, it is difficult to exactly determine the active area of the solar cell devices during deformations. One of the commonly used strategies is to cover a black mask on the device. Although the device suffers deformation and the changes in real area, the projected area in the flat surface is controlled to 0.25 cm². All the photovoltaic parameters including short-circuit current density (J_sc), open-circuit voltage (V_oc), fill factor (FF), and power conversion efficiency were extracted from the characteristic J–V curves recorded at either undeformation or deformations. J_sc was determined as the current density at a zero voltage, while V_oc was the voltage at current of zero. FF and cell efficiency were calculated by the equations.

where P_in was the incident light power, P_max was the maximum power output, J_max (mA cm⁻²) and V_max (V) were the current density and voltage at the point of maximum power output in the J–V curves, respectively.

2.8 Other characterizations

The morphologies of the PAA–PEG matrix and 3-layered matrix of even P3LM were observed on a scanning electron microscope (SEM, SU8020), while the solvothermal-processed (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ precusor was subjected to transmission electron microscope (TEM, JEM2010, JEOL). The cross-sectional morphologies of the 3-layered matrix of even P3LM device were observed by confocal laser scanning microscopy. X-ray diffraction (XRD) pattern was recorded on an X-ray power diffractometer (X'pert MPD Pro, Philips, Netherlands) with Cu Kα radiation. The chemical composition of the perovskite crystals were detected by inductively coupled plasma-atomic emission spectra (ICP-AES). The optical transmission spectra of dried PAA/PEG matrix and swollen PAA/PEG gel were recorded on a UV-vis spectrophotometer at room temperature. Fourier transform infrared spectrometry (FTIR) spectrum was recorded on a PerkinElmer spectrum 1760 FTIR spectrometer. Incident photo-to-current conversion efficiency (IPCE) curves were obtained at the short-circuit condition on an IPCE measurement systems (MS260).

3 Results and discussion

The gel matrix for solar cell architecture is important in enhancing photovoltaic performances, the framework structure and therefore absorption capability are mainly dependent on concentration of acrylic acid, reaction temperature, reaction time, and dosages for N,N′-methylene bisacrylamide and ammounium peroxydisulfate. The polymerization conditions have been optimized in our previous work and directly employed to synthesize PAA–PEG gel matrix. Due to the extraordinary absorption, the three-dimensional PAA–PEG matrix can be employed to hold gigantic aniline monomer, HCl aqueous solution, ammonium peroxydisulfate aqueous solution as well as species for cell architectures. The absorption procedures require longer time in comparison with spin-coating method for state-of-the-art PSC devices.

Fig. 1a₁ shows that the color of mixture from CH₃NH₃I (MAI), 6-ACAI, and PbI₂ changes from yellow to dark brown after a solvothermal process at 120 °C for 12 h, suggesting the perovskite structure of colloidal (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃. The Pb/I ratio was determined to be 1.000 : 2.905. One of the crucial purposes of utilizing solvothermal method is to synthesize high-quality perovskite nanocrystals,¹⁷ which can be seeds for the growth of photosensitive halides during the drying processes. The precusor is subjected to transmission electron microscopy (TEM), as shown in Fig. 1a₂, yielding spheric and cubic nanocrystals with 15 ± 5 nm.

Fig. 1 (a₁) Solvothermal-processed (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ precusor and (a₂) corresponding TEM characterization. Cross-sectional SEM views of (b₁) PAA–PEG matrix and (b₂) PAA–PEG/Ti/PANi:(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/PCBM: (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/Ti cell along with (b₃) high-resolution SEM image and (b₄) SAD pattern of (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ crystal.

By plotting the square of the Kubelka–Munk function [F(R) × hν] versus hν, as shown in Fig. 2a, one can extract the bandgap of around 1.56 eV.¹⁸ The detection of a small diffraction peak at ∼11° indicates that PbI₂ species have nearly completely converted into perovskite (Fig. 2b). A comparatively strong peak at 14.07° is detected for the tetragonal structured perovskite, suggesting a preferred orgiantion of (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ along (110) direction.¹⁹ After being absorbed into 3D framework of PAA–PEG gel matrix, these nanocrystals are believed to be seeds for the growth of large-size (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ crystals. FTIR spectrum is employed to characterize the molecular structure of resultant perovskite (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ crystals, as shown in Fig. S1.† The characteristic band at 2925 cm⁻¹ is assigned to C–H stretching, while the band located at 1402 cm⁻¹ is attributable to C–C bending.²⁰ The detection of C=O stretching at 1661 cm⁻¹ indicates that the 6-ACA has successfully incorporated into crystal CH₃NH₃PbI₃. The peaks between 1000 and 1200 cm⁻¹ suggest that perovskite CH₃NH₃PbI₃ crystals have been formed within 3D PAA–PEG framework.²¹ Fig. 1b₁ shows cross-sectional view of freeze-dried PAA–PEG matrix, displaying homogeneous micropores with size of 5–15 μm. This microporous structure provides gigantic space for holding hole-transporting PANi, electron-transporting PCBM, and (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ nanocrystals. After assembling into a 3-layered matrix of even P3LM solar cell, the micropores have been filled by (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃, as shown in Fig. 1b₂. The high-resolution SEM in Fig. 1b₃ suggests the cubical topology of the final (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ after drying steps. The crystal size ranges from hundreds of nanometers to several millimeters. Due to the fabrication of PSC at atmosphere environment, exposing solvothermal-processed precursor to moisture during cell formation could result in accumulation of moisture within grain boundaries, yielding grain boundary creep, merging adjacent grains, and providing an aqueous environment to enhance diffusion length of the precursor ions. This may increase grain size and a high quality perovskite structure during recrystallization process.²² The selected area diffraction (SAD) confirms the perovskite structure, as shown in Fig. 1b₄. Due to the elastic nature of PAA–PEG gel matrix, the resultant PSC gives expression to ultra-flexibility such as bendability, stretchability, twistability, and compression (Fig. 3a). As drawn in Fig. 3b the energy-level diagram, the photogenerated electrons can jump from conduction band of perovskite crystals to PCBM percolating network and subsequently to Ti electrode and external circuit, leaving holes transfer to PANi. The efficient separation of electron–hole pairs is a prerequisite for realizing photovoltaic power generation in PSCs.

Fig. 2 (a) Tauc plots and (b) XRD pattern of (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ nanocrystals.

Fig. 3 (a) Photographs, schematic diagrams, and stress analysis of the PSCs at deformations. (b) Energy-level diagrams of the PSC device.

Confocal laser scanning microscopy photographs in Fig. 4 demonstrate stacking architecture and clear interfaces at PCBM/(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ and PANi/(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃, which cross-checks the feasible of flexible PSCs using 3D PAA/PEG gel framework (σ = 1.62 × 10⁻⁸ S cm⁻¹ at 25 °C) as a matrix. Moreover, the optical transmission spectra of dried PAA/PEG matrix and swollen PAA/PEG gel are shown in Fig. S2.† Apparently, the dried PAA/PEG matrix is nearly non-transparent, however, the swollen PAA/PEG gel matrix has an optical transmission of ∼10% at light wavelength of 400–1000 nm, yielding partial visible light penetrating into the gel framework for excitation of perovskite crystals.

Fig. 4 Confocal laser scanning microscopy images of (a) PCBM/(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ interface and (b) PANi/(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ interface.

Fig. 5a–d show representative photocurrent density–voltage (J–V) curves for the PSCs at deformations under AM1.5G sunlight, and the photovoltaic parameters are summarized in Tables 1–4. A PCE of 0.88% (V_oc = 0.320 V, J_sc = 10.49 mA cm⁻²) is determined on the ultra-flexible PSC at undeformed state, and this value is enhanced to 2.51% (V_oc = 0.452 V, J_sc = 18.38 mA cm⁻²) at a bending angle of 120°, to 3.04% (V_oc = 0.519 V, J_sc = 18.17 mA cm⁻²) at an elongation of 180%, to 2.35% (V_oc = 0.466 V, J_sc = 15.44 mA cm⁻²) at a twisting angle of 360°, and to 1.79% (V_oc = 0.394 V, J_sc = 17.14 mA cm⁻²) at a compression of 30%, yielding 185.2%, 245.5%, 167.0%, and 103.5% enhancements, respectively. In comparison the state-of-the art flexible PSC devices fabricated on conductive plastics¹⁴ or metal foils¹⁵ or even polymer tandem solar cell¹⁶ the undeformed efficiency of 0.88% is remarkably low. There are two reasons for the low efficiency: (i) the thick device leads to more traps and resistance for charge transport; (ii) the relatively low transparency of gel matrix results in the incomplete excitation of perovskite (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ crystals, offering low electron density and therefore efficiency output. The reasonable explanation for increased performances at deformations would be configurative changes for PANi, PCBM, and perovskite-structured (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ crystals. Moreover, the extracted V_oc and FF are also much lower than the state-of-the-art PSC device with several hundreds of nanometer in thickness,^23,24 which is attributable to markedly enhanced device thickness and reduced electron transport. As demonstrated in Fig. 1c, the PSC will suffer stresses in x-, y-, and z-directions at deformations. Take stretching deformation as an example, the device experiences a compressive stress at y- and z-directions, and a tensile stress at x-direction. The tensile stress along x-axis and compressive stress along y-axis increase the bulk densities of PANi, PANi, and (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃, while the compressive stress at z-direction shortens the charge-transfer distance and enhances the incident light intensity for electron photogeneration. These changes in bulk density and perovskite excitation are prone to increase electron generation, which can be confirmed by the increase in J_sc. Moreover, the deformations may break the intramolecular and intermolecular hydrogen bonds between within PANi or between PANi and PAA–PEG for enhanced electron delocalization, leading to rapid electron–hole separation. This may be the potential reason for enhanced V_oc at deformations. Moreover, the compressive stress at z-direction at deformations can reduce the device thickness and therefore yield increased V_oc values. Till now, the actual reasons for enhanced photovoltaic performances are still not clear, but the proposed mechanisms can explain well the determined evolution of photovoltaic parameters such as J_sc and V_oc. Although the measured PCEs are still incomparable with planar perovskite solar cells made on conductive plastic,^25,26 it still represents a significant step forward, as it is the first time to experimentally realize the multi-deformable quasi-solid-state PSC platforms. From a small standard deviation and limited hysteresis, as shown in Fig. S3 and S4,† we infer that the ultra-flexible PSCs with reasonable photovoltaic performances and high reproducibility can be realized using the method reported here. As shown in Fig. S5,† the IPCE spectrum of the PSC device at undeformed state indicates that the device shows a spectral response in the region from visible to near-infrared (400–800 nm) with a peak IPCE value of around 15% at approximately 500 nm. The observation for increased IPCE values at deformations suggests enhanced light absorption and therefore photoelectric conversion. This result cross-checks the enhancement in photovoltaic parameters during deformations.

Fig. 5 Stacking J–V characteristics of the PAA–PEG/Ti/PANi:(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/PCBM:(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/Ti device at different (a) bending angles, (b) elongations, (c) twisting angles, and (d) compressions.

Table 1 The corresponding PCE, J_sc, and V_oc values for the PSC device at different bending angles

Parameters	Bending angle (°)
	0	30	60	90	120
Voc (V)	0.320	0.393	0.422	0.437	0.452
Jsc (mA cm^−2)	10.49	13.51	15.64	17.31	18.38
PCE (%)	0.88	1.58	1.98	2.29	2.51

---

Table 2 The corresponding PCE, J_sc, and V_oc values for the PSC device at different elongations

Parameters	Elongation (%)
	0	20	40	60	80	100	120	140	160	180
Voc (V)	0.320	0.355	0.434	0.482	0.485	0.495	0.499	0.507	0.515	0.519
Jsc (mA cm^−2)	10.49	12.16	15.35	16.42	16.88	17.03	17.18	17.56	17.94	18.17
PCE (%)	0.88	1.14	1.81	2.47	2.53	2.68	2.83	2.85	3.03	3.04

---

Table 3 The corresponding PCE, J_sc, and V_oc values for the PSC device at different twisting angles

Parameters	Twisting angle (°)
	0	90	180	270	360
Voc (V)	0.320	0.424	0.447	0.456	0.466
Jsc (mA cm^−2)	10.49	11.86	13.31	14.29	15.44
PCE (%)	0.88	1.61	1.92	2.13	2.35

---

Table 4 The corresponding PCE, J_sc, and V_oc values for the PSC device at different compressions

Parameters	Compression (%)
	0	10	20	30
Voc (V)	0.320	0.345	0.357	0.394
Jsc (mA cm^−2)	10.49	11.66	14.00	17.14
PCE (%)	0.88	1.02	1.32	1.79

---

Repeated deformations are also carried out to demonstrate the performance evolution of the solar cell device. The stacking J–V characteristics are shown in Fig. 6a–d and corresponding photovoltaic parameters are summarized in Tables 5–8, yielding an increased PCE, V_oc and J_sc at arbitrary deformations. After suffering 100-cycle bends, the cell efficiency increases to 1.64% (V_oc = 0.349 V, J_sc = 15.57 mA cm⁻²) at 100^th bending, to 1.80% (V_oc = 0.446 V, J_sc = 14.45 mA cm⁻²) at 100^th elongation, to 1.94% (V_oc = 0.399 V, J_sc = 17.22 mA cm⁻²) at 100^th twist, and to 1.78% (V_oc = 0.499 V, J_sc = 13.70 mA cm⁻²) at 50^th compression, yielding 86.4%, 104.5%, 1120.5%, and 102.3% enhancements, respectively. Oppositely, the planar PSCs on conductive plastics will suffer a cell performance reduction at repeated deformations. In order to better understand the mechanical flexibility and failure mechanisms of flexible PSCs, Kelly has repeatedly bending flexible PSC devices.²⁷ The result demonstrates that the formation of cracks is the final originate for performance reduction. However, the PAA/PEG gel matrix is elastic with arbitrary deformation without any cracks, therefore the ultra-flexible PSCs from PAA/PEG gel matrix are expected to have extraordinary cell performances under deformations. The photovoltaic performance enhancement after suffering repeated deformations might correspond to the synergetic effect of two possible reasons. The first one is related to improved electron delocalization and therefore long-range hole transport along conjugated PANi chains at repeated deformations of PSC device, in which the intermolecular hydrogen-bonding between PANi ([double bond splayed left]C=N–, –C–N=) and PAA–PEG (–OH) matrix as well as intramolecular hydrogen-bonding within conjugated PANi chains may be broken during deformations. Another presumption is that the cell materials never fully return to their original state because of low macromolecular mobility, allowing for effective photogenerated electron–hole separation and light harvesting for (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ excitation. In this fashion, the quasi-solid-state PSCs reported here have good tolerance toward repeated deformations.

Fig. 6 Stacking J–V characteristics of the PAA–PEG/Ti/PANi:(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/PCBM:(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/Ti device at repeated (a) bending, (b) stretching, (c) twisting, and (d) compressive cycles. The maximum deformations in each cycle were 90°, 120%, 360°, and 30% for bending angle, elongation, twisting angle, and compression, respectively.

Table 5 The corresponding PCE, J_sc, and V_oc values for the PSC device at repeated bending cycles

Parameters	Bending cycle
	0	10	20	30	40	50	60	70	80	90	100
Voc (V)	0.320	0.329	0.334	0.339	0.343	0.345	0.346	0.347	0.348	0.349	0.349
Jsc (mA cm^−2)	10.49	12.14	12.75	13.62	14.02	14.28	14.59	14.92	15.11	15.33	15.57
PCE (%)	0.88	1.21	1.29	1.40	1.45	1.48	1.52	1.56	1.58	1.61	1.64

---

Table 6 The corresponding PCE, J_sc, and V_oc values for the PSC device at repeated stretching cycles

Parameters	Stretching cycle
	0	10	20	30	40	50	60	70	80	90	100
Voc (V)	0.320	0.333	0.363	0.388	0.407	0.418	0.427	0.433	0.438	0.449	0.446
Jsc (mA cm^−2)	10.49	11.31	12.19	12.84	13.14	13.50	13.79	14.09	14.31	14.52	14.45
PCE (%)	0.88	1.04	1.24	1.41	1.52	1.62	1.71	1.78	1.79	1.84	1.80

---

Table 7 The corresponding PCE, J_sc, and V_oc values for the PSC device at repeated twisting cycles

Parameters	Twisting cycle
	0	10	20	30	40	50	60	70	80	90	100
Voc (V)	0.320	0.322	0.332	0.345	0.359	0.369	0.376	0.385	0.389	0.397	0.399
Jsc (mA cm^−2)	10.49	10.72	11.45	12.33	13.49	14.37	15.23	15.83	16.34	16.78	17.22
PCE (%)	0.88	0.93	1.02	1.16	1.34	1.47	1.61	1.70	1.78	1.86	1.94

---

Table 8 The corresponding PCE, J_sc, and V_oc values for the PSC device at repeated compression cycles

Parameters	Compression cycle
	0	10	20	30	40	50
Voc (V)	0.320	0.372	0.377	0.414	0.457	0.499
Jsc (mA cm^−2)	10.49	11.07	12.24	12.39	13.19	13.70
PCE (%)	0.88	1.07	1.16	1.32	1.51	1.78

---

Despite great success in boosting power conversion efficiency, planar PSCs are still facing a critical challenge in instability of the devices to atmospheric moisture.^28,29 Due to the hydroscopic nature of perovskite-structured CH₃NH₃PbI₃, exposing methylammonium lead halide to moisture during cell operation could damage the crystallinity of the perovskite structure. Motivated by the amphiphilicity of PAA–PEG gel matrix in which the C=O (or –OH) groups in PAA bond with –OH (or –O–) in PEG by hydrogen-bonding interactions,³⁰ we sought to assess if this amphiphilic surface could function as a water-retaining layer. Fig. 7a represents the contact angle test of the device, yielding contact angles higher than 90° on device surface. Subsequently, we monitor the efficiency evolution of the device in ambient air (ca. 70% humidity, 25 °C). Fig. 7b plots the normalized PCE as a function of exposure time. The PCE of PAA–PEG/Ti/PANi/(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/PCBM:(6-ACA)_0.038(CH₃NH₃)_0.962PbI₃/Ti device is maintained over 82.1% of its initial value when exposure in the humidied air over 120 h. Possible mechanism behind the efficiency degradation using perovskite (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ is an amphiphilic surface for partial moisture attack. Future focuses will be placed on designing fully hydrophobic gel matrices having 3D frameworks. The study presented here is far from being fully optimized, especial incomparable cell performances with planar flexible PSCs, but the arbitrary deformations along with cost-effective and scalable matrices suggest the proposed ultra-flexible PSC devices hold great promise in specific applications. For example, the gel matrices can be made for biocompatibility by copolymerizing thermo-sensitive monomers with natural macromolecules such as starch, cellulose, chotosan, and gelatin, therefore the final PSCs may be potentially used as biosensors for clinic applications.

Fig. 7 (a) The contact angle test for the ultra-flexible perovskite solar cell device along with (b) dependence of PCE on exposure time in ambient air (ca. 70% humidity, 25 °C). PCE₀ and PCE correspond to the power conversion efficiencies before and after exposure, respectively.

4 Conclusions

In summary, we have successfully realized the spatial confinement growth of (6-ACA)_0.038(CH₃NH₃)_0.962PbI₃ nanocrystals by filling solvothermal-processed precusor into 3D gel framework, allowing for arbitrary deformations such as bendability, stretchablity, twistability, and compressibility. The preliminary results demonstrate that the ultra-flexible PSC has a reasonable efficiency of 0.88% at undeformed state, the efficiency increases to 2.51%, 3.04%, 2.35%, and 1.79% when suffering a bending angle of 120°, an elongation of 180%, a twisting angle of 360°, and a compression of 30%, respectively. Apart from excellent stability even at high strains and deformation cycles, the resultant device has a low moisture sensitivity, remaining a concern for large-scale device fabrication or persistent operation at atmosphere air.

Acknowledgements

The authors acknowledge financial supports from National Natural Science Foundation of China (21503202, U1037604), Collaborative Innovation Center of Research and Development of Renewable Energy in the Southwest Area (05300205020516009), and Shandong Provincial Natural Science Foundation (ZR2015EM024).

References

1. M. Liu, M. B. Johnston and H. J. Snaith, Nature, 2013, 501, 395–398.
2. Q. H. Li, X. Jin, Y. Yang, H. Wang, H. Xu, Y. Cheng, T. Wei, Y. Qin, X. Luo, W. Sun and S. Luo, Adv. Funct. Mater., 2016, 26, 254–266.
3. X. Jin, W. Sun, Z. Chen, T. Wei, C. Chen, X. He, Y. Yuan, Y. Li and Q. H. Li, ACS Appl. Mater. Interfaces, 2014, 6, 8771–8781.
4. Z. Wei, H. Chen, K. Yan and S. Yang, Angew. Chem., Int. Ed., 2014, 53, 13239–13243.
5. Y. M. Xiao, G. Y. Han, H. H. Zhou and J. H. Wu, RSC Adv., 2016, 6, 2778–2784.
6. Y. M. Xiao, G. Y. Han, Y. P. Li, M. Y. Li and J. H. Wu, J. Mater. Chem. A, 2014, 2, 16856–16862.
7. T. Y. Hsieh, T. C. Wei, K. L. Wu, M. Ikegami and T. Miyasaka, Chem. Commun., 2015, 51, 13294–13297.
8. Z. Ku, X. Xia, H. Shen, N. H. Tiep and H. J. Fan, Nanoscale, 2015, 7, 13363–13368.
9. J. Burschka, N. Pellet, S. J. Moon, R. Humphry-Baker, P. Gao, M. K. Nazeeruddin and M. Grätzel, Nature, 2013, 499, 316–319.
10. K. Wang, Y. Shi, Q. Dong, Y. Li, S. Wang, X. Yu, M. Wu and T. Ma, J. Phys. Chem. Lett., 2015, 6, 755–759.
11. S. S. Shin, W. S. Yang, J. H. Noh, J. H. Suk, N. J. Jeon, J. H. Park, J. S. Kim, W. M. Seong and S. I. Seoik, Nat. Commun., 2015, 6, 7410.
12. S. Das, B. Yang, G. Gu, P. C. Joshi, I. N. Ivanov, C. M. Rouleau, T. Aytug, D. B. Geohegan and K. Xiao, ACS Photonics, 2015, 2, 680–686.
13. S. Ameen, M. S. Akhtar, H. K. Seo, M. K. Nazeeruddin and H. S. Shin, J. Phys. Chem. C, 2015, 119, 10379–10390.
14. C. Roldan-Carmona, O. Malinkiewicz, A. Soriano, G. M. Espallargas, A. Garcia, P. Reinecke, T. Kroyer, M. I. Dar, M. K. Nazeeruddin and H. J. Bolink, Energy Environ. Sci., 2014, 7, 994–997.
15. L. Qiu, J. Deng, X. Lu, Z. Yang and H. S. Peng, Angew. Chem., Int. Ed., 2014, 53, 10425–10428.
16. A. Mei, X. Li, L. Liu, Z. Ku, T. Liu, Y. Rong, M. Xu, M. Hu, J. Chen, Y. Yang, M. Grätzel and H. W. Han, Science, 2014, 345, 295–298.
17. T. Zhang, M. Yang, E. E. Benson, Z. Li, J. Lagemaat, J. M. Luther, Y. Yan, K. Zhu and Y. Zhao, Chem. Commun., 2015, 51, 7820–7823.
18. Q. Jiang, D. Rebollar, J. Gong, E. L. Piacentino, C. Zheng and T. Xu, Angew. Chem., Int. Ed., 2015, 54, 7617–7620.
19. N. Yantara, F. Yanan, C. Shi, H. A. Dewi, P. O. Boix, S. G. Mhaisalkar and N. Mathews, Chem. Mater., 2015, 27, 2309–2314.
20. D. R. Chejara, M. Mabrouk, R. V. Badhe, J. A. S. Mulla, P. Kumar, Y. E. Choonara, L. C. Toit and V. Pillay, Carbohydr. Polym., 2016, 135, 324–333.
21. M. Yang, Y. Zhou, Y. Zeng, C. S. Jiang, N. P. Padture and K. Zhu, Adv. Mater., 2015, 27, 6363–6370.
22. J. You, Y. Yang, Z. Hong, T. B. Song, L. Meng, Y. S. Liu, C. Y. Jiang, H. P. Zhou, W. H. Chang, G. Li and Y. Yang, Appl. Phys. Lett., 2014, 105, 183902.
23. S. Kazim, M. K. Nazeeruddin, M. Grätzel and S. Ahmad, Angew. Chem., Int. Ed., 2014, 53, 2812–2824.
24. Q. Jiang, X. Sheng, Y. Li, X. Feng and T. Xu, Chem. Commun., 2014, 50, 14720–14723.
25. P. Docampo, J. M. Ball, M. Darwich, G. E. Eperon and H. J. Snaith, Nat. Commun., 2013, 4, 2761.
26. J. You, Z. Hong, Y. Yang, Q. Chen, M. Cai, T. B. Song, C. C. Chen, S. Lu, Y. Liu, H. Zhou and Y. Yang, ACS Nano, 2014, 8, 1674–1680.
27. K. Poorkazem, D. Liu and T. L. Kelly, J. Mater. Chem. A, 2015, 3, 9241–9248.
28. Y. Han, S. Meyer, Y. Dkhissi, K. Weber, J. M. Pringle, U. Bach, L. Spiccia and Y. B. Cheng, J. Mater. Chem. A, 2015, 3, 8139–8147.
29. J. Cao, J. Yin, S. Yuan, Y. Zhao, J. Li and N. Zheng, Nanoscale, 2015, 7, 9443–9447.
30. S. S. Yuan, Q. W. Tang, B. L. He and P. Z. Yang, J. Power Sources, 2014, 254, 98–105.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra08816c

---