Ultralong Perovskite Microrods: One- versus Two-Step Synthesis and

May 20, 2016 - ... been devoted to preparing CH3NH3PbI3 perovskite thin films (PTF) ..... are seen in the spectrum (space group P3m1 (164), JCPDS File...
0 downloads 0 Views 4MB Size
Article pubs.acs.org/JPCC

Ultralong Perovskite Microrods: One- versus Two-Step Synthesis and Enhancement of Hole-Transfer During Light Soaking Xiangyang Wu, Jin Wang, and Edwin K. L. Yeow* Division of Chemistry and Biological Chemistry, School of Physical and Mathematical Sciences, Nanyang Technological University, 21 Nanyang Link, Singapore 637371 S Supporting Information *

ABSTRACT: In this work, a two-step synthesis of CH 3NH 3PbI3 perovskite microrods involving first the production of PbI2 microrods by the evaporation of a dimethylformamide (DMF) solution of PbI2 is introduced. The PbI2 rods, acting as a shape template, yield rod-like perovskites of similar lengths and diameters when immersed in a solution of CH3NH3I. When comparable amounts of PbI2 are used, the two-step synthesis produces perovskite microrods that are shorter and more uniform in length than the ones prepared by a one-step synthesis via the evaporation of a DMF mixture of PbI2 and CH3NH3I. This is due to an uncontrolled crystal growth reaction in the one-step synthesis versus a nucleation dominated one in the two-step synthesis. The hole-transfer dynamics from the as-prepared perovskite rods to spiroOMeTAD is also examined. Both the photoluminescence (PL) intensity and lifetime of the perovskite in the absence of spiroOMeTAD are found to be enhanced as the CH3NH3PbI3 rods are continuously exposed to the excitation light. However, in the presence of spiro-OMeTAD, the quenched PL is gradually reduced as light irradiation progresses, and an increase in the holetransfer (HT) rate between the two moieties is observed. We propose that during the light soaking process, trapping of free charge carriers by electronic traps improve the overall electronic coupling between the perovskite and spiro-OMeTAD or/and facilitate the diffusion of free holes to the perovskite/spiro-OMeTAD interface. The phenomenon seen here has important implications as it offers a plausible explanation for the discrepancies in reported HT rates between perovskites and spiroOMeTAD. sequential steps: formation of a solid precursor (e.g., PbI2 film) followed by reaction with a second precursor (e.g., CH3NH3I). Solar cells based on nanostructures (e.g., nanowires) have been demonstrated to possess several advantages over thin-film devices (e.g., reduced reflection, and improved strain relaxation and band gap tuning properties).19 Since perovskite morphology and crystallinity play important roles in device performances, it is pertinent for scientists and engineers alike to be able to judiciously control the shapes and sizes of perovskite nanostructures. One-dimension nanowire perovskites have previously been synthesized using either the vapor phase technique20,21 or the more commonly employed one-step and two-step solution-processed methods.13,14,22−28 In the one-step synthesis, perovskite nanowires of lengths ranging from tens to hundreds of micrometers were observed when a mixture of PbI2 and CH3NH3I in the presence of dimethylformamide (DMF) was treated by the slip-coating process or allowed to undergo an evaporation-induced self-assembly (EISA) reaction on a solid surface.22−24 Nanowires of CH3NH3PbI3 were also

1. INTRODUCTION Organic−inorganic hybrid perovskite solar cells have received a tremendous amount of attention in recent years due to their high solar energy to electricity conversion efficiencies (as high as 20−21%), long-term stability, and low-cost of preparation.1−8 One of the earliest and most common perovskite for use in solar cells, light-emitting diodes, lasers, and photodetectors is methylammonium lead triiodide (CH3NH3PbI3).9−15 Significant efforts have been devoted to preparing CH3NH3PbI3 perovskite thin films (PTF) that can display optimal performances in optoelectronic applications.16 There are currently two conventional solution-processing methods of preparing PTF, namely, the one-step and twostep methods.1,17 The one-step method is a simple and costeffective technique of synthesizing PTF and involves the direct formation of a perovskite film after a precursor mixture of PbI2 and CH3NH3I is deposited (e.g., by spin-coating) onto a substrate. Unfortunately, the one-step method suffers from uncontrolled precipitation and growth of perovskite crystals. To circumvent this problem, the alternative two-step method is used which has a better control of the formation of PTF.18 In this case, polycrystalline perovskite films are synthesized by the © XXXX American Chemical Society

Received: April 10, 2016 Revised: May 19, 2016

A

DOI: 10.1021/acs.jpcc.6b03649 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Despite previous works on the effects of light soaking on the intrinsic photophysical properties of perovskites (e.g., PL intensity and lifetime),29,37−40 there remains many unanswered questions on the relationship between light curing and holetransfer efficiency to a HTM. In this study, we will investigate how exposing an as-prepared CH3NH3PbI3 perovskite rod coated with spiro-OMeTAD to low optical light excitation is able to perturb its overall hole-transfer efficiency. We will also discuss how this phenomenon may further shed light on the inconsistent hole-transfer rates reported in the past. This study is thus 2-fold. A new method of preparing CH3NH3PbI3 microrods via a two-step solution-based method that yields structures with a narrow-size distribution is introduced and the influence of light soaking on the changes in the hole-transfer rate to a HTM will be quantified.

prepared by spin-coating a mixture of CH3NH3I and DMF in isopropanol (IPA) onto a predeposited PbI2 layer.25 The nanowire perovskites formed from this two-step method were found to exhibit improved hole-transfer efficiency to (2,2′,7,7′tetrakis(N,N-di-p-methoxyphenylamine)-9,9-spirobifluorene) (spiro-OMeTAD) when compared to perovskites of other morphologies.25 Very recently, Zhu et al. proposed a method of directly converting a thin film of CH3NH3PbI3 into nanowires of various lengths and widths by spin-coating an IPA solution containing different concentrations of DMF onto the perovskite film.26 A grave concern associated with perovskite solar cells is the large hysteresis observed in current−voltage measurements under working conditions.29−32 There are several explanations provided to rationalize this behavior when a bias is applied: (1) presence of defect sites that trap electrons and holes,32,34 (2) ion migration through the perovskites,35 and (3) ferroelectric effects.36 Furthermore, both bulk samples and single nanoparticles of CH3NH3PbI3 are known to exhibit photoluminescence (PL) enhancement when exposed to continuous light irradiation.29,37−40 The light soaking or curing effect has been attributed to trap filling by photogenerated free charge carriers that reduces the nonradiative quenching pathways.37 Indeed, the passivation of electronic traps found on the surfaces and grain boundaries of perovskites by either pyridine or fullerene was found to enhance the PL of perovskites and increase their performances in solar cells.33,34,41,42 Efficient hole-transfer from perovskite to a hole-transporting material (HTM) such as spiro-OMeTAD followed by impeded charge-recombination are essential steps in high-performance solar cells.12,43−48 Therefore, understanding the photophysics of perovskite and its hole-transfer dynamics to spiro-OMeTAD provide useful information on the design of optimal devices. The reported hole-transfer rate from perovskite to spiroOMeTAD spans a wide range from ∼1 ps to 16 ns.48 Marchioro et al. have attributed the large discrepancy to nonuniform morphologies created during sample preparation.43 For example, large hole-transfer rates are expected when the HTM is well deposited into the pores of the perovskite, whereas poor interfacial contact due to insufficient HTM will lead to poor hole extraction. Makuta and co-workers have also provided an alternative explanation that involves the dependence of hole-transfer rate on photoexcitation intensity (i.e., the hole-transfer rate decreases with increasing excitation intensity).48 In this study, we will introduce a two-step synthesis method of preparing rod-like perovskites that first involves the formation of PbI2 rods (with lengths and diameters in the micrometer range) by evaporating a solution of PbI2 in DMF. The as-prepared PbI2 rods acting as morphology templates are subsequently converted into perovskites without losing their rod-like shapes and sizes. Unlike in previous two-step methods, the technique introduced here does not involve either a PbI2 or CH3NH3PbI3 film as an initial precursor.25,26 The dimensions of the perovskite rods formed from the two-step synthesis are then compared with those obtained from the one-step evaporation-induced self-assembly (EISA) method using comparable PbI2 concentrations. In addition, real-time visualization of the formation of perovskite crystals using time-lapse microscopy is also conducted in this study to gain a better insight into the mechanism of crystal formation (e.g., nucleation-dominated or growth-dominated). The differences between the one-step and two-step methods will be discussed.

2. EXPERIMENTAL METHODS 2.1. Materials. Dimethylformamide (DMF, HPLC grade), toluene (HPLC grade), and isopropanol (IPA, HPLC grade) were obtained from Tedia. PbI2 (98%), chlorobenzene (HPLC grade, 99.9%), poly(methyl methacrylate) (PMMA, average MW 120 000), and γ-butyrolactone (GBL, >99%) were obtained from Sigma-Aldrich. Methylamine (41% w/w aqueous solution) and (2,2′,7,7′-tetrakis(N,N-di-p-methoxyphenylamine)-9,9-spirobifluorene) (spiro-OMeTAD) were obtained from Sinopharm Chemical Reagent Co. Ltd. and Solaronix, respectively. Absolute ethanol (AR grade) and diethyl ether (≥99.5%) were obtained from Fisher Scientific UK and Merck Specialties Pte. Ltd., respectively. Hydroiodic acid (57% w/w aqueous solution) was obtained from Alfa Aesar. All chemicals were used as received. 2.2. Sample Preparation. CH3NH3I was synthesized according to a method described elsewhere.63 Basically, 5.06 mL of methylamine aqueous solution, 5.4 mL of hydroiodic acid, and 25 mL of ethanol were mixed at 0 °C for 2 h. The precipitate was recovered by rotary evaporation at 50 °C, and the raw product was washed with diethyl ether with stirring for 30 min. After filtration, the step was repeated three times and the product recrystallized using a mixture of ethanol and diethyl ether. The final white product was obtained and dried at 60 °C in vacuum for 24 h. In the two-step synthesis, PbI2 rods on different substrates (20 × 20 mm2 silicon wafer and precleaned glass converslip) were first obtained by drop-casting and evaporating a droplet (10 μL) of DMF solution of PbI2 (concentrations of 1 and 5 wt %) on the substrate surface at ambient conditions. The PbI2 rods were subsequently annealed at 75 °C for 15 min in an oven to remove any residual solvents. MAPbI3 perovskite rods were formed by immersing PbI2 rods into an IPA solution of CH3NH3I (0.076 M) for 15 min, and the product was annealed at 75 °C for 15 min. For experiments in reduced moisture and O2, a layer of PMMA was deposited on the MAPbI3 microrods by spin-coating 10 mg mL−1 PMMA/toluene solution (100 μL) at 2000 rpm for 2 min. In the one-step synthesis, a 10 wt % perovskite solution was first prepared by heating a DMF solution of PbI2 (7.3 wt %) and CH3NH3I (2.7 wt %) at 60 °C for 24 h under N2 atmosphere and filtering using a 0.2 μm syringe filter. MAPbI3 perovskite rods were produced by drop-casting a droplet of 10 wt % perovskite solution (50 μL) on a substrate and allowing it to evaporate at ambient conditions. The perovskite rods were then annealed at 75 °C for 15 min. Thin perovskite films were prepared by spin-coating a 10 wt % B

DOI: 10.1021/acs.jpcc.6b03649 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C perovskite/GBL solution (200 μL) at 4000 rpm for 2 min. For the hole-transfer experiments, the perovskites were coated with spiro-OMeTAD by spin-coating a chlorobenzene solution of spiro-OMeTAD (200 mg mL−1) onto the perovskite samples. In this case, the spiro-OMeTAD solution was dropped onto the perovskite and allowed to sit for ∼1 min before it was spincoated at 2000 rpm for 2 min. Solubility of PbI2 in pure DMF or 1.06 M MAI in DMF solution were measured using a dynamic method. Briefly, pure DMF or MAI/DMF solution was slowly added to dissolve a certain amount of PbI2 powder in glass vials, and solubility was determined by visually noting the volume of the solution used when the last trace of PbI2 solid completely disappeared. 2.3. Characterization. The powder X-ray diffraction pattern was obtained using an X-ray diffractometer (D8Advance, Bruker) and Cu Kα radiation. The morphology of the PbI2 and CH3NH3PbI3 microrods was characterized using field emission scanning electron microscopy (FESEM, JEOL-JSM7600F). The wide-field fluorescence microscopy setup consists of a microscope (IX 71, Olympus) with a halogen light source and a 633 nm HeNe laser source (Melles Griot). The excitation light was tuned to be circular polarized using λ/4 and λ/2 waveplates and expanded via a beam expander (ThorLabs) before being focused onto the back-focal plane of an air objective lens (20×, N.A. 0.4, Olympus). An excitation filter (Z633/10, Chroma) was used to filter the excitation light. The fluorescence was passed through a dichroic mirror (Z633rdc, Chroma) and an emission filter (HQ645lp, Chroma) and subsequently detected by a digital camera (Hamamatsu, C11440). The growth evolution of both the PbI2 and perovskite rods was monitored by first depositing a 5 μL droplet of either a PbI2/DMF or PbI2/MAI/DMF on a glass surface. Fluorescence intensity/lifetime images of single MAPbI3 perovskite fibers were obtained using a time-resolved confocal microscope (MicroTime 200, PicoQuant). Basically, an inverted microscope (IX71, Olympus) equipped with a scanning stage (P-733.2CL, Physik Instruments) was used. The excitation source used is a 635 nm pulsed laser diode (LDH-P-C-635B, PicoQuant) and the frequency used is 2.5 MHz for the samples in the absence of spiro-OMeTAD or 10 MHz for samples in the presence of spiro-OMeTAD. Circular polarized excitation light was achieved by using half-wave and quarter-wave plates. Circular polarized excitation light passed through an excitation filter (Z636/10, Chroma) before being focused through an air objective lens (20×, N.A. 0.4, Olympus) which was also used to collect the fluorescence. The fluorescence was then passed through a dichroic mirror (Z638rpc, Chroma) and an emission filter (FGL715, Thorlabs) before being detected by a single-photon avalanche diode (SPAD, SPCM-AQR-15, PerkinElmer).

Figure 1. PXRD patterns of PbI2 rods (black) and perovskite rods prepared by using the one-step (green) and two-step (red) synthesis methods (A). The reflection peaks corresponding to PbI2 and MAPbI3 are labeled as “∗” (12.66°, 38.74°, and 53.32° corresponding to the (001), (003), and (004) lattice planes of 2H PbI2 crystals, respectively) and “Δ” (13.90°, 24.13°, 28.37°, 31.73°, and 40.41° corresponding to the (110), (202), (220), (310), and (224) lattice planes, respectively50), respectively. FESEM images of PbI2 rods prepared from the evaporation of 1 wt % PbI2/DMF solution (B) and 5 wt % PbI2/DMF solution (C). FESEM images of perovskite rods formed by immersing the PbI2 rods obtained in part B in an IPA solution of CH3NH3I (D). FESEM images of perovskite rods from the one-step synthesis method (E). The insets in parts B−E display the magnified images of a single microrod. The scale bars are 100 μm for (B−E), 100 nm for the inset of part B and 1 μm for the insets of parts C−E.

The morphology of the PbI2 microrods is characterized using field emission scanning electron microscopy (FESEM). Figure 1B,C show the FESEM images of microrods formed from the evaporation of 1 and 5 wt % PbI2/DMF solutions on silicon wafer, respectively. The length and diameter distributions of 121 randomly chosen microrods formed from 1 wt % PbI2/DMF solution are fitted to a Gaussian function that yields an average length of 190.7 ± 39.1 μm (Figure 2A) and average diameter of 7.0 ± 2.3 μm (Figure 2B). When a higher concentration of PbI2 is employed (i.e., 5 wt % PbI2/DMF), the average length and diameter of the resulting microrods (73 randomly chosen rods are used in the analysis) are reduced to 21.3 ± 4.1 μm (Figure 2C) and 2.4 ± 0.7 μm (Figure 2D), respectively. In addition, it is observed that the length and diameter of the microrods become less disperse for the more concentrated PbI2 solution. It is worth noting that the formation of PbI2 microrods via solvent evaporation occurs on different surfaces including bare glass and indium tin oxide (ITO)-coated glass. The growth evolution of the PbI2 microrods in real-time is studied using time-lapse microscopy. Video S1 shows the

3. RESULTS AND DISCUSSION 3.1. Two-Step Synthesis of CH3NH3PbI3 Perovskites Microrods. The two-step synthesis of ultralong CH3NH3PbI3 (MAPbI3) perovskite rods first involves the preparation of PbI2 microrods, yellow in color, by the evaporation of a PbI2/DMF solution. Powder X-ray diffraction (PXRD) of the microstructures obtained using 1 wt % PbI2 in DMF on silicon wafer is shown in Figure 1A. Diffraction peaks at 2θ = 12.66°, 38.74°, and 53.32° corresponding to the (001), (003), and (004) lattice planes of 2H PbI2 crystals, respectively, are seen in the spectrum (space group P3m1 (164), JCPDS File No. 07-0235). C

DOI: 10.1021/acs.jpcc.6b03649 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 3. Microscope images of PbI2 rods prepared using a 1 wt % PbI2/DMF solution taken at the start (A) and at t = 194 s (B) of Video S1. In part A, PbI2 particles are labeled 1−4, and a growing PbI2 rod is labeled 5.

case of low supersaturation levels (e.g., 1 wt % PbI2 in DMF), crystal growth occurs more favorably, resulting in the observed longitudinal growth of the PbI2 crystal. Since there is a smaller number of nucleus sites, more monomers are able to attach onto existing crystals to promote growth.49 At high supersaturation levels (e.g., 5 wt % PbI2 in DMF solution), nucleation dominates which leads to the initial formation of a large number of nuclei and the rapid consumption of monomers. Therefore, more crystals with relatively shorter and more homogeneous lengths are formed.49 MAPbI3 perovskite microrods are prepared by immersing the as-prepared PbI2 microrods into an IPA solution of CH3NH3I (0.076 M) for 15 min. The color of the PbI2 microrods immediately turned from yellow to brown when exposed to CH3NH3I (MAI), demonstrating a fast in situ solid−liquid interfacial conversion reaction described by

Figure 2. Histograms of the length and diameter of PbI2 rods obtained using 1 wt % ((A) for length and (B) for diameter; 121 rods) and 5 wt % ((C) for length and (D) for diameter; 73 rods) PbI2/DMF solutions. Histograms of the length and diameter of the MAPbI3 perovskite rods obtained using the two-step synthesis from 1 wt % ((E) for length and (F) for diameter; 102 rods) and 5 wt % ((G) for length and (H) for diameter; 67 rods) PbI2/DMF solutions. The perovskites in parts E−H are obtained by immersing PbI2 microrods into an IPA solution of CH3NH3I (0.076 M) for 15 min.

PbI 2(s) + CH3NH3+(sol) + I−(sol) → CH3NH3PbI3(s) (1)

crystal growth of a typical microrod close to the edge of an evaporating 1 wt % PbI2/DMF droplet. As the DMF solvent evaporates, a supersaturation state is achieved and nucleation occurs. Continuous solvent evaporation stabilizes the nucleus which subsequently undergoes preferential growth along the (001) plane, as suggested by the most intense diffraction peak in Figure 1A. This growth is likely caused by the deposition of PbI2 molecules, either around the microrod or transported from the interior of the evaporating droplet to the edge (e.g., by convection), onto the crystal front. Furthermore, small PbI2 particles within the diffraction-limited area of the illuminating light (e.g., particles labeled 1, 2, 3, and 4 in Figure 3) and close to the growing microrod (labeled 5 in Figure 3) lower their energy by dissolving in the solvent and redepositing onto the microrod (see Video S1). When a higher concentration of PbI2/DMF solution is used (i.e., 5 wt %), the growth of a large number of relatively shorter and thinner crystals (i.e., ∼240 crystals on a surface area of 0.11 mm2) is observed to be completed within 5 s (see Video S2). On the other hand, a smaller number of crystals is formed from the 1 wt % PbI2/DMF solution (i.e., ∼20 crystals on a surface area of 0.11 mm2) where it is noted that crystal growth persists even after more than 2 min (e.g., Video S1). The overall size of microrods formed from the different concentrations of PbI2 can be rationalized using the classical crystallization theory.17 In the

The PXRD spectrum of the products, presented in Figure 1A, reveals the absence of peaks arising from PbI2 whereas new reflection peaks at 2θ = 13.90°, 28.37°, 31.73°, and 40.41° corresponding to the (110), (220), (310), and (224) lattice planes of MAPbI3 perovskites are observed,50 indicating the successful conversion of PbI2 into MAPbI3. The FESEM image in Figure 1D shows the shape of MAPbI3 perovskite obtained after conversion from PbI2 microrods (prepared using 1 wt % PbI2 in DMF). The perovskite retains the rod-like shape of its predecessor PbI2 microrod and exhibits an average length of 196.2 ± 45.7 μm and average diameter of 7.0 ± 2.4 μm (Figure 2E,F). In the case of perovskites prepared from PbI2 microrods using 5 wt % PbI2 in DMF, the average length and diameter are 21.7 ± 3.4 μm and 2.6 ± 0.5 μm, respectively (Figure 2G,H). These values are close to the ones measured for the corresponding PbI2 microrods (Figure 2A− D); indicating that the solution-based immersion method employed here to convert PbI2 to MAPbI3 has little effects on the overall shape of the latter. Yang et al. have shown that at intermediate to high MAI concentrations (>27 mM), PbI2 is dissolved in the iodide-rich solution to form PbI42−:51 PbI 2(s) + 2I−(sol) → PbI4 2 −(sol) D

(2)

DOI: 10.1021/acs.jpcc.6b03649 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

varying amounts of residual solvent on the intermediate and different evaporation rates. An obvious difference between the perovskites formed in the one-step versus two-step synthesis when comparable concentrations of PbI2 are used is the lengths of the rods formed. In the two-step method, significantly shorter MAPbI3 rods are observed (e.g., in the nucleation-controlled crystallization of 5 wt % PbI2 in DMF). In the case of the one-step synthesis, crystal growth is clearly the dominating process responsible for the creation of ultralong rods (>500 μm, see FESEM image in Figure 1E).22−24 The rate of nucleation is governed, in part, by the degree of supersaturation. Therefore, the dominance of nucleation can be impeded by increasing the solubility (or equilibrium concentration) of the solute so that a high supersaturation state is more difficulty to achieve. The solubility of PbI2 in DMF was determined to be ∼347 mg mL−1 in the absence of MAI. On the other hand, in the presence of 1.06 M MAI (molar ratio of PbI2 to MAI is 1.7:1), the solubility of PbI2 in DMF is significantly increased to ∼796 mg mL−1. The inclusion of MA+ in the MAPbI3·DMF intermediate most probably increases its solubility over that of PbI2. Our findings are in line with previous studies that have also shown that MAI aids in the dissolution of PbI2 in other aprotic solvents such as γ-butylrolacetone.17 When a 10 wt % perovskite solution is used in the one-step synthesis (see above), the amount of MAI is sufficiently high (molar ratio of PbI2 to MAI is 1:1.2) to induce low supersaturation states and promote active crystal growth. The role of DMF as a growth directing agent to facilitate the formation of nanowire perovskites has previously been discussed. Horváth et al. have suggested the presence of a stabilizing intermediate adduct between DMF and MAPbI3 (e.g., the dinuclear plumbate) as a plausible driving force behind the directed crystal growth.22 In the case of rod-like PbI2 crystals formed from the evaporation of PbI2/DMF solution, an adduct between DMF and PbI2 is created via Pb− O coordination bonding.55 The association between DMF and MAPbI3 or PbI2 may potentially create an energetic or entropic barrier which optimizes crystallization kinetics and suppresses exaggerated (or encourages directed) growth.56 It must, however, be stressed that the growth-directing mechanism of DMF remains ambiguous. 3.3. Time-Resolved Enhancement of Hole-Transfer during Light Soaking. The luminescence lifetime image of single MAPbI3 perovskite microrods, prepared using the twostep synthesis method with 1 wt % PbI2/DMF solution, on a glass substrate in ambient environment is given in Figure 4A. The power intensity of the excitation light (635 nm) is 0.05 W cm−2. The PL intensity trajectory of a randomly chosen fluorescent point on a MAPbI3 shows an increase in intensity even after the site on the perovskite has been exposed to light for 30 min (Figure 4C). In this case, the initial PL intensity of 12.8 counts ms−1 is enhanced 4.4 times after 30 min of light irradiation. Of the 76 randomly chosen points on 55 MAPbI3 microrods studied, ∼90% of them displayed PL increase during light exposure. The PL enhancement factor calculated from EF = (PL intensity after 30 min of light excitation/initial PL intensity) ranges between 1.05 and 20.8 (Figure 5A). The time-resolved PL lifetime decay profiles at three separate times and over a 2 s time-step for the PL trajectory in Figure 4C (i.e., points X, Y, and Z corresponding to 0, 120, and 1798 s, respectively, in Figure 4C) are described using a monoexponential decay function (Figure 4D). Among the three

We, therefore, hypothesize that in our system, the initially formed perovskite on the surface first dissolves in MAI solution to form PbI42−:52 CH3NH3PbI3(s) + I−(sol) → CH3NH3+(sol) + PbI4 2 −(sol)

(3)

The unreacted PbI2 beneath the surface of the microrod is thus exposed to I−, and the dissolution reaction 2 takes place. At sufficiently high concentrations of PbI42−, MAPbI3 perovskite is formed by the recrystallization reaction 4: CH3NH3+(sol) + PbI4 2 −(sol) → CH3NH3PbI3(s) + I−(sol)

(4)

In addition, MAI may also enter through the pinholes on the PbI2 surface (see insets of Figure 1B,C) to effect direct conversion of PbI2 beneath the surface to perovskite via the in situ reaction 1. The combined processes above ensure the near complete conversion of PbI2 to MAPbI3. Therefore, the continuous dissolution and recrystallization steps may result in MAPbI3 rods that are not larger in diameter compared to their predecessor PbI2 rods. In addition, unlike the PbI2 microrod which contains pinholes, the perovskite microrod is composed of several nonporous MAPbI3 crystal granulas packed together as seen in the inset of Figure 1D. The different packing arrangements of the materials may also lead to the observed similar dimensions for both the PbI2 and MAPbI3 rods. Im et al. have also prepared MAPbI3 nanowires using a twostep spin-coating procedure.25 By depositing an isopropanol solution of MAI containing various concentrations of DMF (50−100 μL) onto a layer of PbI2, the authors found that shorter wires are created when the DMF concentration is increased. This is likely due to a larger amount of predeposited PbI2 dissolving in the DMF when a larger volume of the aprotic solvent is present, leading to both a higher PbI2 concentration and supersaturation level. In this case, crystal nucelation is efficient and short nanowires are created, in line with the observations reported here. 3.2. One-Step Synthesis of CH3NH3PbI3 Perovskites Microrods. The growth evolution of the MAPbI3 perovskite microrods prepared via the one-step synthesis is studied using time-lapse wide-field microscopy (Video S3). A droplet of 10 wt % perovskite solution (i.e., 7.3 wt % of PbI2 and 2.7 wt % of MAI in DMF) was allowed to undergo evaporation on a glass substrate. The growth of the crystals toward the interior of the droplet is recorded in Video S3 (see first 2 min 26 s under white light illumination). When the white light was switched to a 633 nm laser light at time t = 2 min 27 s, no photoluminescence was observed from the microrods. This indicates that the microrods formed are not pure MAPbI3 perovskite and are most likely an intermediate that does not absorb the excitation light. Previous studies have suggested that the intermediate is a dinuclear plumbate ((MA+)2(PbI3−)2· DMF2) where MA+ is strongly bonded to DMF via Hbonding.53,54 After 7 min (i.e., at t = 9 min 28 s in Video S3), PL from the MAPbI3 perovskite rods is observed. This is due to the removal of DMF from the intermediate species by evaporation (i.e., solvent-loss process) which yields pure MAPbI3 microrods that absorb the 633 nm light. In Video S3, it is noted that the emergence of the PL is not uniformly nor simultaneously observed across the screen, possibly due to E

DOI: 10.1021/acs.jpcc.6b03649 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 5. Histograms of the enhancement factor (EF), reduction factor (RF), initial lifetime τi (obtained from the monoexponential PL decays constructed from photons between 0 and 2 s), and final lifetime τf (obtained from the monoexponential PL decays constructed from photons between 1798 and 1800 s) of several randomly chosen points on single perovskite microrods in the absence ((A) for EF, (C) for τi, (E) for τf) and presence ((B) for RF, (D) for τi, (F) for τf) of spiroOMeTAD.

Figure 4. PL lifetime images (size, 80 × 80 μm2) of single perovskite microrods in the absence (A) and presence (B) of spiro-OMeTAD. Typical PL intensity (black, bin-time = 2 s) and lifetime (gray) trajectories of a randomly selected point on a MAPbI3 pervoskite microrod in the absence (C) and presence (E) of spiro-OMeTAD. The PL lifetime decays and corresponding monoexponential fits for profiles constructed at times X, Y, and Z in part C are given in part D. The PL lifetime decays and the corresponding monoexponential fits for profiles constructed at times X and Y in part E are given in part F.

surfaces and grain boundaries of perovskite.29,37,41,57 Typical trap state densities ranging from 1016 to 1017 cm−3 for MAPbI3 have previously been reported.58 At the start of light irradiation, a large portion of the photogenerated electrons are captured by trap states which reduces the number of free electrons available for radiative recombination with free holes.37 This accounts for the relatively low PL intensity and short lifetime observed at the early stages of the measurement (e.g., point X in Figure 4C). As the optical excitation time increases, more trap states are filled resulting in a decrease in trap density (i.e., reduction in nonradiative quenching pathways), more occurrences of the radiative bimolecular electron−hole recombination, and an increase in luminescence (e.g., points Y and Z in Figure 4C). The diffusion length of free carriers (LD) is related to the PL lifetime (τ) and diffusion coefficient (D) via LD ≈ (Dτ)1/2. Therefore, the LD values at the start (point X) and end (point Z) of the light soaking experiment for the perovskite in Figure 4C are 363 and 587 nm, respectively, where D = 0.017 cm2 s−1.47 Since electron detrapping in MAPbI3 is an inefficient process occurring on a slow time scale (e.g., > millisecond time range),37 the density of unfilled defect sites is gradually reduced during light soaking. Therefore, at later illumination times, free charges are able to diffuse over relatively larger displacements, without being captured, before undergoing recombination, resulting in a lengthening of LD. Previous studies have proposed that the photogenerated electron−hole pairs in MAPbI3 perovskite exist as free carriers with only a small fraction (