3 perovskite solar cells

4 downloads 0 Views 5MB Size Report
Jul 14, 2017 - Degradation in perovskite solar cells stored under different ... PbI2 platelets for inverted planar organolead Halide Perovskite solar cells via ...
Home

Search

Collections

Journals

About

Contact us

My IOPscience

Trapping charges at grain boundaries and degradation of CH3NH3Pb(I1−x Br x )3 perovskite solar cells

This content has been downloaded from IOPscience. Please scroll down to see the full text. 2017 Nanotechnology 28 315402 (http://iopscience.iop.org/0957-4484/28/31/315402) View the table of contents for this issue, or go to the journal homepage for more Download details: IP Address: 203.255.190.41 This content was downloaded on 25/08/2017 at 08:26 Please note that terms and conditions apply.

You may also be interested in: Recent progress and challenges of organometal halide perovskite solar cells Liyan Yang, Alexander T Barrows, David G Lidzey et al. Enhanced optoelectronic quality of perovskite films with excess CH3NH3I for high-efficiency solar cells in ambient air Yunhai Zhang, Huiru Lv, Can Cui et al. Origins and mechanisms of hysteresis in organometal halide perovskites Cheng Li, Antonio Guerrero, Yu Zhong et al. Recent progress in efficient hybrid lead halide perovskite solar cells Jin Cui, Huailiang Yuan, Junpeng Li et al. Degradation in perovskite solar cells stored under different environmental conditions Abhishek K Chauhan and Pankaj Kumar PbI2 platelets for inverted planar organolead Halide Perovskite solar cells via ultrasonic spray deposition Gaoda Chai, Shizhen Wang, Zhonggao Xia et al. Enhanced photovoltaic performance of planar perovskite solar cells fabricated in ambient air by solvent annealing treatment method Vincent Obiozo Eze and Tatsuo Mori Recent progress in stability of perovskite solar cells Xiaojun Qin, Zhiguo Zhao, Yidan Wang et al.

Nanotechnology Nanotechnology 28 (2017) 315402 (13pp)

https://doi.org/10.1088/1361-6528/aa727e

Trapping charges at grain boundaries and degradation of CH3NH3Pb(I1−xBrx)3 perovskite solar cells Bich Phuong Nguyen1,3 , Gee Yeong Kim1,3, William Jo1, Byeong Jo Kim2 and Hyun Suk Jung2 1

Department of Physics, Ewha Womans University, Seoul 03760, Republic of Korea School of Advanced Materials Science & Engineering, Sungkyunkwan University, Suwon 16419, Republic of Korea

2

E-mail: [email protected] Received 8 March 2017, revised 18 April 2017 Accepted for publication 11 May 2017 Published 14 July 2017 Abstract

The electrical properties of CH3NH3Pb(I1−xBrx)3 (x = 0.13) perovskite materials were investigated under ambient conditions. The local work function and the local current were measured using Kelvin probe force microscopy and conductive atomic force microscopy, respectively. The degradation of the perovskite layers depends on their grain size. As the material degrades, an additional peak in the surface potential appears simultaneously with a sudden increase and subsequent relaxation of the local current. The potential bending at the grain boundaries and the intragrains is the most likely reason for the change of the local current surface of the perovskite layers. The improved understanding of the degradation mechanism garnered from this study helps pave the way toward an improved photo-conversion efficiency in perovskite solar cells. Supplementary material for this article is available online Keywords: perovskite solar cells, stability, Kelvin probe force microscopy (Some figures may appear in colour only in the online journal) 1. Introduction

stability issues would complete the next step toward commercialization [10]. MAPbI3 is sensitive to moisture, which can also exert significant adverse effects upon the stability of the composition of the perovskite layer, leading to degradation upon longer exposures to high humidity levels. Niu et al [11] proposed the decomposition of MAPbI3 to the PbI2 solid, an aqueous hydrogen iodide (HI), and/or the dissolution of CH3NH2 in water. Frost et al [12] proposed the formation of a colorless monohydrate MAPbI3.H2O crystal in an aqueous solution at room temperature. Christian et al [13] and Yang et al [14] investigated the interaction of MAPbI3 and water vapor under a controlled humidity; here, x-ray diffraction (XRD) patterns showed the characterization of a pale yellow dihydrate (MA)4PbI6.2H2O crystal through the appearance of a new peak. It has been shown that the exposure of MAPbI3 solar cells to a relative humidity greater than ∼90% rapidly degrades the device performance. Recently, Leguy et al [16]

The structure of the methylammonium lead halide (MAPbX3, X = halogen; CH3NH3 = MA) solar cell material is that of a perovskite crystal, and it is considered a promising photovoltaic material because of its high carrier mobility, high extinction coefficient, large absorption coefficient, large carrier-diffusion length, and suitable bandgap [1–4]. The power conversion efficiency (PCE) of perovskite-based solar cells increased from 3.8% [5] in 2009 to 22% [6] several years later. As a result of the rapid progress in this field, numerous studies are being conducted to further improve the efficiency by seeking the optimal preparation methods and compositions [7–9]. In terms of the material stability, however, a number of questions remain outstanding, and the resolution of the 3

Two authors contributed equally.

0957-4484/17/315402+13$33.00

1

© 2017 IOP Publishing Ltd Printed in the UK

Nanotechnology 28 (2017) 315402

B P Nguyen et al

observed that the pale yellow (MA)4PbI6.2H2O film rapidly dehydrated and regenerated the dark brown of the MAPbI3 film by flowing the dry air. They demonstrated the formations of the monohydrate MAPbI3.H2O at the early part of the degradation stage and the dihydrate (MA)4PbI6.2H2O upon longer exposure times. The insulation of the hydrated layer on the intragrains (IGs) resulted in the reduction of the PCE of the perovskite devices. The degradation of the perovskite film started along the grain boundary (GB). Li et al [17] confirmed the change in the film morphology and crystal structure under a high humidity (80%) by using in situ scanning force microscopy and in situ XRD. The appearance of the monohydrate MAPbI3.H2O phase led to the appearance of additional GBs. In contrast to Leguy et al, the monohydrate is unstable under a dry atmosphere and could be reversed to MAPbI3; however, the GB persisted. The presence of the GB could prevent the charge transport and act as the recombination centers. Li et al proposed that the presence of additional GBs caused an increased hysteresis under a high humidity. So, many open questions remain with respect to the degradation mechanism of perovskite for the attainment of a deeper understanding. In this study, the degradation of MAPb(I1−xBrx)3 (x = 0.13) perovskite materials of various grain sizes were investigated in a dark ambient environment under 50% humidity. The significant structural change upon the exposure to the humidity leads to a change of the electrical behavior before and after the degradation. Here, a description is provided regarding the way that scanning probe microscopy (SPM) can be used to obtain quantitative data on the electrical degradation of perovskite materials. This method allows for the characterization of the devices at the nanometer scale, whereby new insights into the degradation processes are gained.

The morphology of the perovskite device was investigated using scanning electron microscopy (FE-SEM, JEOL, JSM-6700F). The current–voltage characteristics of the solar cells were measured under solar-simulated light (Newport Oriel Solar 3A Class AAA, 64023A) with the aid of a potential (CHI 600D, CH Instruments). The local electrical properties were characterized using the measurements of Kelvin probe force microscopy (KPFM) and conductive atomic force microscopy (C-AFM) for which a commercial atomic force microscope (AFM) (Nanofocus Inc., n-Tracer) with a Pt/Ir-coated silicon cantilever (Nanosensor) was employed under atmospheric conditions (humidity ∼40%) (figure S2, available online at stacks.iop.org/NANO/28/ 315402/mmedia). The topography and surface potential images were obtained using the non-contact mode through the application of an AC voltage with an amplitude of 1.0 V. The applied frequency was 70 kHz. The scan size of the images was 3 μm × 3 μm. The scan speed was set to 0.5 Hz to reduce the topological signal. Local dark current maps were obtained from C-AFM measurements. The external sample bias was swept at 1 V with a time dependence. The as-grown reference samples were examined at the beginning of the KPFM and C-AFM measurements. The SPM-based measurements allowed for the quantification of the degradation level in the absorber material. SPM is a powerful tool that can be used to investigate the surface topography in the nanoscale domain; in addition to the topography, the electrical properties of thin films and devices can be investigated at the nanometer scale using a variety of different modes [19–21]. KPFM measures the height variation on the surface of the sample and simultaneously applies an electrostatic force between the sample and the tip, finally yielding the contact potential (UCP). The work function Fsample of the sample can be obtained from the measured UCP via the following simple formula: Fsample = Ftip - eUCP,

(1 )

where e is the elementary electric charge of an electron. Furthermore, KPFM can provide a map of the electric potential across the surface. The work function, surface charges or dipoles, and charge separation all affect the measured UCP [47]. As a result, KPFM is useful for studying the degradation mechanism of an active layer. In addition, the local current, grain boundaries, and electrical properties can be measured using C-AFM and KPFM [22–24]. Coffey et al [25] used C-AFM to map the local current in an active organic photovoltaic layer. They found that the average current measured with C-AFM is in sound agreement with the current for bulk devices and can therefore be correlated with the device performance as well. Also, XRD (D8 Advance (Bruker)) with CuKα radiation was utilized to compare the crystalline structures of various new and aged devices.

2. Experimental methods MAI and MABr powders were synthesized using a procedure that has been detailed elsewhere [18]. A 40 wt% solution of MAPb(I1−xBrx)3 (x ∼ 0.13, measured value by energy dispersive x-ray spectroscopy (EDS)) was made by reacting MAI and MABr and PbI2 at a 1:1 mole ratio in N, N-dimethylformamide (DMF) at 60 °C for 12 h. The MAPb(I1−xBrx)3 (x ∼ 0.13) in DMF was deposited on a mesoporous TiO2/compact TiO2/fluorine-added tin oxide (FTO) film. The S1 and S2 samples were then annealed on a hot plate at 95 °C for 1.5 h and for 15 min, respectively. Then, 25 μl of the hole transport layer (HTM) was deposited using a spin coater at 2500 rpm for 45 s. With the use of the spin coater, 25 μl of the HTM mixture (consisting of 80 mg of 2,20,7,70-tetrakis(N, N-di-pmethoxyphenyl-amine)-9, 90spirobifluorene (spiro-MeOTAD), 8.4 ml of 4-tert-butylpyridine, and 51.6 ml of a bis (trifluoromethane) sulfonamide lithium salt (Li-TFSI) solution (154 mg ml−1 in acetonitrile)) was deposited after its dissolution in 1 ml of chlorobenzene. Lastly, a top Ag electrode was deposited by thermal evaporation.

3. Results and discussion The MAPb(I0.87Br0.13)3 perovskite thin films were deposited on mesoporous TiO2/compact TiO2/FTO substrates with different grain sizes. The grain size of the S1 sample 2

Nanotechnology 28 (2017) 315402

B P Nguyen et al

Figure 1. Topography images of the surfaces of the S1 sample (a)–(e) and the S2 sample (k)–(o) with the time dependence, and the

corresponding KPFM images (f)–(j) and (p)–(t), respectively. The surface potential range of the S1 sample is from ∼70 mV to ∼100 mV, and that of the S2 sample is from ∼90 mV to ∼150 mV.

(meso-TiO2 + perovskite layer) is ∼300 nm, while the grain size of the S2 sample is ∼100 nm, as is shown in figure S1. In figure S1, the measured J–V curves of the devices are displayed under AM1.5 solar irradiation. The PCE improved as the grain size increased, from 7.4% for the S2 sample to 10.73% for the S1 sample. Both the short circuit current (JSC) of ∼14.69 mA cm−2 to ∼17.91 mA cm−2 and the open circuit voltage (VOC) of ∼0.88 V to ∼0.98 V also improved when the size was enhanced. For the measurement of the degradation of photovoltaic devices, the decreases in the JSC and VOC were determined. However, because many of the thin film layers that are used as hole- or electron-blocking layers form an interface with the perovskite material, they might influence the degradation chemistry at their

interface [14]. Here, the semi-structure device (glass/FTO/ compact TiO2/meso-TiO2/MAPb(I0.87Br0.13)3) was used to investigate the degradation behavior of the perovskite material. The morphological, chemical, and electrical properties during the degradation process was examined over 310 h under an ambient condition. In figure S3, the initial color of the as-grown thin films is a dark brownish black while the degraded films changed to a yellowish color. In particular, the S2 sample became almost yellow after the 310 h, while the dark color in the S1 sample was partially retained. Figure 1 presents the topography map and the local work function of the MAPb(I0.87Br0.13)3 film with different size casts on the FTO. The data for the ‘as-grown film’ were obtained during

3

Nanotechnology 28 (2017) 315402

B P Nguyen et al

Figure 2. Local work function distribution of the two samples at the start and end of the KPFM measurements. The widths of the surface function distribution increased after 310 h of storage. The numerous work function peaks indicate that a diverse phase with a different composition ratio might exist in the surface.

the start of the KPFM measurements under 50% humidity. The topography images show a uniform morphology and a better surface coverage of the perovskite on the mesoporous TiO2 of the S1 sample. The grain size of the surface is ∼300 nm for the S1 sample and ∼100 nm for the S2 sample. The size of the perovskite material increased as a function of the time, and after the 310 h, the achievements for the S1 and S2 samples are ∼500 nm and ∼300 nm, respectively. The

work function distribution obtained from the KPFM images, calculated from equation (1), for the two samples that were stored in an atmospheric environment are shown in figure 2 as a function of the time. The average work function of the two samples before degradation is low with a value of ∼4.72 eV compared with the more typical work function of 4.8 eV [32]. The average value of the work function in the dark for the S1 sample does not vary much during the degradation process; it 4

Nanotechnology 28 (2017) 315402

B P Nguyen et al

It can be expected that the formation of a decomposed phase will lead to the change in the surface work function distribution [35, 36]. The work function distribution fits well with a Gaussian distribution. The peak at 4.7 eV might be the major peak of the perovskite. The peak above 4.8 eV can be attributed to the decomposed phase because it appears at 288 h with the S1 sample and at 48 h with the S2 sample during the degradation of the perovskite layer. In previous studies, MAPbI3 perovskite changed to hexagonal PbI2 and/ or an intermediate hydrated phase during the degradation [11–17]. The PbI2 phase and the hydrated compound (MA)4PbI6.2H2O were detected using grazing incidence x-ray diffraction (GIXRD) [14]. In other studies, the disappearance of the characteristic perovskite phase under 90% humidity occurs after 336 h, and the presence of a new peak associated with crystalline PbI2 was observed using femtosecond transient absorption spectroscopy [29, 31]. The present sample shows the peak of the PbI2 crystal in the XRD pattern after 192 h (in figure S4). Also, the work function of the PbI2 film was measured using the KPFM measurement. The range of the work function of the PbI2 film is from 4.8–5.1 eV (in figure S5), compared with the work function of the bulk PbI2 obtained using ultraviolet photoemission spectrometry (UPS) is 6.35 eV [38]. It is therefore expected that the peak in the range from 4.8–5.0 eV represents a PbI2 phase in the perovskite layer. Figure 3 shows the change of the work function of the MAPb(I0.87Br0.13)3 and the PbI2 in the perovskite thin films as a function of the time. Clearly, the decomposed phase in the perovskite layer of the S2 sample forms more rapidly than that of the S1 sample. From the distribution of the work function, the full widths at half maximum (FWHM) that were derived from the work function peaks in the KPFM measurements were obtained. The FWHM value of the work function can be used to compare the phase distributions. If the FWHM value of the work function of one sample is lower than that of the other sample, then the phase distribution of the sample with the low FWHM value is more uniform. In figure 3, the FWHM value of the PbI2 peak of the S2 sample is approximately 0.1 from 50 h and increases to 0.4 at 310 h, while that of the S1 sample is 0.25 from 288 h and increases to 0.31 at 310 h. The morphology of the perovskite layer could affect the decomposition rate of the perovskite material. Kim et al [39] found that the samples with larger particles in the perovskite layer have a greater stability because the larger grain size reduces the moisture permeability and tends to form an air barrier. Therefore, the crystalline structure of the S1 sample with larger particles on the top of the mesoporous TiO2 layer prevented the degradation of the underlying perovskite layer; the PbI2 peak only appeared after 288 h of exposure. Interestingly, although the S2 sample perovskite layer decomposed the quickest, the main peak remained at the same position until the 244 h mark. Han et al [40] observed that the decomposition rate slowed with the increasing of the grain size. The results demonstrate that larger perovskite grain sizes improve the stability of the perovskite absorber layer [39, 40].

Figure 3. A potential peak is evident at 4.7 eV, and this is the major perovskite peak. A potential peak is evident at 4.82 eV, and this is the PbI2 peak. (a) Position of the perovskite peak and the FWHM value of the perovskite film with different grain size as a function of the storage time. (b) Position of the lead iodide peak and its FWHM value.

is centered between 4.65 eV ∼ 4.7 eV; however, the situation is different for the S2 sample, where a broader range of the work function values is present, depending on the time. The range of the work function is between 4.8–5.05 eV. The work function distribution of the S2 sample started changing after 48 h. These results are similar to those of previous studies [15, 33]. Noh et al [15] reported that MAPb(I1−xBrx)3 (x < 0.2) solar cells are very sensitive to humidity. The MAPbI3 structure is tetragonal while the MAPbBr3 structure is cubic; because the cubic structure is more stable than the tetragonal structure, the MAPb(I1−xBrx)3 (x … 0.2) films are more stable. In addition, Misra et al [34] reported that Brcontaining perovskite materials have a better stability due to the formation of a stronger bond of shorter length with Pb–Br than with Pb–I. The MAPb(I0.87Br0.13)3 perovskite decomposed to PbI2 and/or a hydrated phase in air, as indicated by the color change from dark brown to yellow. In the present results, however, the surface work functions of the S1 sample are almost unchanged after a 96 h storage period, indicating that the grain size of the perovskite material exerts an influence on the degradation. 5

Nanotechnology 28 (2017) 315402

B P Nguyen et al

Figure 4. Statistical characterization of the surface potential focusing on the intragrains based on the line-profile results of the many regions of the intragrains of the perovskite films with the S1 sample (grain size ∼300 nm) (a)–(e), and the S2 sample (grain size ∼100 nm) (f)–(j). All of the samples show a high positive potential rate at the second step of the degradation process. The change in the potential distribution at the grain boundaries of the S2 sample is faster than that of the S1 sample. Note: scale bar is −500–500 mV in (e), (i), and (j).

6

Nanotechnology 28 (2017) 315402

B P Nguyen et al

Figure 5. Statistical characterization of surface potential focusing on the grain boundaries based on the line-profile results of the many regions of the grain boundaries of the perovskite films with the S1 sample (grain size ∼300 nm) (a)–(e), and the S2 sample (grain size ∼100 nm) (f)– (j). All of the samples show a high positive potential rate at the first step of the degradation process. The change in the potential distribution at the grain boundaries of the S2 sample is faster than that of the S1 sample. Note: scale bar is −500–500 mV in (e), (i), and (j).

7

Nanotechnology 28 (2017) 315402

B P Nguyen et al

Figure 6. Local current map images of the perovskite thin films with the time dependence at a bias voltage of 1 V. The topographies of the S1 sample (a)–(e) and the S2 sample (k)–(o); and the corresponding C-AFM results (h)–(j) and (p)–(t), respectively.

perovskite layer toward the GBs, whereas the majority carriers (holes) are repelled due to the barrier. Moreover, the surface potential difference between the GBs and IGs creates the barrier to prevent the electron–hole recombination. The larger potential barrier of the S1 sample, ∼30 mV compared with the ∼10 mV of the S2 sample, provides a more efficient suppression of the electron–hole recombination after 48 h. During the exposure under the ambient condition the positive surface potential ratio was increased. The S1 sample shows 77% of the higher surface potential ratio at the GBs (33% increase), while the S2 sample shows 68% (36% increase) after 96 h. These results explain the faster rate of the GB formation of the S2 sample. The GB formation is fairly significant in terms of the degradation process due to the water as an absorber [17, 26].

To obtain a more in-depth understanding of the mechanism behind the observed degradation, the role of the GB in the perovskite layers was considered, especially the potential distribution at the GBs that express the local surface state of the perovskite layer. Figures 4 and 5 show the statistical characterization of the local work function focus regarding the GBs and the IGs. The histograms of the work function distribution are indicated at the GBs or the IGs based on the results of the line profiles in many regions in each sample. Before the degradation, figures 4 and 5 show higher surface potentials of the two samples at the GBs compared to the IGs. It is suggested that the positive surface potential at the GBs forms a downward band bending [28, 30]. The downward band bending in the energy band diagram lead to the attraction of the minority carriers (electrons) in the p-type 8

Nanotechnology 28 (2017) 315402

B P Nguyen et al

sample is broader and higher (−150 mV) than that of the S1 sample (−80 mV); this surface potential increase in the IGs may be due to the formation of PbI2. Ahn et al [26] proposed that the charges are trapped along the GBs. The irregular fields formed by the trapped charges could distort the structure of the hydrated perovskite, which causes the beginning of the degradation process. In addition, the distortion of the perovskite structure leads to an easy penetration of the water molecules into the perovskite structure. The faster GB formation in the perovskite layer causes the faster degradation of the S2 sample. The surface potential distribution of the S1 sample after 310 h is similar to the distribution of the S2 sample at 144 h, and this means the S1 sample starts to degrade. The S2 sample is now yellow, which is from the degradation after 310 h. The surface potential in the IGs is the broadest and highest (−150–300 mV), and this potential distribution can express the following two parts: the pre-degradation potential range of −150–150 mV is similar, and the potential range of 150–300 mV is the increased potential that is due to the formation of the PbI2. To demonstrate the way that the surface potential observed in the KPFM data affects the current flow in the perovskite layer, the current flow across the MAPb (I0.87Br0.13)3 film was investigated using C-AFM measurement. The local current distribution was characterized by a resolution of a few nanometers, and it is well-suited for the detection of changes in the electrical properties during the degradation process. The local current mapping images and topography maps shown in figure 6 were obtained by applying a positive voltage bias of 1 V in the dark as a function of the time. Clearly, the local current passes through the large grains of the perovskite thin films. The S1 thin film exhibited a higher local current flow through the surface (∼800 pA) while the S2 sample exhibited a lower predegradation value (∼500 pA). Our group also reported that the local current flows through the grains of the MAPb(I1−xBrx)3 thin films [46]. Certainly, the current flow disappeared after the 310 h exposure under the ambient condition due to the decomposition of the perovskite material. The average local currents of the two samples in storage are shown in figure 7(a). The variation in the average local current depends on the thickness of the perovskite film. The local current flow of the two samples increased during the first few days of the storage under the ambient conditions (first step), but decreased quickly thereafter (second step). The plot of the current distribution is consistent with the stability trends of Br-containing perovskite solar cells [15]. From the KPFM measurement, the beneficial effect of the GBs in the degradation process is shown. The GBs were increasingly formed and acted as channels for the current flow, so that the local current flow could increase in the first-step degradation; however, as there is no current in the second step, the authors suggest that the new decomposed phase species that appear in the perovskite layer are responsible for these findings. Li et al [41] reported that the poorly conducting grains in the perovskite layer showed no current flow in the C-AFM measurement. The poorly conducting grains are indicative of the PbI2 phase that is generated in the annealing process

Figure 7. The average local current value plotted from the current

mapping at the bias voltage of 1 V of the two perovskite samples with different grain size. (a) The current values were obtained from the C-AFM measurements in an ambient environment over time. (b) Fitting of the current distribution to explain the degradation mechanism of the perovskite solar cells depending on the grain size of the absorber layer. Table 1. Fitting parameters of the MAPb(I0.87Br0.13)3 perovskite solar cells.

Equation

⎛ t ⎞n I = I0 ⎜ ⎟ + Ii ⎝ t0 ⎠

⎧ ⎛ t - t0 ⎞ p ⎫ I = I0 exp ⎨-⎜ ⎟ ⎬ ⎩ ⎝ t0 ⎠ ⎭

S1

I0_S1

571.91 pA

nS1

1.10

S2

t0_S1 I0_S2

94.63 h 285.30 pA

pS1 nS2

1.40 0.90

t0_S2

41.58 h

pS2

0.62

These GBs enhance the minority carrier collection and provide the current path for the minority carriers, leading to a more effective charge transport and suppressing recombination. Besides, the increase of the negative surface potential ratio indicates the upward potential bending at the IGs. After 144 h, the positive surface potential distributions of both samples are different. The negative surface potential distribution of the S2 9

Nanotechnology 28 (2017) 315402

B P Nguyen et al

Figure 8. Statistical characterization of the local current focusing on the grain boundaries and the intragrains based on the line-profile results

of the many regions of the grain boundaries and the intragrains of the perovskite films with the S1 sample (grain size ∼300 nm) (a)–(e) and the S2 sample (grain size ∼100 nm) (f)–(j). Note: scale bar is 0–200 pA in (e), (i), and (j). 10

Nanotechnology 28 (2017) 315402

B P Nguyen et al

Figure 9. The energy level diagram: (a) before degradation, (b) in the second-step degradation.

[27, 42]. Leguy et al [16] proposed the hydration of the MAPbI3 that leads to a separation of the grains, which increased the charge recombination at the grain interface. Besides, Li et al [17] demonstrated a reversibility between the hydrate and perovskite phases during the degradation process. Clearly, in figure 6, the roughness of the thin films of the two samples increased as a function of the time. The change in the topography with the time corresponds to the conversion from the cubic MAPb(I0.87Br0.13)3 phase to the hexagonal PbI2 phase [37]; this observation is consistent with the penetration of moisture into the perovskite film along the GBs. The large grain size within a perovskite thin film acts to form an air barrier [39], resulting in a slower conversion to the MAPbI3 hydrate in the S1 sample compared with the S2 sample. After the 310 h exposure, an increase in the intensity of the peak attributed to the PbI2 phase in the film is such that the local currents in both films are almost eliminated; therefore, the local current could be influenced by the decomposed PbI2 phase. Figure 7(b) displays a mathematical fit to the current distribution to help explain the dependency of the perovskite solar cell degradation process on the absorber layer thickness. I0 is the maximum average current value and t0 is the time at the maximum average current. The following equation was applied to the increased-current region: I = I0

t n t0

()

+ Ii ,

where n is the ionic migration constant. In the region of the low current, the following exponential function was applied: t - t0 p t0

{ ( ) },

I = I0 exp -

(3 )

where p is the decay constant [43, 44]. The detailed current fitting results are shown in table 1. Based on the present current fitting results, the degradation mechanism occurs in two steps. The increased-current region may be due to the enhancement of the carrier separation and the suppression of the charge recombination at the GBs. After that, the current sharply decreased; due to this, the perovskite layer can be considered as acting as a current-blocking layer through its division into the major PbI2. To confirm this hypothesis, the statistical characterization of the local current focus on the GBs and IGs were collected, as shown in figure 8. The histograms of the local current distribution are indicated at the GBs or the IGs based on the results of the line profiles in many regions in each sample. Clearly, the local current flow at the GBs, which is obtained at the microscopic level, is enhanced after the absorption of the water molecules in the short time. The ion migration causes the local current flow in the MAPb(I1−xBrx)3. The active energy (EA) of MAPb(I1−xBrx)3 is 0.27 eV, which is close to that of pure halide perovskite [43]. The enhanced ion migration resulted in the redistribution of the ions along the GBs in our recent research [35] and this was also observed in previous research [30, 44]. In fact, the ion migration device efficiency ion in perovskite solar cells actually improves the

(2 )

11

Nanotechnology 28 (2017) 315402

B P Nguyen et al

References

device efficiency rather than reducing the device efficiency [13, 15, 29, 39]. Therefore, the GBs are beneficial to the solar cell performance, but they are harmful to the long term stability. The more GB formation causes the easy penetration of the water molecule into the perovskite layer. The rapid formation of the PbI2 phase blocks the electron transport due to its wide bandgap, leading to the quenching of the local current flow. Recently, Hu et al [45] demonstrated that the efficiency and stability of MAPb(I1−xBrx)3 are also increased under illumination by the reduced GB areas. Figure 9 shows the energy band diagram across the GBs before and after the degradation. At the pre-degradation, the local current flows through the grains of the MAPb(I1−xBrx)3 thin films, however, the downward band bending in the energy band diagram lead to the electrons to be attracted toward the GBs, whereas the majority carriers (holes) are repelled. At the first step of degradation, after absorbing the water molecules the perovskite material produces the band bending around the GBs and the IGs regions. The upward and downward band bending at the IGs and GBs help the separation of the excitons and the charge transportation under humidity exposure. The higher local current flows through the GBs when the applied voltage overcomes the barrier created by the band bending, indicating that the GBs effectively act as either charge dissociation sites or local current pathways. At the second step degradation, the S1 sample slowly degraded into the PbI2 phase when compared with the quickly degraded S2 sample due to more GB formation, the charges are trapped along the GBs due to the PbI2-phase formation at the IGs. Through the reduction of the GB formation and the improved crystallinity, the MAPb(I1−xBrx)3 device is more stable under illumination. In the present work, an enhancement of the stability of perovskite under humidity is achieved through the control of the grain size. Further study is needed to determine the role of GBs in terms of the perovskite stability under high temperatures.

[1] Park N–G 2015 Mater. Today 18 65–72 [2] Wang B, Xiao X and Chen T 2014 Nanoscale 16 12287–97 [3] Green M A, Ho-Baillie A and Snaith H J 2014 Nat. Photon. 8 506–14 [4] Jung H S and Park N-G 2015 Small 11 10–25 [5] Kojima A, Teshima K, Shirai Y and Miyasaka T 2009 J. Am. Chem. Soc. 131 6050–1 [6] Green M A, Emery K, Hishikawa Y, Warta W and Dunlop E D 2016 Prog. Photovolt., Res. Appl. 24 905–13 [7] He M, Zheng D, Wang M, Lin C and Lin Z 2014 J. Mater. Chem. A 2 5994–6003 [8] Conings B, Baeten L, De Dobbelaere C, D’Haen J, Manca J and Boyen H-G 2014 Adv. Mater. 26 2041–6 [9] Kim B J et al 2015 Energy Environ. Sci. 8 916–21 [10] Rajamanickam N, Kumari S, Vendra V K, Lavery B W, Spurgeon J, Druffel T and Sunkara M K 2016 Nanotechnology 27 235404 [11] Niu G, Guo X and Wang L 2015 J. Mater. Chem. A 3 8970–80 [12] Frost J M, Butler K T, Brivio F, Hendon C H, Schilfgaarde M V and Walsh A 2014 Nano Lett. 14 2584–90 [13] Christians J A, Herrera P A M and Kamat P V 2015 J. Am. Chem. Soc. 137 1530–8 [14] Yang J, Siempelkamp B D, Liu D and Kelly T L 2015 ACS Nano 9 1955–63 [15] Noh J H, Im S H, Heo J H, Mandal T N and Seok S I 2013 Nano Lett. 13 1764–9 [16] Leguy A M A et al 2015 Chem. Mater. 27 3397−407 [17] Li D et al 2016 J. Phys. Chem. C 120 6363–8 [18] Heo J H et al 2013 Nat. Photon. 7 486–91 [19] Berger R, Domanski A L and Weber S A L 2013 Eur. Polym. J. 49 1907–15 [20] Dante M, Peet J and Nguyen T-Q 2008 J. Phys. Chem. C 112 7241–9 [21] Hoppe H, Glatzel T, Niggemann M, Hinsch A, Lux-Steiner M C and Sariciftci N S 2005 Nano Lett. 5 269–74 [22] Jiang C–S et al 2015 Nat. Commun. 6 8397 [23] Yun J S, Ho-Baillie A, Huang S, Woo S H, Heo Y, Seidel J, Huang F, Cheng Y–B and Green M A 2015 J. Phys. Chem. Lett. 6 875–80 [24] Xu X et al 2015 Nano Lett. 15 6514–20 [25] Coffey D C, Reid O G, Rodovsky D B, Bartholomew G P and Ginger D S 2007 Nano Lett. 7 738–44 [26] Ahn N, Kwak K, Jang M S, Yoon H, Lee B Y, Lee J-K, Pikhitsa P V, Byun J and Choi M 2016 Nat. Commun. 7 13422 [27] Liu D, Gangishetty M K and Kelly T L 2014 J. Mater. Chem. A 2 19873–81 [28] Li J-J, Ma J-Y, Ge Q-Q, Hu J-S, Wang D and Wan L-J 2015 ACS. Appl. Mater. Interfaces 7 28518–23 [29] Christians J A, Manser J S and Kamat P V 2015 J. Phys. Chem. Lett. 6 2086–95 [30] Shao Y et al 2016 Energy Environ. Sci. 9 1752–9 [31] Bass K, McAnally R E, Zhou S, Djurovich P I, Thompson M E and Melot B C 2014 Chem. Commun. 50 15819–22 [32] Dymshits A, Henning A, Segev G, Rosenwaks Y and Etgar L 2015 Sci. Rep. 5 8704 [33] Jiang M, Wu J, Lan F, Tao Q, Gao D and Li G 2015 J. Mater. Chem. A 3 963–7 [34] Misra R K, Aharon S, Li B, Mogilyansky D, Visoly-Fisher I, Etgar L and Katz E A 2015 J. Phys. Chem. Lett. 6 326–30 [35] Kim D, Kim G Y, Ko C, Pae S R, Lee Y S, Gunawan O, Ogletree D F, Jo W and Shin B 2016 J. Phys. Chem. C 120 21330–5

4. Conclusion In summary, the electrical degradation of perovskite devices was investigated in the dark under an ambient condition using KPFM and C-AFM measurements. The perovskite with the large particle size coverage on the mesoporous TiO2 surface exhibited a slow degradation compared with the small size coverage. The GBs are both beneficial and harmful to the long-term stability of perovskite solar cells. These results help in the elucidation of the fundamental decomposition pathways in the perovskite thin films, which in turn are expected to lead to the development of more stable materials, and perhaps onward to the design of commercially viable devices.

Acknowledgments This work was supported by a National Research Foundation of Korea (NRF) grant funded by the Korean Government (MSIP) (No. 2015001948). 12

Nanotechnology 28 (2017) 315402

B P Nguyen et al

[42] Chen Q, Zhou H, Song T-B, Luo S, Hong Z, Duan H-S, Dou L, Liu Y and Yang Y 2014 Nano Lett. 14 4158–63 [43] Hoke E T, Slotcavage D J, Dohner E R, Bowring A R, Karunadasa H I and McGehee M D 2015 Chem. Sci. 6 613–7 [44] Yuan Y and Huang J 2016 Acc. Chem. Res. 49 286–93 [45] Hu M, Bi C, Yuan Y, Bai Y and Huang Y 2016 Adv. Sci. 3 1500301 [46] Kim G Y, Oh S H, Nguyen B P, Jo W, Kim B J, Lee D G and Jung H S 2015 J. Phys. Chem. Lett. 6 2355–62 [47] Sadewasser S and Glatzel T 2011 Kelvin Probe Force Microscopy: Measuring and Compensating Electrostatic Forces (New York: Springer)

[36] Kim G Y, Kim J R, Jo W, Lee K D, Kim J Y, Nguyen T T and Yoon S 2014 Curr. Appl. Phys. 14 1665–8 [37] Coning B et al 2015 Adv. Energy Mater. 5 1500477 [38] Cao D H, Stoumpos C C, Malliakas C D, Katz M J, Farha O K, Hupp J T and Kanatzidis M G 2014 APL Mater. 2 091101 [39] Kim J H, Williams S T, Cho N, Chueh C-C and Jen A K-Y 2015 Adv. Energy Mater. 5 1401229 [40] Han Y, Meyer S, Dkhissi Y, Weber K, Pringle J M, Bach U, Spiccia L and Cheng Y-B 2015 J. Mater. Chem. A 3 8139–47 [41] Li J–J, Ma J–Y, Ge Q–Q, Hu J–S, Wang D and Wan L–J 2015 ACS Appl. Mater. Interfaces 7 25818–8523

13