SKPM study on organic-inorganic perovskite materials

We report Atomic Force Microscopy (AFM) and Scanning Kelvin Probe Microscopy (SKPM) studies on the surface morphology and surface potential properties of CH3NH3PbI3, CH3NH3PbI3-xClx, CH3NH3PbI3-xBrx and CH3NH3PbBr3-xClx, respectively. For CH3NH3PbI3 rod structure, its surface potential is independent of the precursor concentration, suggesting a robust electronic feature. Surface potential studies of CH3NH3PbI3 particle reveal that the Fermi level within CH3NH3PbI3 is strongly influenced by the substrate. In the case of CH3NH3PbI3-xClx, its surface potential depends on precursor concentrations and we suspect that chlorine concentrated solutions might lead to more chlorine incorporation in the final products, thus lowering its Fermi level. Also, we studied the surface potentials of CH3NH3PbI3-xBrx and CH3NH3PbBr3-xClxwith specified halide ratios. The surface potential differences between different samples are related to their work function variations. These results are helpful to the understanding of the st...


I. INTRODUCTION
Since their first introduction in solar cells in 2009, 1 methylammonium lead halide perovskite materials (e.g., CH 3 NH 3 PbX 3 ,X = Cl, Br, I) have attracted much attention in photovoltaic research community. [2][3][4][5][6][7][8][9] The power conversion efficiency of perovskite based solar cells has already surpassed 20% in a few years. 10 Various investigations have revealed their great potential as a prospective candidate for photovoltaic application, featuring with favorable bandgap, high absorption coefficient, high carrier mobility and long electron/hole diffusion length. [11][12][13][14] For iodide/chloride mixed perovskite CH 3 NH 3 PbI 3-x Cl x , a variety of experiments are conducted to investigate the role of chlorine element. [15][16][17][18][19][20][21][22][23][24][25] It shows that Cl ions in the precursor solution have a profound and beneficial effect on the film morphology, 15,16 carrier transport, 13 and stability, 17,26 even though Cl remains at a very low content in the perovskite lattice. 24,25,27,28 For iodide/bromide and chloride/bromide mixed perovskites, their optical and electronic properties can be tuned by changing the halide ratios. [29][30][31] These perovskites are mostly wide bandgap and are not suitable for photovoltaic applications due to their transparency to most of the solar spectrum. However, recent studies have shown promising prospects for their use in light emission devices whose spectral response can be chemically tuned across the entire visible spectrum. 30,31 Despite the fast development for perovskite materials, more studies are needed to clearly clarify their structure-property relations. Scanning Kelvin Probe Microscopy (SKPM) is a surface potential detection technique based on AFM. It measures the surface potential or contact potential difference (V CPD ) by compensating the electrostatic forces between the microscopic probe and sample. The contact potential difference is generally determined by the effective work function of two materials and the electrostatic potential difference. The former depends on the surface properties of the two materials, and the latter is determined by the bias and surface charges of the sample. 32 SKPM is widely used in multiple disciplines to study the morphology and surface electronic properties of materials simultaneously. [33][34][35] It also has been used for perovskite films to study grain boundary (GB) effects. [36][37][38][39][40] Their SKPM studies revealed a downward band bending in the energy band diagram at the GBs which act as effective charge dissociation interfaces and photocurrent transduction pathways, enhance minority carrier collection and boost the overall performance of the device. However, as SKPM directly gives the local work function/Fermi level of materials, or at least shows the relative energy levels with careful controls on experimental conditions in air, it provides a useful approach to obtain insights on the electronic properties of perovskite and the interaction between perovskite materials and substrates.
In this work, we systematically conduct AFM and SKPM experiments to investigate the morphology and surface potential of CH 3 NH 3 PbI 3 , CH 3 NH 3 PbI 3-x Cl x , CH 3 NH 3 PbI 3-x Br x and CH 3 NH 3 PbBr 3-x Cl x , respectively. For triiodide perovskite, two different structures are formed. The rod structure is made by single-step pre-mixed precursor deposition, particle structure by two-step deposition. The concentration effect and substrate effect on the energy barrier at the perovskite/substrate interfaces are studied. Then by studying CH 3 NH 3 PbI 3-x Cl x perovskite films deposited from three precursor concentrations, we establish a connection between its Fermi level and chlorine effect. Also, we studied the surface potential of I/Br and Br/Cl mixed perovskites with specified ratios and the surface potential is correlated with their work functions.

A. Morphology and surface potential study of CH 3 NH 3 PbI 3
We first choose to investigate the most widely studied perovskite CH 3 NH 3 PbI 3 . 1 Structures are both formed by solution approaches. The common rod structure is made by one-step pre-mixed precursor deposition, and the particle structure by two-step deposition. Their typical topography images are shown in Figures 1(a) and 1(b), respectively. Four precursor solutions with different concentrations are used to obtain perovskite rods with micrometer-long sizes. Increasing concentration from 4.8 to 29 wt%, the rod length grows from ∼4.5 µm to ∼20 µm and its average height increases from 5.8 nm to 211.7 nm. For particle perovskite, FTO, PEDOT:PSS/FTO and PEDOT:PSS/glass are used as substrates. The size distributions of obtained rods and particles are summarized in Table I. More topography images can be found in the supplementary material( Figure S1, S4).
Figures 1(g) and 1(h) shows the absorption spectra and PL spectra of both perovskite particles and rods, respectively. Perovskite particles show more pronounced absorption peaks around 420 nm and 500 nm. The absorption spectra of the both structures display a distinct absorption onset around the energy band gap (∼760 nm), indicating the formation of perovskite. This is also confirmed by their steady state PL spectra, which show emission peaks around 764 nm. A slight shift in the emission peak (∼756 nm) is observed for rod perovskite.
In SKPM measurements, the surface potential is determined by the effective work function of both the probe and sample. For meaningful comparison, a same potential reference is needed. In this section, we choose the substrate potential as reference point as different probes are used in these experiments. To semi-quantitatively discuss this problem, we choose a simple Mott-Schottky (MS) model for inorganic semiconductor/metal contact. 41 Owing to work function difference between the substrate and perovskite semiconductor, electrons are free to move throughout the interface, reaching a thermal equilibrium, leading to Fermi-level alignment and band-bending phenomena at the metal/semiconductor contact. 35 The V CPD , which coincides to the difference of work function between the substrate and the semiconducting solid, reflects an energy barrier that a hole needs to overcome in order to be transferred from perovskite to substrate at the interface. 36,41 The energy level alignment diagram at the CH 3 NH 3 PbI 3 /substrate interfaces is depicted in Figure 2.
Typical potential images for rod and particle structures are shown in Figures 1(d) and 1(e), respectively. Quantitative analyses are summarized in Table I. For rod perovskite deposited on n-type FTO substrate, it shows that when precursor concentration is decreased from 29 wt% to 4.8 wt%, the surface potential shows a little variations around 45 meV. There is no obvious FIG. 1. Topography, surface potential and optical properties of perovskites. (a-c) AFM topography images of rod CH 3 NH 3 PbI 3 , particle CH 3 NH 3 PbI 3 and CH 3 NH 3 PbI 3-x Cl x deposited on FTO substrate, respectively. (d-f) Corresponding contact potential difference obtained by SKPM. (g) Similar absorption spectra both display a distinct absorption onset around 760 nm, indicating the formation of final perovskites. (h) PL spectra of rod CH 3 NH 3 PbI 3 and CH 3 NH 3 PbI 3-x Cl x share a nearly identical emission peak around 750 nm. A slight shift in the emission peak (∼756 nm) is observed for rod CH 3 NH 3 PbI 3 .
connections between the surface potential and the rod size. This suggests that the electronic properties of this rod structure perovskite is robust or that CH 3 NH 3 PbI 3 is defect tolerance as small variations in the processing/reaction could lead to an incomplete conversion or structural heterogeneity across the bulk film, resulting defects or different phases in perovskite films. The topography and potential images of these rods made from these precursor solutions are shown in supplementary material ( Figure S1, S2, S3). The surface potential of particle perovskite depends on the substrates used. On the n-type FTO substrate, SKPM measurement exhibits an energy barrier around 11 meV. When another p-type PEDOT:PSS film is spin coated on FTO or glass substrates before the perovskite deposition, the barrier is lowered to around 0 ∼4 meV. These results agree with literature which reveals that the Fermi level within CH 3 NH 3 PbI 3 is strongly influenced by the substrate. 42 The n-type substrates (FTO used here) yield strongly n-type surfaces, while the p-type substrates (PEDOT:PSS used here) result in perovskite surfaces that are more p-type like. The small potential difference at perovskite/PEDOT:PSS interface indicates that their Fermi levels are very close before reaching equilibrium as depicted in Figure 2.

B. Morphology and surface potential study of CH 3 NH 3 PbI 3-x Cl x
Iodide/chloride mixed perovskite (CH 3 NH 3 PbI 3-x Cl x ) is another widely used perovskite. It possesses the same basic crystalline structure and similar electronic structure as its CH 3 NH 3 PbI 3 analogue. 28,43 However, its carrier lifetime and diffusion length are found to increase by more than an order of magnitude. 13 For example, experiment shows that electron-hole diffusion lengths in CH 3 NH 3 PbI 3-x Cl x exceed 1 µm compared with ∼100 nm in CH 3 NH 3 PbI 3 . In this section, we study the film morphology and surface potential of CH 3 NH 3 PbI 3-x Cl x by AFM and SKPM, respectively. The CH 3 NH 3 PbI 3-x Cl x films are deposited by spin coating a precursor solution with respective 32.9 wt%, 24.7 wt% and 7.6 wt% concentrations. Its typical topography and surface potential images are compared with CH 3 NH 3 PbI 3 rod in Figure 1. More topography and potential images can be found in supplementary material ( Figure S7). Figures 1(g) and 1(h) shows that these two perovskites share nearly identical absorption and PL spectra. However, their topography images are different. Instead of the rod structure made by one-step method for CH 3 NH 3 PbI 3 , the structure of CH 3 NH 3 PbI 3-x Cl x is mainly isolated fused-like crystalline islands with several micrometer in size at least. It is caused by their different crystal growth kinetics. It is reported that the PbCl 2 addition affected perovskite crystal formation kinetics by acting as heterogeneous nucleation sites due to its limited solubility in DMF. 44 The surface potential of CH 3 NH 3 PbI 3-x Cl x are also different compared with CH 3 NH 3 PbI 3 . From the SKPM measurements, its surface potential depends on the concentrations of the precursor used. The surface potential data are summarized in Table I. Its surface potential decreases from 175 meV for 32.9 wt% solution to 71 meV and 24 meV for 24.7 wt% and 7.6 wt% solution, respectively. This indicates that its Fermi level is lowered with the increasing precursor concentrations.
FIG. 2. Energy diagram of CH 3 NH 3 PbI 3 on n-type or p-type substrates before and after contact. (a) N-type substrates (FTO) yield strongly n-type CH 3 NH 3 PbI 3 . (b) P-type substrates (PEDOT:PSS) result in p-type like CH 3 NH 3 PbI 3 . After contact, electrons are free to move across the interface, leading to Fermi-level alignment and band-bending at the metal/semiconductor contact according to Mott-Schottky model. By comparing their topography and potential line profiles in Figure S9 in supplementary material, the potential variation is not following its height variation. Thus, the contribution of the substrate to the potential data is neglected.
Here, we speculate that the dependence of surface potential on the precursor concentrations might be related to the presence of chloride in perovskite films. It is expected that more chlorine concentrated solutions might lead to more chlorine incorporation although the content is very low, thus lowering its Fermi level. In a previous study about As-doped CdTe, SKPM experiments revealed that increasing As concentrations in CdTe led to a decrease in V CPD , a downward shift in the CdTe Fermi level and an increase in the CdTe work function. 45 We would expect the same case here. It is also consistent with a recent literature about CH 3 NH 3 PbI 3 (Cl)-based perovskite solar cells which found that chlorine incorporation decreased the V CPD compared with CH 3 NH 3 PbI 3 , resulting a lower Fermi level. 43 XPS and UPS experiments also confirmed that CH 3 NH 3 PbI 3-x Cl x has a slight higher work function than CH 3 NH 3 PbI 3 . 46 More topography and surface potential can be found in the supplementary material ( Figure S7, S8, S9).

C. Morphology and optical properties of I/Br mixed perovskite films
For CH 3 NH 3 PbI 3-x Cl x perovskite, the chlorine element is believed to be at a very low concentration. 24,25,27,28 It is caused by the limited chloride solubility in the iodide derivative due to the large difference in the ionic radii of ClandI anions. 24 However, for I/Br and Cl/Br mixed perovskites, the composition can be widely tuned from 0 ∼1 by varying the halide ratio in the precursor solution. [29][30][31] Similar conclusions were drawn for other halide compounds, such as CH 3 NH 3 SnX 3 , 47 CH 3 NH 3 I 1-x Br x and CH 3 NH 3 Br 1-x Cl x ( Figure S10 in supplementary material).
I/Br mixed perovskites CH 3 NH 3 Pb(I 1-x Br x ) 3 are deposited by spin coating precursors with specified CH 3 NH 3 I/CH 3 NH 3 Br ratios mixed with PbAc 2 in DMF solvent. As the halide element comes only from the CH 3 NH 3 I/CH 3 NH 3 Br mixtures, Pb source from PbAc 2 , the final perovskite composition are more easily controlled. Studies also show that the PbAc 2 are facilitating the removal of the by-products during crystallization, thus obtaining more continuous films with less pinholes. 48 Figures 3(a)-3(f) shows topography images of CH 3 NH 3 Pb(I 1-x Br x ) 3 on FTO substrate with I/Br ratios of 0/1, 0.2/0.8, 0.4/0.6, 0.6/0.4, 0.8/0.2, 1/0,respectively, for areas of 30 µm*30 µm. It shows a little variation of grain size between different I/Br ratio mixed perovskite. The 0.4/0.6 ratio film has a more compact structure than the other ratio films which have less connections between grains. The 0/1 and 0.2/0.8 ratio films both consisted by more separated grains. And the 0.8/0.2, 1/0 ratio film have a very similar network structure. One thing to note is that the voids between grains may be also covered by the perovskite materials as the void surface is different from the bare FTO substrate. Figures 3(g) and 3(h) shows the corresponding photographs and absorption spectra of CH 3 NH 3 Pb(I 1-x Br x ) 3 with specified I/Br ratios respectively. Through the ratio control of CH 3 NH 3 I/CH 3 NH 3 Br, the film colors are tuned from dark brown for CH 3 Figure 4 shows the enlarged surface topography and the corresponding surface potential maps of perovskite films with I/Br ratio of 0/1, 0.2/0.8, 0.4/0.6, 0.6/0.4, 0.8/0.2, 1/0, respectively for areas of 5 µm*5 µm. As the same probe is used for this group of samples, the surface potential can be directly compared. In this case, the V CPD is defined as the effective work function difference between the tip and sample. The average surface potential of different I/Br ratios mixed films are summarized in Table II. The potential increases from -172 meV for CH 3 NH 3 PbBr 3 (I/Br=0/1) to 31 meV (I/Br=0.6/0.4). Further increasing the I/Br ratio to 1/0, however, the potential decreases to -123 meV. The surface potential of CH 3 NH 3 Pb(Br 1-x Cl x ) 3 with Cl/Br ratios of 0/1, 0.2/0.8, 0.4/0.6, 0.6/0.4, 0.8/0.2, 1/0, are also studied here. The corresponding AFM topography and SKPM surface potential images, along with their optical absorption spectra are shown in the supplementary material ( Figure S11, S12, S13, S14). Their potential data are summarized in Figure 5. The work functions of CH 3 NH 3 Pb(Br 1-x Cl x ) 3 from literature decreases first and then increases, with increasing the ratio of Cl/Br. 50 Similar trend is seen in the average potential of CH 3 NH 3 Pb(Br 1-x Cl x ) 3 while quite large fluctuations also exist. As in amplitude modulation SKPM, both the tip apex and the cantilever would contribute to the measured V CPD . 51 Also, the experiments are conducted in ambient conditions. Thus, the measured V CPD differences among samples do not fully correspond to their work function differences. 52,53 Not only the average work function can be obtained from the SKPM measurements, the local inhomogeneities in materials can also be analysed from its potential images. For CH 3 NH 3 Pb(I 1-x Br x ) 3 , work function variations appear for both samples and are most pronounced in 0/1 and 0.2/0.8 mixed films. They are highlighted in the potential images in Figure 6. In the 0/1 ratio mixed perovskite, area 1 and area 3 with similar potentials both appear on the grain boundary. The potential difference between area 1 (about -190 meV) and area 2 (about -227 meV) is 37 meV. These grain-to-grain variations appear to depend on the physical orientation of each grain and its associated crystallographic face. In 0.2/0.8 ratio mixed perovskite, the interior grain area 1 (about -172 meV) has a lower potential than edge area 2 (about -133 meV) and the potential difference is 40 meV. Regarding the reasons of these local work function shift, it might be attributed to local contaminants, such as chemisorbed or physisorbed H 2 O molecules, solvent residues etc., resulting in a decrease of the local vacuum energy and the electron affinity. 54 However, as the experiments are conducted in similar conditions and the inhomogeneities are most pronounced in 0/1 and 0.2/0.8 mixed films, we believe that the variations are most likely attributed to compositional inhomogeneities. Previous studies show that for I/Br mixed halide systems, there is compositional inhomogeneity. 55

III. CONCLUSION
In summary, we have studied the morphology and surface potential of tri-iodine perovskite, I/Br mixed perovskite and Br/Cl mixed perovskite, respectively. For tri-iodide perovskite, two different structures are formed. The SKPM measurement reveals that the energy barrier at rod/FTO interface is independent of the precursor concentrations used. The potential of particle perovskite is directly correlated with the type of substrate used. The n-type FTO substrate yield an n-type surface, while the p-type PEDOT:PSS film results in perovskite surfaces that are more p-type with a lower energy barrier. For chloride mixed perovskite, measurements show the energy barrier is decreased from 175 meV to 71 meV and 24 meV for 32.9 wt%, 24.7 wt% and 7.6 wt%, respectively. It is expected that chlorine concentrated solutions might lead to more chlorine incorporation, thus lowering its Fermi level. Also, we studied the surface potential of I/Br and Br/Cl mixed perovskites with specified ratios. The surface potential trend between samples is directly related to their work functions. We believe these studies are helpful to understand the structural, optical and electronic properties of perovskite materials.

Methylammonium halide synthesis
Methylammonium halide (CH 3 NH 3 I,CH 3 NH 3 Br,CH 3 NH 3 Cl) was synthesized by the reaction of methylamine aqueous solution and the corresponding hydrohalide (HCl,HBr,HI) acid in a beaker at room temperature for 2 h with constant stirring. The precipitate was recovered by carefully evaporating the solvents at 70 • C. The raw product CH 3 NH 3 X was then rinsed with diethyl ether and dried at 60 • C in a vacuum oven overnight.
Methylammonium mix halides (CH 3 NH 3 I 1-x Br x and CH 3 NH 3 I 1-x Cl x ) were synthesized by evaporating the specified ratio of CH 3 NH 3 I(CH 3 NH 3 Cl) and CH 3 NH 3 Br in ethanol solution at 70 • C. The raw product was rinsed with diethyl ether and dried at 60 • C in a vacuum oven overnight.

B. Precursor preparation and film formation
Before the film deposition, FTO coated glass substrates and glass substrates were cleaned with DI water, acetone and ethanol, for 10 min successively in ultra sonicator. Then the substrates were rinsed by DI water and dried with clean air.
To deposit tri-iodide perovskite rod, CH 3 NH 3 I powder and PbI 2 were dissolved in DMF at 1: 1 molar ratio with required concentrations. The obtained solution was stirred overnight before use. Then the precursor was spin coated on FTO substrate at the rotation speed of 3000 rpm for 40 s. The film was finally annealed at 100 • C for 30 min on the hotplate.
To deposit tri-iodide perovskite particle, CH 3 NH 3 I powder and PbI 2 were dissolved in IPA and DMF with a concentration of 30 mg/ml and 100 mg/ml, respectively. For FTO substrate based sample, hot PbI 2 precursor solution (70 • C) was directly spin coated on FTO surface at a rotation speed of 2000 rpm for 40 s without further annealing. For PEDOT:PSS/FTO or PEDOT:PSS/glass substrate based samples, PEDOT:PSS solution was first spin coated on the FTO or glass substrate at a rotation speed of 2000 rpm for 40 s, followed by annealing at 130 • C for 15 min. Then the hot PbI 2 precursor solution (70 • C) was spin coated on the PEDOT:PSS surface at a rotation speed of 2000 rpm for 40 s. After that, the prepared CH 3 NH 3 I solutions were spin coated on FTO, PEDOT:PSS/FTO and PEDOT:PSS/glass substrates at a rotation speed of 2000 rpm for 40 s. During another spin process, several IPA drops were dripped on the films to clean extra CH 3 NH 3 I solutions. The obtained films were finally annealed at 100 • C for 30 min.
For chloride mixed perovskite, CH 3 NH 3 I powder and PbCl 2 were dissolved in DMF at 3: 1 molar ratio with required concentrations. The obtained solution was stirred overnight before use. Then the precursor was spin coated on FTO substrate at a rotation speed of 3000 rpm for 40 s. The film was finally annealed at 100 • C for 45 min.
To prepare I/Br and Cl/Br mixed perovskite films, specified CH 3 NH 3 I(or CH 3 NH 3 Cl)/CH 3 NH 3 Br ratios and PbAc 2 were dissolved in DMF at 3:1 molar ratio at ∼40 wt% concentrations. The obtained solutions were stirred overnight before use. Then the precursors were spin coated on FTO substrate at a rotation speed of 3000 rpm for 40 s. The films were finally annealed at 100 • C for 10 min.

C. UV-Vis characterization
A UV-3600 spectrophotometer (Shimadzu, Japan) was used for the absorption spectra of samples. PL spectra were recorded by a commercial Micro Raman spectrometer (Horiba), using an excitation laser beam with a wavelength of 532 nm.

D. Morphology and surface potential characterization
SKPM is implemented on an AFM platform (MFP-3D-BIO, Asylum Research). The measurement is carried out in a two-pass technique. The first pass is conducted as a standard AC mode scan and is used to record the surface topography. In the second pass, the tip is then retraced from the first topography scan with an additional z distance of 10 ∼50 nm. An AC voltage at a frequency of 60-65 kHz and a tunable DC voltage are applied between the tip and the substrate in the second pass. The DC voltage is adjusted at each point to cancel the force at the AC frequency, and accordingly the value is recorded as the potential difference. In this article, Pt/Cr coated silicon tips (Multi75E-G, Budget sensor) are used. It has an oscillation frequency of 60 kHz, spring constant of 1-5 N/m, and a quality factor Q of ∼186. All experiments are conducted in ambient conditions, in dark.

SUPPLEMENTARY MATERIAL
See supplementary material for the detailed data.

ACKNOWLEDGMENTS
This work was supported by the National Natural Science Foundation of China (Grant No. 11474345, No. 11674043).