Research Article J. Electrochem. Sci. Technol.

, 2022, 13(1), 71-77

Ion Migration in Metal Halide Perovskites

Anafi Nur'aini2,3, Seokwon Lee1, and Ilwhan Oh1,3*
1
Departments of Applied Chemistry, Kumoh National Institute of Technology, Gumi, Gyeongbuk, 39177 South Korea
2
Chemical Engineering, Kumoh National Institute of Technology, Gumi, Gyeongbuk, 39177 South Korea
3
Department of Energy Convergence Engineering, Kumoh National Institute of Technology, Gumi, Gyeongbuk, 39177 South Korea

ABSTRACT
Metal halide perovskites are promising photovoltaic materials, but they still have some issues that need to be solved. Hys-
teresis is a phenomenon that strongly is correlated with ion migration; thus, a fast, easy, and low-temperature method for
measuring ion migration is required. Through selective blocking, ion migration can be measured separately, apart from elec-
tron migration. In this study, ion migration in metal halide perovskites was measured using a vertical device. At different
temperatures, ionic activation energies were obtained for a range of perovskite compositions such as MAPbI , FAPbI , CsP- 3 3

bI , and MAPbBr . By comparing the measured ionic activation energies with the theoretical values, we conclude that
3 3

- -
among other possibilities, I is the migrating ion in MAPbI , FAPbI , CsPbI and Br is the migrating in MAPbBr .
3 3 3, 3

Keywords : Ion Migration, Activation Energy, Perovskite, Selective Blocking

Received : 28 January 2021, Accepted : 8 June 2021

1. Introduction that hysteresis is caused by scan speed [6], film fabri-

cation and crystallinity [8,9], charge trapping [10],
In 2009, for the first time, metal halide perovskite ferroelectricity [11], and ion migration [4,6,11].
film was introduced as a solar absorber into dye-sen- Because perovskite conducts both electrons and ions,
sitized solar cells. At that time, device efficiency was thus ion transport may become an issue. However,
only 3.8% [1]. Perovskite has been intensively devel- because the electron and ion conduction are mixed, it
oped into higher efficiency due to its high potential for is difficult to measure ion conduction quantitatively.
solar energy converter and easy fabrication. Recently, Several experimental techniques have been used to
perovskite has successfully achieved an efficiency of discover evidence of ion migration, such as photo
25.2% [2]. Even though perovskite can attain high effi- thermal-induced resonance microscopy (PTIR) [12],
ciency, its stability during long operation remains an Tubandt method [13], impedance spectroscopy [14],
issue [3,4]. Stability is one of the gold standards for time of flight secondary ion mass spectrometry
commercialization of photovoltaic devices. (TOF-SIMS) [15], electrochemical, and selective
Investigations into perovskite characteristics have blocking [16,17]. Selective blocking measurement is
shown some anomalies, such as slow photocurrent easily accomplished and can be conducted at low
response [5], switchable photocurrent [6], and hyster- temperatures. This method allows the separation of ion
esis [4,7]. Hysteresis is a common phenomenon conduction from total conductivity [16]. A constant
when the shape of the current-voltage (I-V) graph dif- current is applied to measure the steady-state of volt-
fers between forward and backward scans. This phe- age as a function of time; thus, electron and ion con-
nomenon leads to low performance and efficiency ductivity can be extracted [17]. Initially, both ions and
degradation [8,9]. Previous studies have identified electrons pass through the perovskite film. Then, the
number of passing ions is reduced until only electrons
*E-mail address: [email protected] directly pass through the perovskite film (Fig. 1a) [16].
DOI: https://doi.org/10.33961/jecst.2021.00136 In this study, ionic activation energies of the per-
This is an open-access article distributed under the terms of the Creative Commons
Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0)
ovskite films MAPbI3, FAPbI3, CsPbI3, and MAPb-
which permits unrestricted non-commercial use, distribution, and reproduction in any Br 3, were measured using the selective blocking
medium, provided the original work is properly cited.

− 71 −
72 Anafi Nur'aini et al. / J. Electrochem. Sci. Technol., 2022, 13(1), 71-77

Fig. 1. (a) Schematic diagram of selective blocking measurement, (b) Structure of vertical device used in this study.

method. Ionic activation energy is influenced by cess was performed in ambient air. On the top of
crystal structure, radius of ion, and charge of ion [4]. PEDOT: PSS layer, perovskite solution (MAPbI 3,
For MAPbI3, the migrating ions might be I -, Pb2+, MAPbBr3, FAPbI3, or CsPbI3) was spin-coated at
MA+, or H+ [4,18]. If cation MA+ is substituted into 1000 rpm for 40 s, ramp 5.0, and 5000 rpm for 20 s.
FA+ and Cs+, these ions may also migrate in the per- During spin-coating of MAPbI3, FAPbI3, CsPbI3, at
ovskite film. The ion migration in FAPbI3 can be sup- t = 64 s, 800 μL of toluene was dropped on the top of
pressed by FA+ , which has a larger diameter than perovskite film. For MAPbBr 3, the toluene was
MA+ [15]. Substitution of MA+ with monovalent cat- dropped at t = 57 s. Toluene acted as an antisolvent for
ion Cs+ can eliminate the effects of ion migration precursor; therefore, toluene drip facilitated fast crys-
enhancement due to light intensity in MAPbI3 [17]. tallization of perovskite film. The perovskite film was
Anion substitution of I- into Br- may provide another dried on the hot plate for 10 min at 100oC for MAPbI3,
insight because Br- is also predicted to play a role in 170oC for FAPbI3, 350oC for CsPbI3, and 75oC for
ion migration. Understanding ion migration makes it MAPbBr 3. The deposition process was conducted
possible to design a new strategy to improve the sta- under N2 gas inside a glove box. Finally, an 80 nm Au
bility of perovskites film. layer was deposited on the top of the perovskite layer
using a thermal evaporator at a rate of 1 Å/s.
2. Experimental
2.2 Sample characterization
2.1 Fabrication of vertical devices X-ray diffraction measurements (XRD) were car-
The structure of a vertical device is illustrated in ried out using Rigaku X-MAX from 10o-40o to exam-
Fig. 1b. This sandwich-like structure consisted of ine the structure of each perovskite material. The
indium tin oxide (ITO), poly-(3,4-ethylene dioxyth- vertical structure of each sample was confirmed by
iophene): poly(styrene sulfonic acid) (PEDOT: PSS), cross-section observation using Field Emission Scan-
a perovskite film, and gold. ITO glass was cut into ning Electron Microscope (FE-SEM) MAIA III TES-
1×1 inch. A distance of 7 mm from one of the edges CAN and JEOL JSM-6701F.
was etched using Zn powder and HCl solution to
remove the ITO layer. ITO glass was cleaned by son- 2.3 Selective blocking measurement
ication in acetone, isopropyl alcohol, and deionized Selective blocking measurements were performed
water for 10 min each. The cleaning process was con- by a four-point probe system with a Faraday cage and
tinued by oxygen plasma for 20 min to improve the a black shield. The black shield provided a dark con-
adhesion between the substrate and thin film. dition for the measurement because, in this condition,
PEDOT: PSS was filtered using 0.45 μm syringe ions migrate faster than electrons. The four-point
filter then spin-coated onto clean ITO at 4000 rpm for probes were connected to the AutoLab controller
15 s and ramp 5.0. PEDOT: PSS layer was dried on with the PGSTAT302N module. NOVA Metrohm
the hot plate for 15 min at 125oC. The deposition pro- Autolab B.V version 1.10.1.9 was used as software.
 Anafi Nur'aini et al. / J. Electrochem. Sci. Technol., 2022, 13(1), 71-77 73

During measurement, two probes were connected to peak (110) indicates strain in the crystal structure
the cell. Hot chuck MH4R and controller MST [23,24]. In Fig. 2d, peaks of (100) and (200) corre-
1000 H were used to control the temperature of the spond to MAPbBr3 with cubic formation [25].
sample. Chronopotentiometry (Δt >1 ms) was used to FE-SEM confirmed the vertical structure of the
characterize the sample in a 100 nA current flow for device. The devices were constructed like a sandwich
20 s. The voltage-time (V-t) curve from that measure- (ITO/PEDOT: PSS/perovskite/Au) with perovskite
ment was then used to calculate total resistance and as an active layer. FE-SEM images are presented in
electronic resistance. Finally, the resistivity and con- Fig. 3a-d.
ductivity of both electrons and ions can be obtained. Cross-section images show that all layers are pre-
sented and vertically stacked. The perovskite thick-
3. Results and Discussions ness was approximately 200 nm. Using the Scherrer
equation, average grain size can be calculated from
3.1 Characterization of materials and devices XRD peaks. The average grain size for MAPbI 3 is
X-ray diffraction measurements were conducted to 43 ± 3 nm. CsPbI 3 has a larger grain size of 51 ±
examine the perovskite structure. Each perovskite 4 nm. MAPbBr 3 has the largest average grain size
solution was deposited onto ITO or FTO glass with- (55 ± 4 nm), whereas FAPbI3 has the smallest aver-
out Au and PEDOT: PSS layer. The structure of each age grain size (29 ± 7 nm). These average grain sizes
perovskite is presented in Fig. 2a-d. Fig. 2a shows are much lower than other groups whose minimum
(110), (220), and (310) peaks corresponding to the grain sizes are approximately 300 nm [26,27].
formation of orthorhombic MAPbI3 [19,20]. Fig. 2b However, grain size is correlated to the grain
confirms the formation of FAPbI3, showing (101), boundaries. Small grain sizes lead to extensive grain
(202), and (211) peaks, which are assigned to trigonal boundaries vice versa. Because grain boundaries
α- FAPbI3 or black phase FAPbI3 [21,22]. Fig. 2c facilitate the ion migration pathway [4], ion migra-
shows maximum intensity for (100) and (200), indi- tion is easier in the film with small grain sizes [6].
cating the cubic black phase of CsPbI3, and the lower

Fig. 2. XRD patterns of (a) MAPbI3, (b) FAPbI3, (c) CsPbI3, and (d) MAPbBr3.
74 Anafi Nur'aini et al. / J. Electrochem. Sci. Technol., 2022, 13(1), 71-77

Fig. 3. FE-SEM cross sections of (a) MAPbI3, (b) FAPbI3, (c) CsPbI3, and (d) MAPbBr3.

3.2 Selective blocking mixed conduction of electrons and ions. At the cur-
Selective blocking measurements were performed rent onset, perovskite film exhibits a lower resistance
in the dark, and ionic conductivities were measured. because both ionic and electronic conduction occurs.
When a constant current was applied to the vertical However, as the ionic conduction is blocked at the
device, the potential difference between two termi- interface between the perovskite and metal electrode,
nals was measured as a function of time. The poten- the resistance is gradually increased and reaches a
tial difference was affected by electronic resistance steady-state level, where ionic conduction is com-
(Re) and ionic resistance (Rion), which can be written pletely blocked, and only electrons can flow. On the
as Eqn. (1) [13]: other hand, when the temperature is increased, the
resistance of the perovskite is decreased because
R e R ion
V = i ---------------------
- (1) charge carriers (electrons and ions) are thermally acti-
R e + R ion
vated. Using the Arrhenius relation between the con-
As ions keep migrating in one direction, the ions ductivity and temperature, the activation energy (Ea)
were depleted at one end and accumulated at another for conduction can be obtained as shown in Fig. 5a-d.
end. In this stage, only electrons migrated, which cor-
respond to i = ie and iion = 0 [13]. Thus, the voltage 3.3 Ionic activation energy
can be written as Eqn. (2) : Activation energy (Ea) is the minimum energy that
must be overcome by an ion to jump from its initial
V = iR e (2) position to nearby vacancy sites [26]. After calculat-
ing ionic conductivities at each temperature, these
By measuring the initial and final voltages, and values can be further processed to extract the E a
combining Eqn. (1) and (2), Re and Rion can be deter- value for each perovskite materials. Following
mined. Furthermore, the ionic (σion) and electronic Arrhenius dependence, ionic activation energy can be
conductivities (σe) can be calculated from Rion and Re. calculated using Eqn. (3) [17]:
In this study, each sample was measured at room Ea
temperature (20-40oC), which perovskite devices σ i T = σ 0 exp -------- (3)
kb T
operate typically.
Fig. 4a-d show the voltage-time (V-t) graphs from where σi is conductivity, σ0 is the pre-exponential
selective blocking measurements for various per- factor, Ea is the activation energy, kb is Boltzmann
ovskite materials, which show typical behavior of constant, and T is temperature. That equation can be
 Anafi Nur'aini et al. / J. Electrochem. Sci. Technol., 2022, 13(1), 71-77 75

Fig. 4. V-t graphs from selective blocking measurements for (a) MAPbI3, (b) FAPbI3, (c) CsPbI3, and (d) MAPbBr3.

Fig. 5. Arrhenius plots of ionic conductivities for (a) MAPbI3, (b) FAPbI3, (c) CsPbI3, and (d) MAPbBr3.
76 Anafi Nur'aini et al. / J. Electrochem. Sci. Technol., 2022, 13(1), 71-77

Table 1. Ionic activation energy (Ea) of various perovskite materials

Perovskite materials Ea from our experiments (eV) Ea from previous studies (eV) Ionic radius (nm)
MAPbI3 0.48 ± 0.01 0.43 [13], 0.45 [28], 0.58 [29] 0.220 (I-) 0.216 (MA+)
MAPbBr3 0.073 ± 0.004 0.25 [20], 0.09 [30] 0.196 (Br-)
FAPbI3 0.29 ± 0.01 0.6 [32], 0.48, 0.55 [33] 0.253 (FA+)
CsPbI3 0.38 ± 0.03 0.36 [34], 0.53 [35] 0.167 (Cs+)

rearranged into Eqn. (4): I- [30,31].

In this study, we also investigated cation substitu-
Ea
ln ( σ i T ) = ln ( σ o ) + – -------- (4) tion by substituting MA + with FA + and Cs + . The
kb T
experimental result showed that FAPbI3 (0.29 eV)
In Fig. 5 ln(σiT) versus 1/T is plotted, and a linear has lower ionic activation energy than MAPbI 3
E (0.48 eV). Noting that FA+ (0.253 nm) is larger than
fit is applied. The slope corresponds to – -----a- , from
kb MA+ (0.216 nm), the observed reduction of Ea value
which Ea value can be determined. The Ea value of for FAPbI3 can be explained by the fact that FAPbI3
MAPbI3, MAPbBr3, FAPbI3, and CsPbI3 are 0.49 eV, has a larger lattice constant, leading to a lower barrier
0.07 eV, 0.29 eV, and 0.39 eV, respectively. These for I migration. For CsPbI3, even though the ionic
values are presented in Table 1 and compared with radius of Cs + is smaller than MA+ and the smaller
other studies to analyze the possible migrating ions in crystal might increase the barrier for I- migration, the
the perovskite. measured Ea value is smaller than that for MAPbI3.
The measured Ea value for MAPbI3 (0.48 eV) is This result is not clearly understood yet. We expect
close to those from previous literature, which is that other factors than crystal lattice size might come
0.43 eV [13] and 0.45 eV [28]. According to a study into play, such as simultaneous cation (Cs+) migra-
using density functional theory (DFT) calculation, tion or ion migration along the grain boundary.
the activation energy of I - is 0.58 eV [29]. It is In this experiment, a perovskite film was fabricated
assumed that I- has the shortest pathway to I- vacan- by a solution process followed by annealing, which
cies through the edge of PbI6 octahedron. In contrast, allowed evaporation of solvent after spin coating. Con-
Pb2+ has the longest pathway because it needs to pass sequently, grain boundaries were generated on the sur-
along the diagonal direction <110>. Meanwhile, face, which might have facilitated ion migration. If the
MA+ migrates to the nearest V’MA cage of the unit ion migrates in the grain boundaries, the ionic activa-
cell face composed of four I ions. Both Pb2+ and MA+ tion energy is only half of the ionic activation energy
have longer paths than I - ; thus Pb + (2.31 eV) and from ion migration in bulk [4]. Therefore, the low
MA+ (0.84 eV) have higher activation energies than I- ionic activation energy of metal halide perovskite in
(0.58 eV) [29]. this experiment compared to others might have been
The Ea value for MAPbBr3 (0.073 eV) in Table 1 caused by the abundance of grain boundaries.
has important implications in identifying the migrat-
ing species in the perovskite materials. With the sub- 4. Conclusions
stitution of I- with Br-, MAPbBr3 has a much lower
Ea value than MAPbI3 (0.48 eV). The lowered Ea for Ion migration in metal halide perovskite was inves-
the anion –substituted perovskite indicates that the tigated by measuring ionic activation energy. The
migrating species might be the anion. As the ionic ionic activation energies from the vertical structure
radius of Br- (0.196 nm) is smaller than I- (0.220 nm), devices of MAPbI3, MAPbBr3, FAPbI3, and CsPbI3
Br- will more easily migrate in the perovskite crystal were examined using selective blocking. The ionic
despite the lattice size of the Br- substituted per- conductivity was extracted from mixed conductivity
ovskite in smaller than generic material. In contrast, at 20 o C to 40 o C. The calculated ionic activation
other groups suggested that according to computa- energy from MAPbI3, FAPbI3, CsPbI3, and MAPb-
tional study, Br- has comparable energy activation to Br 3 , are 0.49 eV, 0.29 eV, 0.39 eV, and 0.07 eV,
 Anafi Nur'aini et al. / J. Electrochem. Sci. Technol., 2022, 13(1), 71-77 77

respectively. These values were compared with other [16] Y. C. Zhao, W. K. Zhou, X. Zhou, K. H. Liu, D. P. Yu,
results in the literature. Variations in ionic activation and Q. Zhao, Light Sci. Appl., 2017, 6(5), e16243-
energy are understandable because every research e16243.
[17] W. Zhou, Y. Zhao, X. Zhou, R. Fu, Q. Li, Y. Zhao, K.
group has different experimental techniques, fabrica-
Liu, D. Yu, and Q. Zhao, J. Phys. Chem. Lett., 2017,
tion processes, and measurement conditions. This 8(17), 4122-4128.
study found that I- is a possible ion that migrates in [18] D. A. Egger, L. Kronik, and A. M. Rappe, Angew.
MAPbI3, FAPbI3, and CsPbI3, whereas Br- is the most Chem. Int. Ed., 2015, 54(42), 12437-12441.
mobile ion in MAPbBr3. [19] G. Jang, H. C. Kwon, S. Ma, S. C. Yun, H. Yang, and J.
Moon, Adv. Energy Mater., 2019, 9(36), 1901719.
[20] L. J. Phillips, A. M. Rashed, R. E. Treharne, J. Kay, P.
Acknowledgments
Yates, I. Z. Mitrovic, A. Weerakkody, S. Hall, and K.
Durose, Sol. Energy Mater. Sol. Cells., 2016, 147, 327-
This research was supported by the sabbatical of 333.
Kumoh National Institute of Technology. [21] L. K. Ono, E. J. Juarez-Perez, and Y. Qi, ACS Appl.
Mater. Interfaces., 2017, 9(36), 30197-30246.
References [22] G. E. Eperon, D. Bryant, J. Troughton, S. D. Stranks, M.
B. Johnston, T. Watson, D. A. Worsley, and H. J. Snaith,
[1] A. Kojima, K. Teshima, Y. Shirai, and T. Miyasaka, J. J. Phys. Chem. Lett., 2015, 6(1), 129-138.
Am. Chem. Soc., 2009, 131(17), 6050-6051. [23] Y. G. Kim, T. Y. Kim, J. H. Oh, K. S. Choi, Y. J. Kim,
[2] N. R. E. L. (NREL), http://www.nrel.gov/ncpv/images/ and S. Y. Kim, Phys. Chem. Chem. Phys., 2017, 19(8),
efficiency_chart.jpg. 2016. 6257-6263.
[3] L. Meng, J. You, and Y. Yang, Nat. Commun., 2018, [24] G. E. Eperon, G. M. Paternò, R. J. Sutton, A. Zampetti,
9(1), 1-4. A. A. Haghighirad, F. Cacialli, and H. J. Snaith, J.
[4] Y. Yuan and J. Huang, Acc. Chem. Res., 2016, 49(2), Mater. Chem. A., 2015, 3(39), 19688-19695.
286-293. [25] K. H. Wang, L. C. Li, M. Shellaiah, and K. W. Sun, Sci.
[5] W. Nie, J. C. Blancon, A. J. Neukirch, K. Appavoo, H. Rep., 2017, 7(1), 1-14.
Tsai, M. Chhowalla, M. A. Alam, M. Y. Sfeir, C. Katan, [26] J. W. Lee, S. G. Kim, J. M. Yang, Y. Yang, and N. G.
J. Even, S. Tretiak, J. J. Crochet, G. Gupta, and A. D. Park, APL Mater., 2019, 7(4), 041111.
Mohite, Nat. Commun., 2016, 7(1), 1-9. [27] B. Yang, C. C. Brown, J. Huang, L. Collins, X. Sang, R.
[6] Z. Xiao, Y. Yuan, Y. Shao, Q. Wang, Q. Dong, C. Bi, P. R. Unocic, S. Jesse, S. V. Kalinin, A. Belianinov, J.
Sharma, A. Gruverman, and J. Huang, Nat. Mater., Jakowski, D. B. Geohegan, B. G. Sumpter, K. Xiao, and
2015, 14(2), 193-197. O. S. Ovchinnikova, Adv. Funct. Mater., 2017, 27(26),
[7] H. J. Snaith, A. Abate, J. M. Ball, G. E. Eperon, T. 1700749.
Leijtens, N. K. Noel, S. D. Stranks, J. T. W. Wang, K. [28] O. Almora, I. Zarazua, E. Mas-Marza, I. Mora-Sero, J.
Wojciechowski, and W. Zhang, J. Phys. Chem. Lett., Bisquert, and G. Garcia-Belmonte, J. Phys. Chem. Lett.,
2014, 5(9), 1511-1515. 2015, 6(9), 1645-1652.
[8] H. S. Kim and N. G. Park, J. Phys. Chem. Lett., 2014, [29] C. Eames, J. M. Frost, P. R. F. Barnes, B. C. O’Regan,
5(17), 2927-2934. A. Walsh, and M. S. Islam, Nat. Commun., 2015, 6(1),
[9] Y. Shao, Z. Xiao, C. Bi, Y. Yuan, and J. Huang, Nat. 1-8.
Commun., 2014, 5(1), 1-7. [30] J. M. Azpiroz, E. Mosconi, J. Bisquert, and F. De
[10] S. H. Kim and D. Lee, J. Phys. Chem. C., 2019, Angelis, Energy Environ. Sci., 2015, 8(7), 2118-2127.
123(14), 9629-9633. [31] Y. Nah, O. Allam, H. S. Kim, J. I. Choi, I. S. Kim. J.
[11] T. Zhang, H. Chen, Y. Bai, S. Xiao, L. Zhu, C. Hu, Q. Byun, S. O. Kim, S. S. Jang, and D. H. Kim, ACS Nano,
Xue, and S. Yang, Nano Energy., 2016, 26, 620-630. 2021, 15(1), 1486-1496.
[12] Y. Yuan, J. Chae, Y. Shao, Q. Wang, Z. Xiao, A. [32] G. Y. Kim, A. Senocrate, Y. Wang, D. and Moia, J.
Centrone, and J. Huang, Adv. Energy Mater., 2015, Maier, Angew. Chem. Int. Ed., 2021, 60, 820-826.
5(15), 1500615. [33] J. Haruyama, K. Sodeyama, L. Han, and Y. Tateyama, J.
[13] T. Y. Yang, G. Gregori, N. Pellet, M. Grätzel, and J. Am. Chem. Soc., 2015, 137(32), 10048-10051.
Maier, Angew. Chemie - Int. Ed., 2015, 54(27), 7905- [34] S. Meloni, T. Moehl, W. Tress, M. Franckeviius, M.
7910. Saliba, Y. H. Lee, P. Gao, M. K. Nazeeruddin, S. M.
[14] D. Klotz, G. Tumen-Ulzii, C. Qin, T. Matsushima, and Zakeeruddin, U. Rothlisberger, and M. Graetzel, Nat.
C. Adachi, RSC Adv., 2019, 9(57), 33436-33445. Commun., 2016, 7(1), 1-9.
[15] T. Zhang, X. Meng, Y. Bai, S. Xiao, C. Hu, Y. Yang, H. [35] S. R. G. Balestra, J. M. Vicent-Luna, S. Calero, S. Tao,
Chen, and S. Yang, J. Mater. Chem. A., 2017, 5(3), and J. A. Anta, J. Mater. Chem. A., 2020, 8(23), 11824-
1103-1111. 11836.