Quantification of Ion Migration in CH3NH3PbI3 Perovskite

Solar Cells by Transient Capacitance Measurements

Moritz H. Futscher1, Ju Min Lee1, Lucie McGovern1, Loreta A. Muscarella1, Tianyi Wang1,

Muhammad Irfan Haider2, Azhar Fakharuddin2, Lukas Schmidt-Mende2 and Bruno Ehrler1*

1. Center for Nanophotonics, AMOLF, Science Park 104,

1098 XG Amsterdam, The Netherlands

2. Department of Physics, University of Konstanz, Universitätsstraße 10,

78457 Konstanz, Germany

AUTHOR INFORMATION

Corresponding Author

* ehrler@amolf.nl
Abstract

Solar cells based on organic-inorganic metal halide perovskites show efficiencies close to

highly-optimized silicon solar cells. However, ion migration in the perovskite films leads to

device degradation and impedes large scale commercial applications. We use transient ion-

drift measurements to quantify activation energy, diffusion coefficient, and concentration of

mobile ions in methylammonium lead triiodide (MAPbI3) perovskite solar cells, and find that

their properties change close to the tetragonal-to-orthorhombic phase transition

temperature. We identify three migrating ion species which we attribute to the migration of

iodide (I-) and methylammonium (MA+). We find that the concentration of mobile MA+ ions is

one order of magnitude higher than the one of mobile I- ions, and that the diffusion

coefficient of mobile MA+ ions is three orders of magnitude lower than the one for mobile I-

ions. We furthermore observe that the activation energy of mobile I- ions (0.29 ± 0.06 eV) is

highly reproducible for different devices, while the activation energy of mobile MA+ depends

strongly on device fabrication. This quantification of mobile ions in MAPbI3 will lead to a

better understanding of ion migration and its role in operation and degradation of

perovskite solar cells.

Introduction

Organic-inorganic metal halide perovskites have proven to be a promising candidate for low-

cost photovoltaic devices. Due to their large charge-carrier diffusion length, long charge-

carrier lifetime, and high defect tolerance, perovskite solar cells already reach record

efficiencies greater than 23%, outperforming any other solution-processed solar cell

technology.1–4 Hybrid perovskites benefit from a continuously tunable bandgap, making

them favourable for multi-junction devices with potential power conversion efficiencies

above 30%.5–9

Unlike conventional inorganic solar cell materials, hybrid perovskites are ionic solids that

exhibit ion migration, complicating the efficiency measurements and the definition of a

steady-state condition in these cells.10 This ion migration has also been shown to be a

pathway for the degradation of perovskite solar cells.11,12 The understanding of ion

migration within perovskite solar cells is therefore crucial for the fabrication of stable

perovskite devices.

In methylammonium lead triiodide (MAPbI3), both anions (I-) and cations (methylammonium

MA+, Pb2+) can migrate due to the presence of vacancies, interstitials, or antisite

substitutions. A large variety of activation energies for ion migration have been published,

both experimentally and theoretically. Theoretical calculations predict activation energies

between 0.08 and 0.58 eV, 0.46 and 1.12 eV, and 0.80 and 2.31 eV for the migration of I-,

MA+, and Pb2+, respectively.13–17 Attempts to experimentally determine the activation energy

have given a similar variety of results.18–25 Most experimental techniques further fail to

distinguish between the charge of the ions (anions and cations), which can lead to mis-

assignment of the ion species.

Here we quantify the activation energy, diffusion coefficient, sign of charge, and

concentration of mobile ions in MAPbI3 using transient ion-drift, one of the most powerful

methods to quantify ion migration.26,27 We show that probing the capacitance change

associated with ion migration requires to measure the capacitance transients on the

timescale of seconds. Using transient ion-drift we identify footprints of distinct mobile ion

species which we attribute to the migration of I- (activation energy 0.29 eV) and MA+ (0.39 -

0.90 eV). We find that the concentration of mobile MA+ ions is one order of magnitude

higher than the one of mobile I- ions, and that the diffusion coefficient of mobile MA+ ions is

three orders of magnitude lower than the one for mobile I- ions. As a result, the migration of

MA+ ions leads to a capacitance transient with a time scale of seconds, where the migration

of I- ions results in a transient with a time scale of less than a millisecond at 300 K. This

quantification leads to a better understanding of ion migration, which is a crucial step

towards stable perovskite solar cells.

Results and discussion

Transient ion-drift measurements rely on the external application of an electric field. We use

a diode configuration to study ion migration. Our diode consists of an inverted planar

perovskite solar cell architecture with a solution-processed NiOx film as a hole-transporting

layer and C60 as an electron-transport layer28, as shown in Figure 1a. We chose the inverted

solar cell structure over the standard one due to the strong tendency to accumulate charges,

both electronic and ionic, at the TiO2/perovskite interface resulting in a capacitive hysteresis

and additional dielectric contributions, which is reduced in the inverted structure (see

section S1 in the Supporting Information (SI)).29 In the inverted solar cell architecture,

PEDOT:PSS (poly(3,4-ethylenedioxythiophene)-poly(styrenesulfonate)) is the most widely

used hole-transport material, however, its high acidity and tendency to absorb water might

lead to unwanted device degradation.30 We furthermore avoid using spiro-MeOTAD

(2,2′,7,7′-tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9′-spirobifluorene) since typical

dopants such as lithium salts lead to instabilities due to their high sensitivity to moisture,

and can show misleading features in the transient ion-drift measurements due to the

additional dopant ion migration.31,32 The current-voltage characteristics of a perovskite solar

cell in the dark in forward and reverse voltage scans are shown in Figure 1b (see Figure S3 in

the SI for current-voltage characteristics measured at 1-Sun). We observe a significant

difference between the forward and reverse scanning direction at 300 K. When cooling the

perovskite solar cell, this current-voltage hysteresis is strongly reduced and almost vanishes

at 180 K (see inset Figure 1b, and Figure S4 in the SI for current-voltage curves measured

between 180 and 330 K). This has previously been attributed to the inhibition of the ion

migration at low temperatures.33,34

Figure 1. Inverted MAPbI3 device characteristics. (a) Cross-section scanning electron microscopy, (b) current-

voltage characteristics measured in the dark, and (c) Impedance spectroscopy measured in the dark at 0 V with

an AC perturbation of 20 mV. The high capacitance at low frequencies is attributed to the high ionic

conductivity mediated by defect states.

The transient ion-drift technique relies on probing the ion migration in the perovskite layer

using capacitance transients at different temperatures. To find the suitable AC frequency

regime for measuring capacitance transients, we study the frequency-dependence of the

capacitance of the perovskite diode in the dark (see Figure 1c). At low frequencies, the

capacitance is dominated by mobile ions which accumulate near the contact interfaces.35

When reducing the temperature, both the current-voltage hysteresis and the ionic

capacitance contribution are strongly reduced until they disappear close to 180 K. At high

frequencies, the capacitance is reduced due to the series resistance of the conductive

contact layers reducing the cut-off frequency of the device. Between the two limits lies a

relatively constant plateau, corresponding to the geometric capacitance of the device, which

is related to the perovskite permittivity (see section S3 in the SI for details).

 + -
Figure 2. Influence of ion migration on the band energies. (a) At short circuit, mobile MA and I ions

accumulate at the interface partially screening the built-in electric field. (b) When applying a forward bias (V0),

mobile ions will drift towards the bulk, (c) resulting in a uniform ion distribution within the perovskite layer. (d)

After removing the forward bias, the built-in electric field will drive mobile ions towards the interfaces. This

drift of mobile ions towards the interfaces results in a capacitance transient used to quantify ion migration. EC

is the conduction band energy, and EV the valance band energy.

We chose to measure the capacitance at 10 kHz, at the plateau of the capacitance. Transient

ion-drift uses the transient capacitance response following a voltage pulse at different

temperatures (see schematic Figure 2). We apply a forward bias of 0.4 V for 1 second, which

reduces the width of depletion region and leads to a new equilibrium distribution within the

previously depleted region (Figure 2a to c), changing the capacitance of the device. This

change in capacitance did not increase further with longer pulse widths, indicating that a

uniform distribution of ions was reached after the 1 second voltage pulse duration (see

Figure S6 in the SI).26 We avoid using higher external voltages since Yuan et al. found that

external electrical fields as low as 3 V/μm at 330 K can lead to the formation of PbI2.36 After

turning off the voltage pulse, the built-in electric field will drive both the mobile ions and

electric charges back to the contact interfaces (Figure 2d). Mobile anions will follow the

electrons and mobile cations will follow the holes, resulting in a capacitance transient. We

measure this capacitance transient at temperatures between 180 and 350 K (see Figure 3a),

above the first-order phase transition from tetragonal to orthorhombic near 165 K.37,38 We

see no capacitance transient at low temperatures (< 190 K), while a negative capacitance

change grows in between 190 and 280 K until the capacitance decay it is too fast to
measure. At higher temperatures, we observe a positive change in capacitance. Assuming

the total ion concentration is conserved, the electric field varies linearly across the depletion

region, and thermal diffusion of ions is negligible compared to ion drift, the change in

capacitance depends only on the temperature, activation energy, diffusion coefficient, and

concentration of mobile ions as39

t
(1)
C(t) = C(∞) + ΔC (1 − s e−τ )

where 𝛥𝐶 is the change in capacitance due to the drift of mobile ions towards the interfaces,

𝐶(∞) the steady-state capacitance, 𝑠 the sign of the charge, and 𝜏 a time constant given by

k B T ε0 ε (2)
τ=
q2 D N

where 𝑘𝐵 is the Boltzmann constant, and 𝑇 the temperature (see section S5 in the SI for

𝐸
− 𝐴
details). 𝐷 = 𝐷0 𝑒 𝑘𝐵 𝑇
is the ion-diffusion coefficient where 𝐷0 is the attempt-to-escape

frequency for ion migration and 𝐸𝐴 the activation energy. The assumption that the electric

field varies linearly across the depletion region is supported by recent studies showing that

the electric field varies linearly within the perovskite layer when the perovskite layer is

subjected to an external or internal electric field.40,41 We can describe the measured

capacitance transients using exponential functions, which further corroborates that the

assumption of a linear field is valid in our devices (see section S6 in the SI).26 We note that

Weber et al. found an additional interface dipole at the perovskite/SnO2 interface.40 This

interface dipole is deliberately omitted in our structure by using NiO x and C60 as extraction

layers (see section S1 in the SI). As metals are prone to reacting with I- ions,42 we have

ensured to perform our measurements shortly after the fabrication of the diodes. In

addition, we have carefully chosen the AC frequency to ensure that the measured
capacitance is not affected by potential ion diffusion through the transport layers. (see

section S7 in the SI).

To identify processes associated to these capacitance changes we use the rate window

analysis, originally introduced by Lang to analyse deep-level transient spectroscopy (DLTS)

measurements.43 The capacitance change extracted by this method is given by

𝛥𝐶 = 𝐶(𝑡1 ) − 𝐶(𝑡2 ), where 𝑡1 and 𝑡2 depend on the typical decay times of the capacitance

at a certain temperature to extract a peak associated with each activation energy. When

choosing 𝑡1 = 2𝑡2 from milliseconds to seconds we find three peaks corresponding to three

separate processes, which we label A1, C1, and C2 (see Figure 3b). The capacitance change

associated with C1 and C2 both are positive and describe the migration of a cation. A1 is

negative and describes the migration of an anion. We hence assign A1 to the migration of I -

ions and C1 and C2 to the migration of MA+ ions. We exclude the migration of Pb2+ ions since

theoretical studies suggest that they are unlikely to migrate.17 Note that we cannot rule out

the migration of H+ ions, which was calculated to have an activation energy of 0.29 eV.44

However, the predicted concentration of H+ ions in MAPbI3 is in the order of 1011 cm-3,45

orders of magnitude lower than what we have measured.

The temperature dependence of the peaks in the rate window analysis together with their

time scales can be used to obtain activation energy and diffusion coefficient of ion

migration. This method, however, uses only two points of each transient to extract the time

scales. To quantify ion migration using all data points, we fit the measured capacitance

changes to exponential decays to obtain the time constants 𝜏 at different temperatures

(Equation 1). By means of an Arrhenius plot we can extract both the activation energy and

diffusion coefficient (see Figure 3c). We again identify the three species, C1, C2, and A1,

where A1 occurs at much faster timescales and lower temperatures.

Figure 3. Ion migration in MAPbI3. (a) Capacitance transient measurements between 180 and 350 K with steps

of 10 K measured at 0 V with an AC voltage of 10 mV at 10 kHz after a voltage pulse of 0.4 V for 1 second. (b)

Rate-window plot of measured capacitance transients with different time constants ranging from milliseconds

(red) to seconds (blue) reveal three ion species with different thermal emission rates. We attribute A1 to the
- +
migration of I ions and C1 and C2 to the migration of MA ions. (c) Arrhenius plot of the observed thermal

emission rates as a function of temperature. The linear fit reveals the activation energy and the diffusion

coefficient of the mobile ion species.

To estimate the sample-to-sample, and lab-to-lab variation we measured solar cells

fabricated at AMOLF and at the University of Konstanz, with power conversion efficiencies

ranging from 1 to 12% (see section S9 in the SI for details). The obtained characteristics of

mobile ions for all the devices are shown in Figure 4 and the mean values are summarized in

Table 1. We find that the activation energy for the migration of I- ions is very reproducible

across all devices, while the activation energy for the migration of MA+ ions depends strongly

on the fabrication conditions, which is consistent with the wide distribution of activation

energies for the migration of MA+ ions in literature. The wide distribution of activation

energies for the migration of I- ions in the literature could be explained by the

misinterpretation of mobile ion species, since most techniques cannot distinguish between the

migration of anions and cations. The transient ion-drift measurements are able to
simultaneously distinguish between mobile cations and anion, and detect low

concentrations of mobile impurities (~ 0.01% of the doping density). Our measurements thus

show that many theoretical calculations cannot be experimentally verified within the margin

of error.

Interestingly, we obtain a diffusion coefficient of 10-9 cm2 s-1 for I- ions which is three orders

of magnitude higher than the diffusion coefficient for MA + ions of 10-12 cm2 s-1 (see Table 1).

The diffusion coefficients measured here are very close to the diffusion coefficients

measured with NMR (10-9 cm2 s-1 for I- and 10-15 - 10-12 cm2 s-1 for MA+),46,47 and to those

obtained in recent studies by Li et al. (5 x 10-8 to 6 x 10-9 cm2 s-1 for I-) and Bertoluzzi et al.

(8 x 10-9 cm2 s-1 for I-).48,49 Solute-dopant pairing can significantly slow down the ionic

diffusion,26 which could be the reason for the slow diffusion of MA + ions. Only the MA+ ions

have a transient decay time in the order of seconds at typical operation temperatures (< ms

for I-). Thus, our results suggest that mobile MA+ ions are the origin of the observed current-

voltage hysteresis in MAPbI3 perovskite solar cells. Previously, also I- has been assigned

responsible for the current-voltage hysteresis,17 however, the sensitivity of transient ion-

drift to the sign of the ion excludes this possibility.

Close to the tetragonal-cubic phase-transition temperature (327 K)45 we observe a decrease

in activation energy and an increase in diffusion coefficient for one of the migrating MA+ ions

(C1), with the exception for the device with a power conversion efficiency of 1%. A similar

behaviour has previously been observed and attributed to the volume change in the unit cell

at temperatures close to the tetragonal-cubic phase transition.50,51 Note that C2 might show

a similar behaviour at lower temperatures, however, the activation energy of C2 in the

tetragonal phase could not be resolved in our measurements due to its long-time constant.

The obtained activation energies of the two migration pathways for MA + (C1 and C2) in the
cubic phase are comparable, yet the diffusion coefficient of C1 is somewhat higher than the

diffusion coefficient of C2. Using Kelvin probe force microscopy, Yun et al. found that ion

migration near grain boundaries is much faster than inside the grains due to higher ionic

diffusivity at grain boundaries.52 We thus speculate that these are both mobile MA+ species

where C1 has a higher diffusion coefficient, which could be due to ion movement in vicinity

to grain boundaries.

Figure 4. Characteristics of mobile ions in MAPbI3. (a) Activation energy, (b) diffusion coefficient at 300 K, (c)

and concentration of mobile ions in MAPbI3 perovskites obtained by transient ion-drift. The downward and the

upward triangle represents measurements below and above the tetragonal-to-cubic phase-transition

temperature. The mean values are summarized in Table 1.

We measure the concentration of mobile ions from the change in capacitance following the

voltage pulse. Since the capacitance 𝐶(∞) ∝ √𝑁, where 𝑁 is the doping density, the

concentration of mobile ions 𝑁𝑚𝑜𝑏𝑖𝑙𝑒 within the probed depletion region can be estimated

as53

𝟐 (3)
𝚫𝑪 + 𝑪(∞) 𝑵𝒎𝒐𝒃𝒊𝒍𝒆 + 𝑵
( ) ∝( ).
𝑪(∞) 𝑵

The obtained concentrations for mobile I- and MA+ ions are summarized in Figure 4c. We

note that we assume a typical doping density of 1 x 1017 cm-3 for all the measured perovskite

films and temperatures (see section S3 in the SI).54 Although the density of mobile ions

depends on the fabrication, we find that the concentration of the mobile MA+ ions is
 systematically about one order of magnitude higher than that of the mobile I- ions. The

measured mobile ion concentration is rather low compared to other studies, which report

values of around 1018 cm-3.49 However, several recent studies measure a mobile ion

concentration comparable to what we measure, on the order of 1015 cm-3,40,41 suggesting

that less than 10% of the screening of the electric field within the perovskite layer is

produced by the presence of mobile ions. We note that electrical neutrality is still given, as

the concentration obtained is the concentration of mobile ions within the perovskite film,

not all ions present in the perovskite film.

Table 1. Characteristics of mobile ions in MAPbI3.

A1 C1 C2
Migrating ion species I- MA+ MA+
Charge negative positive positive
Concentration (cm-3) (1.1 ± 0.9) x 1015 (1.3 ± 0.8) x 1016 (5.0 ± 4.0) x 1015
Phase structure tetragonal tetragonal cubic cubic
Activation energy (eV) 0.29 ± 0.06 0.90 ± 0.45 0.46 ± 0.25 0.39 ± 0.24
Diffusion coefficient
(3.1 ± 2.8) x 10-9 (3.4 ± 3.3) x 10-12 (6.8 ± 5.3) x 10-12 (1.6 ± 0.8) x 10-12
at 300 K (cm2 s-1)

Capacitance transients such as the ones observed here could also originate from deep-level

charge traps. A powerful method to measure charge-carrier traps is DLTS,43 a method which

is very similar to transient ion-drift. In DLTS available states in the bandgap are filled with

charge carriers by applying a voltage pulse. Trapped charge carriers can then be thermally

excited to conducting states and swept out of the depletion region by the junction potential,

resulting in a capacitance transient. DLTS has been used to study fast (< milliseconds) charge

trapping in perovskite solar cells,55 in contrast, ion migration in perovskites typically

proceeds on long timescales of milliseconds to seconds.56,57 Furthermore, the ratio of rise

and decay times of the capacitance in DLTS and transient ion-drift is different, so that we can
distinguish ion migration from trapping and de-trapping of charge carriers (see section S8 in

the SI for details).27 We can therefore attribute the observed transients as the result of ion

migration rather than deep-level charge traps. Atomistic simulations furthermore suggest

that deep-level defects require such high formation energies that their formation is

unlikely.1

To conclude, we have shown that transient ion-drift is a fast and accurate method to

quantify, with high precision, the activation energy, diffusion coefficient, sign of charge, and

concentration of mobile ions in perovskite solar cells. In MAPbI3 perovskites we observe that

both MA+ and I- are migrating. We find that the concentration of mobile MA+ ions is

significantly higher than the concentration of mobile I- ions and that the diffusion coefficient

of I- ions is three orders of magnitude higher than the diffusion coefficient of MA+ ions. On

timescales associated with current-voltage measurements, only the migration of MA+ ions is

slow enough to cause a current-voltage hysteresis in MAPbI3 solar cells. The migration of I- is

still relevant for the device operation, and the degradation of perovskite solar cells. The

migration of mobile I- ion is very reproducible across devices fabricated in different

laboratories, while the migration of mobile MA+ ions strongly depends on the fabrication,

which explains the wide distribution of activation energies for the migration of MA+ ions in

literature. Our measurements guide the future theoretical investigation into ion migration in

halide perovskites and offer quantitative insight into the parameters of the mobile ion

species, and hence the degradation pathways of perovskite solar cells.

Experimental methods

Diode fabrication: Laser patterned indium tin oxide (ITO) glass substrates were cleaned by

ultra-sonication for 20 minutes subsequently in detergent in deionized water, deionized water,

acetone, and isopropanol, followed by oxygen plasma for 20 minutes at 100 W. Nickel oxide

(NiOx) precursor solution (0.1 M nickel(II) nitrate hexahydrate (Aldrich) in ethanol) filtered with

a 0.45 µm PTFE membrane was spun on the cleaned ITO glass at 4000 rpm for 30 seconds. This

step was then repeated two times.28 Annealing at 350 °C for 1 hour with a ramping speed of

3 °C/min induced NiOx film formation. The MAPbI3 perovskite precursor solution was prepared

by mixing of total 1.35 M of methylammonium iodide (MAI, solaronix) and lead(II) iodide (PbI2,

Aldrich) with 1:1 molar ratio dissolved in N,N-dimethylformamide (anhydrous, Aldrich) at

60 °C. The MAPbI3 precursor solution was filtered through a 0.45 µm PTFE membrane spun

onto NiOx coated substrates at 5000 rpm for 25 seconds in a nitrogen filled glove box. 5

seconds after the beginning of the rotation, 180 μL of chlorobenzene anti-solvent (anhydrous,

Aldrich) was quickly dropped onto the substrate. After the MAPbI3 spinning process, the

substrates were annealed at 100 °C for 15 minutes. 30 nm of C60 (0.5 Å/s rate, 99.9%, Aldrich),

8 nm of bathocuproine (0.2 Å/s, 99.99%, Aldrich), 50 nm of silver (1 Å/s, 99.99%, Kurt J. Lesker)

and 150 nm of gold electrode (1 Å/s, 99.999%, Kurt J. Lesker) were sequentially deposited on

top of MAPbI3 layer by thermal sublimation/evaporation at pressures below 2 × 10-7 mbar.

Differences in the sample fabrication for devices made at AMOLF and the University of

Konstanz can be found in section S10 in the SI.

Electrical measurements: To avoid air exposure, the sample was loaded into a Janis VPF-100

liquid nitrogen cryostat inside a nitrogen-filled glovebox. Current-voltage, impedance

spectroscopy, capacitance-voltage, and transient ion-drift measurements were performed at a

pressure below 2 x 10-6 mbar in the dark using a commercially available DLTS system from

Semetrol. To ensure thermal equilibrium the temperature of the sample was held constant for

at least 30 minutes before current-voltage, impedance spectroscopy, and capacitance-voltage

measurements. The capacitance was modelled by a capacitor in parallel with a conductance.

Capacitance transient measurements were performed from 180 to 350 K in steps of 2 K with a

heating rate of about 2 K per minute. The sample was held at 180 K for one hour before

starting the transient ion-drift measurement.

Imaging of device cross-section: To obtain a clean cross-section of the device, it was immersed

in liquid nitrogen for 60 seconds and cleaved in the center. The cross-sectional image was

taken with a FEI Verios 460 scanning electron microscope in the secondary electron mode. An

acceleration voltage of 5 kV and a working distance of 4 mm were used and field immersion

mode was applied for an optimized resolution.

Acknowledgements

The authors thank Erik C. Garnett for carefully reading and commenting on the manuscript.

This work is part of the research program of the Netherlands Organization for Scientific

Research (NWO).

References

(1) Yin, W. J.; Shi, T.; Yan, Y. Unusual Defect Physics in CH3NH3PbI3 Perovskite Solar Cell
Absorber. Appl. Phys. Lett. 2014, 104 (6), 063903. https://doi.org/10.1063/1.4864778.
(2) Kojima, A.; Teshima, K.; Shirai, Y.; Miyasaka, T. Organometal Halide Perovskites as
Visible-Light Sensitizers for Photovoltaic Cells. J. Am. Chem. Soc. 2009, 131, 6050–
6051. https://doi.org/10.1021/ja809598r.
(3) Polman, A.; Knight, M.; Garnett, E. C.; Ehrler, B.; Sinke, W. C. Photovoltaic Materials –
Present Efficiencies and Future Challenges. Science 2016, 352, 307.
https://doi.org/10.1126/science.aad4424.
(4) Jiang, Q.; Chu, Z.; Wang, P.; Yang, X.; Liu, H.; Wang, Y.; Yin, Z.; Wu, J.; Zhang, X.; You, J.
Planar-Structure Perovskite Solar Cells with Efficiency beyond 21%. Adv. Mater. 2017.
https://doi.org/10.1002/adma.201703852.
(5) Hörantner, M. T.; Snaith, H. J. Predicting and Optimising the Energy Yield of
Perovskite-on-Silicon Tandem Solar Cells under Real World Conditions. Energy
Environ. Sci. 2017, 10 (9), 1983–1993. https://doi.org/10.1039/c7ee01232b.
(6) Futscher, M. H.; Ehrler, B. Modeling the Performance Limitations and Prospects of
Perovskite/Si Tandem Solar Cells under Realistic Operating Conditions. ACS Energy
Lett. 2017, 2 (9), 2089–2095. https://doi.org/10.1021/acsenergylett.7b00596.
(7) Lee, J. M.; Futscher, M. H.; Pazos-Outon, L. M.; Ehrler, B. Highly Transparent Singlet
Fission Solar Cell with Multi-Stacked Thin Metal Contacts for Tandem Applications.
Prog. Photovoltaics Res. Appl. 2017. https://doi.org/10.1002/pip.2919.
(8) Karani, A.; Yang, L.; Bai, S.; Futscher, M. H.; Snaith, H. J.; Ehrler, B.; Greenham, N. C.;
 Di, D. Perovskite/Colloidal Quantum Dot Tandem Solar Cells: Theoretical Modeling
and Monolithic Structure. ACS Energy Lett. 2018, 3 (4), 869–874.
https://doi.org/10.1021/acsenergylett.8b00207.
(9) Futscher, M. H.; Ehrler, B. Efficiency Limit of Perovskite/Si Tandem Solar Cells. ACS
Energy Lett. 2016, 1 (4), 2–7. https://doi.org/10.1021/acsenergylett.6b00405.
(10) Snaith, H. J.; Abate, A.; Ball, J. M.; Eperon, G. E.; Leijtens, T.; Noel, N. K.; Stranks, S. D.;
Wang, J. T.-W.; Wojciechowski, K.; Zhang, W. Anomalous Hysteresis in Perovskite
Solar Cells. J. Phys. Chem. Lett. 2014, 5 (9), 1511–1515.
https://doi.org/10.1021/jz500113x.
(11) Nandal, V.; Nair, P. R. Predictive Modeling of Ion Migration Induced Degradation in
Perovskite Solar Cells. ACS Nano 2017, 11 (11), 11505–11512.
https://doi.org/10.1021/acsnano.7b06294.
(12) Bowring, A. R.; Bertoluzzi, L.; O’Regan, B. C.; McGehee, M. D. Reverse Bias Behavior of
Halide Perovskite Solar Cells. Adv. Energy Mater. 2018.
https://doi.org/10.1002/aenm.201702365.
(13) Azpiroz, J. M.; Mosconi, E.; Bisquert, J.; De Angelis, F. Defect Migration in
Methylammonium Lead Iodide and Its Role in Perovskite Solar Cell Operation. Energy
Environ. Sci. 2015, 8 (7), 2118–2127. https://doi.org/10.1039/C5EE01265A.
(14) Haruyama, J.; Sodeyama, K.; Han, L.; Tateyama, Y. First-Principles Study of Ion
Diffusion in Perovskite Solar Cell Sensitizers. J. Am. Chem. Soc. 2015, 137 (32), 10048–
10051. https://doi.org/10.1021/jacs.5b03615.
(15) Meloni, S.; Moehl, T.; Tress, W.; Franckeviius, M.; Saliba, M.; Lee, Y. H.; Gao, P.;
Nazeeruddin, M. K.; Zakeeruddin, S. M.; Rothlisberger, U.; et al. Ionic Polarization-
Induced Current-Voltage Hysteresis in CH3NH3PbX3perovskite Solar Cells. Nat.
Commun. 2016, 7, 10334. https://doi.org/10.1038/ncomms10334.
(16) Delugas, P.; Caddeo, C.; Filippetti, A.; Mattoni, A. Thermally Activated Point Defect
Diffusion in Methylammonium Lead Trihalide: Anisotropic and Ultrahigh Mobility of
Iodine. J. Phys. Chem. Lett. 2016, 7 (13), 2356–2361.
https://doi.org/10.1021/acs.jpclett.6b00963.
(17) Eames, C.; Frost, J. M.; Barnes, P. R. F.; O’Regan, B. C.; Walsh, A.; Islam, M. S. Ionic
Transport in Hybrid Lead Iodide Perovskite Solar Cells. Nat. Commun. 2015, 6, 7497.
https://doi.org/10.1038/ncomms8497.
(18) Li, C.; Guerrero, A.; Zhong, Y.; Huettner, S. Origins and Mechanisms of Hysteresis in
Organometal Halide Perovskites. J. Phys. Condens. Matter 2017, 29 (19), 193001.
https://doi.org/10.1088/1361-648X/aa626d.
(19) Yuan, Y.; Chae, J.; Shao, Y.; Wang, Q.; Xiao, Z.; Centrone, A.; Huang, J. Photovoltaic
Switching Mechanism in Lateral Structure Hybrid Perovskite Solar Cells. Adv. Energy
Mater. 2015, 5 (15), 1500615. https://doi.org/10.1002/aenm.201500615.
(20) Game, O. S.; Buchsbaum, G. J.; Zhou, Y.; Padture, N. P.; Kingon, A. I. Ions Matter:
Description of the Anomalous Electronic Behavior in Methylammonium Lead Halide
Perovskite Devices. Adv. Funct. Mater. 2017, 27 (16), 1606584.
https://doi.org/10.1002/adfm.201606584.
(21) Pockett, A.; Eperon, G. E.; Sakai, N.; Snaith, H. J.; Peter, L. M.; Cameron, P. J.
Microseconds, Milliseconds and Seconds: Deconvoluting the Dynamic Behaviour of
Planar Perovskite Solar Cells. Phys. Chem. Chem. Phys. 2017, 19 (8), 5959–5970.
https://doi.org/10.1039/C6CP08424A.
(22) Yang, T. Y.; Gregori, G.; Pellet, N.; Grätzel, M.; Maier, J. The Significance of Ion
Conduction in a Hybrid Organic-Inorganic Lead-Iodide-Based Perovskite
 Photosensitizer. Angew. Chemie - Int. Ed. 2015, 54 (27), 7905–7910.
https://doi.org/10.1002/anie.201500014.
(23) Xing, J.; Wang, Q.; Dong, Q.; Yuan, Y.; Fang, Y.; Huang, J. Ultrafast Ion Migration in
Hybrid Perovskite Polycrystalline Thin Films under Light and Suppression in Single
Crystals. Phys. Chem. Chem. Phys. 2016, 18 (44), 30484–30490.
https://doi.org/10.1039/C6CP06496E.
(24) Yu, H.; Lu, H.; Xie, F.; Zhou, S.; Zhao, N. Native Defect-Induced Hysteresis Behavior in
Organolead Iodide Perovskite Solar Cells. Adv. Funct. Mater. 2016, 26 (9), 1411–1419.
https://doi.org/10.1002/adfm.201504997.
(25) DeQuilettes, D. W.; Zhang, W.; Burlakov, V. M.; Graham, D. J.; Leijtens, T.; Osherov, A.;
Bulović, V.; Snaith, H. J.; Ginger, D. S.; Stranks, S. D. Photo-Induced Halide
Redistribution in Organic-Inorganic Perovskite Films. Nat. Commun. 2016, 7, 11683.
https://doi.org/10.1038/ncomms11683.
(26) Heiser, T.; Weber, E. Transient Ion-Drift-Induced Capacitance Signals in
Semiconductors. Phys. Rev. B 1998, 58 (7), 3893–3903.
https://doi.org/10.1103/PhysRevB.58.3893.
(27) Heiser, T.; Mesli, A. Determination of the Copper Diffusion Coefficient in Silicon from
Transient Ion-Drift. Appl. Phys. A Solids Surfaces 1993, 57 (4), 325–328.
https://doi.org/10.1007/BF00332285.
(28) Yin, X.; Yao, Z.; Luo, Q.; Dai, X.; Zhou, Y.; Zhang, Y.; Zhou, Y.; Luo, S.; Li, J.; Wang, N.; et
al. High Efficiency Inverted Planar Perovskite Solar Cells with Solution-Processed NiOx
Hole Contact. ACS Appl. Mater. Interfaces 2017, 9 (3), 2439–2448.
https://doi.org/10.1021/acsami.6b13372.
(29) Garcia-Belmonte, G.; Bisquert, J. Distinction between Capacitive and Noncapacitive
Hysteretic Currents in Operation and Degradation of Perovskite Solar Cells. ACS
Energy Letters. American Chemical Society October 14, 2016, pp 683–688.
https://doi.org/10.1021/acsenergylett.6b00293.
(30) Manders, J. R.; Tsang, S. W.; Hartel, M. J.; Lai, T. H.; Chen, S.; Amb, C. M.; Reynolds, J.
R.; So, F. Solution-Processed Nickel Oxide Hole Transport Layers in High Efficiency
Polymer Photovoltaic Cells. Adv. Funct. Mater. 2013, 23 (23), 2993–3001.
https://doi.org/10.1002/adfm.201202269.
(31) Zhang, Y.; Liu, M.; Eperon, G. E.; Leijtens, T. C.; McMeekin, D.; Saliba, M.; Zhang, W.;
de Bastiani, M.; Petrozza, A.; Herz, L. M.; et al. Charge Selective Contacts, Mobile Ions
and Anomalous Hysteresis in Organic–inorganic Perovskite Solar Cells. Mater. Horiz.
2015, 2 (3), 315–322. https://doi.org/10.1039/C4MH00238E.
(32) Luo, Q.; Zhang, Y.; Liu, C.; Li, J.; Wang, N.; Lin, H. Iodide-Reduced Graphene Oxide with
Dopant-Free Spiro-OMeTAD for Ambient Stable and High-Efficiency Perovskite Solar
Cells. J. Mater. Chem. A 2015, 3 (31), 15996–16004.
https://doi.org/10.1039/C5TA02710A.
(33) Ginting, R. T.; Jung, E.-S.; Jeon, M.-K.; Jin, W.-Y.; Song, M.; Kang, J.-W. Low-
Temperature Operation of Perovskite Solar Cells: With Efficiency Improvement and
Hysteresis-Less. Nano Energy 2016, 27, 569–576.
https://doi.org/10.1016/J.NANOEN.2016.08.016.
(34) Bruno, A.; Cortecchia, D.; Chin, X. Y.; Fu, K.; Boix, P. P.; Mhaisalkar, S.; Soci, C.
Temperature and Electrical Poling Effects on Ionic Motion in MAPbI3Photovoltaic
Cells. Adv. Energy Mater. 2017, 7 (18), 1700265.
https://doi.org/10.1002/aenm.201700265.
(35) Almora, O.; Zarazua, I.; Mas-Marza, E.; Mora-Sero, I.; Bisquert, J.; Garcia-Belmonte, G.
 Capacitive Dark Currents, Hysteresis, and Electrode Polarization in Lead Halide
Perovskite Solar Cells. J. Phys. Chem. Lett. 2015, 6 (9), 1645–1652.
https://doi.org/10.1021/acs.jpclett.5b00480.
(36) Yuan, Y.; Wang, Q.; Shao, Y.; Lu, H.; Li, T.; Gruverman, A.; Huang, J. Electric-Field-
Driven Reversible Conversion between Methylammonium Lead Triiodide Perovskites
and Lead Iodide at Elevated Temperatures. Adv. Energy Mater. 2016, 6 (2), 1501803.
https://doi.org/10.1002/aenm.201501803.
(37) Weller, M. T.; Weber, O. J.; Henry, P. F.; Di Pumpo, A. M.; Hansen, T. C. Complete
Structure and Cation Orientation in the Perovskite Photovoltaic Methylammonium
Lead Iodide between 100 and 352 K. Chem. Commun. 2015, 51 (20), 4180–4183.
https://doi.org/10.1039/C4CC09944C.
(38) Onoda-Yamamuro, N.; Yamamuro, O.; Matsuo, T.; Suga, H. P-T Phase Relations of
CH3NH3PbX3 (X = Cl, Br, I) Crystals. J. Phys. Chem. Solids 1992, 53 (2), 277–281.
https://doi.org/10.1016/0022-3697(92)90056-J.
(39) Zamouche, A.; Heiser, T.; Mesli, A. Investigation of Fast Diffusing Impurities in Silicon
by a Transient Ion Drift Method. Appl. Phys. Lett. 1995, 66 (5), 631.
https://doi.org/10.1063/1.114142.
(40) Weber, S. A. L.; Hermes, I. M.; Turren-Cruz, S.-H.; Gort, C.; Bergmann, V. W.; Gilson, L.;
Hagfeldt, A.; Graetzel, M.; Tress, W.; Berger, R. How the Formation of Interfacial
Charge Causes Hysteresis in Perovskite Solar Cells. Energy Environ. Sci. 2018, 11 (9),
2404–2413. https://doi.org/10.1039/C8EE01447G.
(41) Birkhold, S. T.; Precht, J. T.; Liu, H.; Giridharagopal, R.; Eperon, G. E.; Schmidt-Mende,
L.; Li, X.; Ginger, D. S. Interplay of Mobile Ions and Injected Carriers Creates
Recombination Centers in Metal Halide Perovskites under Bias. ACS Energy Lett. 2018,
3 (6), 1279–1286. https://doi.org/10.1021/acsenergylett.8b00505.
(42) Bi, E.; Chen, H.; Xie, F.; Wu, Y.; Chen, W.; Su, Y.; Islam, A.; Grätzel, M.; Yang, X.; Han, L.
Diffusion Engineering of Ions and Charge Carriers for Stable Efficient Perovskite Solar
Cells. Nat. Commun. 2017, 8, 15330. https://doi.org/10.1038/ncomms15330.
(43) Lang, D. V. Deep-Level Transient Spectroscopy: A New Method to Characterize Traps
in Semiconductors. J. Appl. Phys. 1974, 45 (7), 3023–3032.
https://doi.org/10.1063/1.1663719.
(44) Egger, D. A.; Kronik, L.; Rappe, A. M. Theory of Hydrogen Migration in Organic-
Inorganic Halide Perovskites. Angew. Chemie - Int. Ed. 2015, 54 (42), 12437–12441.
https://doi.org/10.1002/anie.201502544.
(45) Frost, J. M.; Walsh, A. What Is Moving in Hybrid Halide Perovskite Solar Cells?
Accounts of Chemical Research. 2016, pp 528–535.
https://doi.org/10.1021/acs.accounts.5b00431.
(46) Senocrate, A.; Moudrakovski, I.; Kim, G. Y.; Yang, T.-Y.; Gregori, G.; Grätzel, M.; Maier,
J. The Nature of Ion Conduction in Methylammonium Lead Iodide: A Multimethod
Approach. Angew. Chemie Int. Ed. 2017, 56 (27), 7755–7759.
https://doi.org/10.1002/anie.201701724.
(47) Senocrate, A.; Moudrakovski, I.; Acartürk, T.; Merkle, R.; Kim, G. Y.; Starke, U.; Grätzel,
M.; Maier, J. Slow CH3NH3+Diffusion in CH3NH3PbI3under Light Measured by Solid-
State NMR and Tracer Diffusion. Journal of Physical Chemistry C. American Chemical
Society September 27, 2018, pp 21803–21806.
https://doi.org/10.1021/acs.jpcc.8b06814.
(48) Li, C.; Guerrero, A.; Huettner, S.; Bisquert, J. Unravelling the Role of Vacancies in Lead
Halide Perovskite through Electrical Switching of Photoluminescence. Nat. Commun.
 2018, 9 (1), 5113. https://doi.org/10.1038/s41467-018-07571-6.
(49) Bertoluzzi, L.; Belisle, R. A.; Bush, K. A.; Cheacharoen, R.; McGehee, M. D.; O’Regan, B.
C. In Situ Measurement of Electric-Field Screening in Hysteresis-Free
PTAA/FA0.83Cs0.17Pb(I0.83Br0.17)3/C60 Perovskite Solar Cells Gives an Ion Mobility
of ∼3 × 10–7 Cm2/(V s), 2 Orders of Magnitude Faster than Reported for Metal-Oxide-
Contacted. J. Am. Chem. Soc. 2018, 140 (40), 12775–12784.
https://doi.org/10.1021/jacs.8b04405.
(50) Bag, M.; Renna, L. A.; Adhikari, R. Y.; Karak, S.; Liu, F.; Lahti, P. M.; Russell, T. P.;
Tuominen, M. T.; Venkataraman, D. Kinetics of Ion Transport in Perovskite Active
Layers and Its Implications for Active Layer Stability. J. Am. Chem. Soc. 2015, 137 (40),
13130–13137. https://doi.org/10.1021/jacs.5b08535.
(51) Hoque, M. N. F.; Islam, N.; Li, Z.; Ren, G.; Zhu, K.; Fan, Z. Ionic and Optical Properties of
Methylammonium Lead Iodide Perovskite across the Tetragonal–Cubic Structural
Phase Transition. ChemSusChem 2016, 9 (18), 2692–2698.
https://doi.org/10.1002/cssc.201600949.
(52) Yun, J. S.; Seidel, J.; Kim, J.; Soufiani, A. M.; Huang, S.; Lau, J.; Jeon, N. J.; Seok, S. Il;
Green, M. A.; Ho-Baillie, A. Critical Role of Grain Boundaries for Ion Migration in
Formamidinium and Methylammonium Lead Halide Perovskite Solar Cells. Adv.
Energy Mater. 2016, 6 (13), 1600330. https://doi.org/10.1002/aenm.201600330.
(53) Lyubomirsky, I.; Rabinal, M. K.; Cahen, D. Room-Temperature Detection of Mobile
Impurities in Compound Semiconductors by Transient Ion Drift. J. Appl. Phys. 1997, 81
(10), 6684–6691. https://doi.org/10.1063/1.365563.
(54) Almora, O.; Aranda, C.; Mas-Marzá, E.; Garcia-Belmonte, G. On Mott-Schottky Analysis
Interpretation of Capacitance Measurements in Organometal Perovskite Solar Cells.
Appl. Phys. Lett. 2016, 109 (17), 173903. https://doi.org/10.1063/1.4966127.
(55) Heo, S.; Seo, G.; Lee, Y.; Lee, D.; Seol, M.; Lee, J.; Park, J.-B.; Kim, K.; Yun, D.-J.; Kim, Y.
S.; et al. Deep Level Trapped Defect Analysis in CH 3 NH 3 PbI 3 Perovskite Solar Cells
by Deep Level Transient Spectroscopy. Energy Environ. Sci. 2017, 10 (5), 1128–1133.
https://doi.org/10.1039/C7EE00303J.
(56) Sanchez, R. S.; Gonzalez-Pedro, V.; Lee, J. W.; Park, N. G.; Kang, Y. S.; Mora-Sero, I.;
Bisquert, J. Slow Dynamic Processes in Lead Halide Perovskite Solar Cells.
Characteristic Times and Hysteresis. J. Phys. Chem. Lett. 2014, 5 (13), 2357–2363.
https://doi.org/10.1021/jz5011187.
(57) Gottesman, R.; Haltzi, E.; Gouda, L.; Tirosh, S.; Bouhadana, Y.; Zaban, A.; Mosconi, E.;
De Angelis, F. Extremely Slow Photoconductivity Response of CH3NH 3PbI3
Perovskites Suggesting Structural Changes under Working Conditions. J. Phys. Chem.
Lett. 2014, 5 (15), 2662–2669. https://doi.org/10.1021/jz501373f.
SUPPLEMENTARY INFORMATION FOR

Quantification of Ion Migration in CH3NH3PbI3 Perovskite

Solar Cells by Transient Capacitance Measurements

Moritz H. Futscher1, Ju Min Lee1, Lucie McGovern1, Loreta A. Muscarella1, Tianyi Wang1,

Muhammad Irfan Haider2, Azhar Fakharuddin2, Lukas Schmidt-Mende2 and Bruno Ehrler1*

1. Center for Nanophotonics, AMOLF, Science Park 104,

1098 XG Amsterdam, The Netherlands

2. Department of Physics, University of Konstanz, Universitätsstraße 10,

78457 Konstanz, Germany

AUTHOR INFORMATION

Corresponding Author

* ehrler@amolf.nl
S1 EFFECT OF CONTACT LAYERS

By measuring the current-voltage characteristic at different scan rates, a distinction can be

made between capacitive and non-capacitive hysteresis.1 Figure S1 illustrates the effect of

capacitive hysteresis in perovskite solar cells using TiO2 as an electron transport material. This

effect is attributed to the accumulation of both ionic and electronic charges at the

TiO2/perovskite interface in a highly reversible manner resulting in a double-layer structure.1

This effect is not observed in the inverted structure using NiOx.

Figure S1. Dark current-voltage characteristics of diodes based on (a) TiO2/MAPbI3/spiro-OMeTAD and (b)

NiOx/MAPbI3/C60 illustrating different hysteretic effects as a function of the scan rate.

Figure S2 shows the capacitance versus frequency at different temperatures for the inverted

and the regular perovskite solar cell structure. At the temperature range between 180 and

300 K no phase transition of the MAPbI3 layer is to be expected. However, the regular devices

structure shows a strong change in the high-frequency region at these temperatures,

indicating dielectric contributions of contact layers play an important role. For the

measurement of transient ion-drift, such interfacial effects shall be minimized.

Figure S2. Impedance spectroscopy measured at different temperatures in the dark at 0 V with an AC

perturbation of 20 mV of (a) an inverted (NiOx/MAPbI3/C60) and (b) a regular (TiO2/C60/MAPbI3/spiro-OMeTAD)

and perovskite solar cell.

S2 DARK AND LIGHT CURRENT-VOLTAGE CHARACTERISTICSs

Inverted perovskite solar cells have been shown to have only little hysteresis when

illuminated,2–5 but hysteresis may still be present in the dark. Figure S3 shows the current-

voltage characteristic of an inverted perovskite solar cell measured with a scan rate of

300 mV/second under 1-Sun from a solar simulator (Oriel 92250A) using a Keithley 2636A

source-measure unit after 15 minutes light soaking. During this measurement, the sample

was masked and placed in nitrogen inside an air-tight sample holder. Figure S4 shows current-

voltage characteristics of an inverted perovskite solar cell in the dark, measured between 180

and 330 K. We note that there is a significant difference between the current-voltage

hysteresis measured in the dark and under illumination. This difference may be caused by

photo-induced ion migration, as it has been shown that the activation energy for ion migration

is reduced by illumination.6,7 When the perovskite solar cell is cooled, the current-voltage

hysteresis is reduced and almost vanishes at 180 K.

Figure S3. Current-voltage characteristics of an inverted perovskite solar cell (NiOx/MAPbI3/C60) measured at

1-Sun at room temperature.

Figure S4. Temperature dependent current-voltage hysteresis measured in the dark at (a) 180 K, (b) 210 K,

(c) 240 K, (d) 270 K, (e) 300 K, and (f) 330 K.

S3 MOTT-SCHOTTKY CHARACTERISTICS

Figure S5 shows the capacitance as a function of voltage measured at 10 kHz, where the

measured capacitance corresponds to the geometric capacitance and the series resistance can

be neglected (see Figure 1c). We observe a plateau at low voltage, which indicates full
depletion under short-circuit conditions. In such a case of full depletion, the geometrical

capacitance can be related to the perovskite permittivity. Assuming a parallel plate capacitor

with the thickness of the perovskite layer, we obtain a permittivity of 15.3 for the perovskite

layer, averaged over the measured temperatures, somewhat lower than the calculated value

of 24.1 from electronic structure calculation in the absence of molecular reorientations. 8

Figure S5. Mott-Schottky characteristics of an inverted perovskite solar cell (NiOx/MAPbI3/C60) measured at 300

and 180 K in the dark with an AC perturbation of 10 mV at 10 kHz.

When a voltage 𝑉 is applied in forward direction, the depletion capacitance 𝐶𝐷 is increased.

This increase in capacitance is correlated to a decrease in depletion-layer width. The

depletion capacitance as a function of applied voltage can be approximated by the Mott-

Schottky relation as

𝑞 𝜀0 𝜀 𝑁
𝐶𝐷 = √
2 (𝑉𝐵 − 𝑉)

where 𝑞 is the elementary charge, 𝜀0 the vacuum permittivity, 𝜀 the perovskite permittivity,

𝑁 the doping density, and 𝑉𝐵 the built-in potential.9 From the 𝐶 −2 (𝑉) plot we obtain a built-

in potential of 0.92 V and a doping density of 7.0 x 1016 cm-3. The slope of the Mott-Schottky

plot furthermore suggests a p-type MAPbI3 layer. Theoretical calculations predict that the

p-type doping of MAPbI3 originates from negatively charged Pb2+ and MA+ vacancies, where
positively charged I- vacancies might result in n-type doping.10 Note that the Mott-Schottky

analysis is only meaningful when the depletion capacitance can be clearly identified.11 Since

the ionic capacitance contribution dominates the depletion capacitance at high

temperatures, we performed the Mott-Schottky analysis at 180 K.

For the calculation of the concentration of mobile ions, a constant doping density of

1 x 1017 cm-3 is assumed for all devices. Since our obtained doping concentration at 180 K is

close to typical vales at room temperature (1 x 1017 cm-3),11 we believe that the measured

temperature window lies within the extrinsic region in which the doping density is

reasonably constant.

S4 ION REDISTRIBUTION

Figure S6. (a) Amplitude of the capacitance transient when applying a voltage of 0.4 V as a function of filling

pulse duration. The grey line indicates the filling pulse duration used for the transient ion-drift measurements.

(b) Background capacitance and capacitance transient after applying a filling pulse of 0.4 V for 1 second at

300 K. The capacitance transient was measured after the background capacitance had reached a steady state.

S5 TRANSIENT ION-DRIFT

Transient ion-drift is a powerful method to quantify mobile ions in perovskite materials with

very high accuracy in a fast and non-destructive way. By measuring the capacitance
transient, the technique is uniquely able to distinguish between mobile cations and anions,

with concentrations as low as 0.01% of the doping density.

Transient ion-drift measures the change of capacitance over time under a constant bias.

Assuming thermal diffusion to be negligible against drift and that the total ion concentration is

conserved, the ion diffusion equation is given by:

𝛿𝑁𝑖𝑜𝑛 𝛿𝑁𝑖𝑜𝑛
= 𝜇𝐸
𝛿𝑡 𝛿𝑥

where 𝜇 is the ion mobility and 𝐸 is the electric field. Assuming that the electric field varies
𝑥
linearly across the depletion region, the electric field can be written as 𝐸(𝑥) = 𝐸0 (1 − 𝑊 ),
𝐷

where 𝑊𝐷 is the depletion width.12 Assuming that the ions are initially uniformly distributed,

the capacitance transient induced by ion drift is given by:

𝑡
𝑁𝑖𝑜𝑛 (𝑡) = 𝑁𝑖𝑜𝑛0 exp
𝜏
𝑊 𝐷𝑞
where 𝑁𝑖𝑜𝑛0 is the initial ion concentration and 𝜏 = 𝜇𝐸𝐷 . Using the Einstein relation (𝜇 = 𝑘 𝑇)
𝑀 𝐵

𝑞 𝑊𝐷 𝑁
and expressing the electric field as a function of the doping density 𝑁 as 𝐸0 = , where 𝑞
𝜀0 𝜀

is the elementary charge, 𝜀0 is the vacuum permittivity, and 𝜀 is the perovskite permittivity,

the time constant can be written as:

𝑘𝐵 𝑇 𝜀0 𝜀
𝜏=
𝑞2𝐷 𝑁
𝐸
− 𝐴
where 𝑘𝐵 is the Boltzmann constant and 𝑇 the temperature. 𝐷 = 𝐷0 𝑒 𝑘𝐵 𝑇
is the ion

diffusion-coefficient where 𝐷0 is the attempt-to-escape frequency for ion migration and 𝐸𝐴

the activation energy.

To quantify ion migration within the perovskite layer, one has to carefully chose a perovskite

device structure to avoid capacitive hysteresis, interfacial effects, and ion migration within

the transport layers (see also section S1). It is not possible to distinguish between mobile
ions within the perovskite layer and mobile ions within the other layers of the device

structure. It is thus important to avoid, as far as possible, mobile ions within the other layers

of the device structure. For measuring capacitance transients, one furthermore has to

carefully chose an AC measurement frequency to measure the capacitance change of the

perovskite layer. Most commercially available capacitance transient measurement systems

use a fixed AC measurement frequency of 1 MHz. Measuring capacitance transients with

such a high frequency requires devices with a very low series resistance. For most

perovskite-based devices, however, the series resistance of the transparent conductive

oxide starts to dominate the impedance response at such high frequencies (see also

section S6). We thus conclude that an AC measurement frequency of 1 MHz is not suitable

for measuring capacitance transients of perovskite devices in the majority of cases.

S6 FITTING CAPACITANCE TRANSIENTS

Figure S7 shows measured capacitance transients from 180 to 350 K with steps of 10 K

measured at 0 V after applying a voltage pulse of 0.4 V for 1 second (identical to Figure 3a of

the main text). The grey dotted lines in Figure S7(a) indicate the modelled capacitance decay

due to ionic drift according to the values in Table S1. For the Arrhenius plot we limited the

data analysis to temperatures where the number of exponentials to use was evident from the

scan and the fit quality was good (as indicated by the colors in Figure 3c of the main text).

Figure S7(b) to (e) exemplary show fits to capacitance decays at 200, 240, 290, and 340 K.
Figure S7. (a) Measured capacitance transients together with the modelled capacitance decay due to ionic drift

of A1, C1, and C2 shown as black dotted lines. Exemplary fits of measured capacitance decay at 200 (b), 240 (c),

290 (d), and 340 K (d). One exponential decay function is used to fit (b) and (d) and two exponential decay

functions to fit (c) and (e). Grey points are experimentally measured data and blue lines are obtained fits.

S7 IMPEDANCE SPECTROSCOPY

In a perovskite solar cell, mobile ions can migrate through the transport layer towards the
13
electrodes. As metals are prone to reacting with I- ion, this ion migration can induce an

additional series resistance related to contact degradation.14 It is thus important to carefully

choose an AC measuring frequency for which the impedance response corresponds to the

capacitive character of the device, as oppose to the resistances.

Figure S8 shows the impedance response of the perovskite solar cells at different

temperatures. A phase angle close to −90° indicate that the impedance corresponds to the

capacitance of the device. At frequencies above 10 kHz, the phase angle increases while the

modulus approaches the series resistance of the device, indicating that the capacitance

response is governed by the series resistance. We therefore measure the capacitance as a

function of voltage and time with an AC frequency of 10 kHz, where the impedance response

is dominated by the capacitance of the device over the whole temperature range of interest.

Figure S8. Impedance spectroscopy of an inverted perovskite solar cell (NiOx/MAPbI3/C60) measured at 0 V with

an AC perturbation of 20 mV in the dark, separated in (a) modulus and (b) phase angle.

S8 TRANSIENT ION-DRIFT VERSUS DEEP-LEVEL TRANSIENT SPECTROSCOPY

To distinguish between ion diffusion and electronic effects such as trapping and de-trapping,

we compare the rise and decay time of capacitance following the forward bias and returning

to short circuit conditions.12 For mobile ions, it is expected that the time required to lead to a

uniform ion distribution after applying a forward bias is longer than the time required for ions

to drift back to the interfaces after removal of the forward bias. In contrast, for traps the

capture rate is much higher than the emission rate. Figure S9 shows the measured capacitance

transient of an inverted perovskite solar cell measured at 210 K, showing that the measured

capacitive transient is due to the diffusion of mobile ions.

Figure S9. Capacitance transient of an inverted perovskite solar cell (NiOx/MAPbI3/C60) measured in the dark (a)

while applying a voltage pule of 0.4 V and (b) at short circuit, after removing the voltage bias. The dashed lines

are fits obtained using an exponential decay function with the timescale indicated in the inset.

S9 STATISTICS AND REPRODUCIBILITY ACROSS LABORATORIES

Many examples in the literature show that the performance of perovskite solar cells depend

heavily on the fabrication, even in the same laboratory. Also, ions are presumably affected

by, and affecting degradation. To study this, we measure seven different devices and

compare the transient ion-drift response.

Device #1 corresponds to the device shown in Figure 1 and Figure 3 the main text. Device #2

corresponds to a device fabricated in the same way as device #1. Device #3 represent a poor

performing device. Devices #4, #5, and #6 are devices fabricated in the laboratory of the

University of Konstanz. All devices have the same device structure (NiOx/MAPbI3/C60), but

the MAPbI3 layer has an average thickness of 105 nm in devices #1, #2 and #3 and 275 nm in

devices #4, #5 and #6. The values obtained for activation energy, diffusion coefficient, and

concentration for mobile ions for the measured samples are summarized in Table S1 – S6.

Note that we could not resolve the slow MA+ species in the devices manufactured at the

University of Konstanz.
 A typical current-voltage characteristic curve together with an external quantum efficiency

spectrum of an inverted perovskite solar cell fabricated in Konstanz is shown in Figure S10.

We furthermore note that in devices with an average MAPbI3 thickness of 275 nm, we

observed an initial capacitance decay at high temperatures related to the redistribution of

ions inside the depletion layer (see Figure S11).15 For DLTS, such an initial decay in

capacitance is not expected, as the drift of free charge carriers out of the depletion layer is

much faster than the emission rate. The capacitance change due to transient ion-drift,

however, can be due to a combination of both the redistribution inside the depletion layer

and the drift of mobile ions towards the contacts.

2
Table S1. Characteristics of mobile ions in device #1 with a short-circuit current density of 17.7 mA/cm , an
open-circuit voltage of 0.98 V, a fill factor of 56%, and a power-conversion efficiency of 9.6%.

A1 C1 C2

Migrating ion species I- MA+ MA+

Charge negative positive positive
Concentration (cm-3) (1.7 ± 0.1) x 1015 (2.5 ± 0.1) x 1016 (1.1 ± 0.1) x 1016
Phase structure tetragonal tetragonal cubic cubic
Activation energy (eV) 0.37 ± 0.01 0.95 ± 0.02 0.28 ± 0.01 0.43 ± 0.01
Diffusion coefficient
(3.2 ± 1.4) x 10-9 (1.8 ± 2.4) x 10-12 (4.7 ± 2.7) x 10-12 (4.4 ± 3.7) x 10-13
at 300 K (cm2 s-1)

2
Table S2. Characteristics of mobile ions in device #2 with a short-circuit current density of 13.4 mA/cm , an
open-circuit voltage of 0.88 V, a fill factor of 48%, and a power-conversion efficiency of 5.7%.

A1 C1 C2

Migrating ion species I- MA+ MA+

Charge negative positive positive
Concentration (cm-3) (2.1 ± 0.1) x 1015 (3.9 ± 0.1) x 1015 (2.4 ± 0.1) x 1015
Phase structure tetragonal tetragonal cubic cubic
Activation energy (eV) 0.39 ± 0.01 0.40 ± 0.01 0.23 ± 0.02 0.04 ± 0.03
Diffusion coefficient
2 -1 (11.6 ± 2.5) x 10-9 (3.4 ± 2.1) x 10-12 (6.4 ± 11.1) x 10-12 (1.6 ± 3.7) x 10-12
at 300 K (cm s )
 2
Table S3. Characteristics of mobile ions in device #3 with a short-circuit current density of 3.6 mA/cm , an
open-circuit voltage of 0.74 V, a fill factor of 37%, and a power-conversion efficiency of 1.0%.
A1 C1 C2

Migrating ion species I- MA+ MA+

Charge negative positive positive
Concentration (cm-3) (2.1 ± 0.1) x 1015 (3.4 ± 0.1) x 1015 (1.7 ± 0.1) x 1015
Phase structure tetragonal tetragonal cubic cubic
Activation energy (eV) 0.23 ± 0.01 0.26 ± 0.01 0.62 ± 0.01 0.71 ± 0.01
Diffusion coefficient
(2.2 ± 0.8) x 10-9 (11.5 ± 2.1) x 10-12 (20.1 ± 16.6) x 10-12 (2.8 ± 2.7) x 10-12
at 300 K (cm2 s-1)

2
Table S4. Characteristics of mobile ions in device #5 with a short-circuit current density of 14.9 mA/cm , an
open-circuit voltage of 0.96 V, a fill factor of 62%, and a power-conversion efficiency of 8.8%.

A1 C1

Migrating ion species I- MA+

Charge negative positive
Concentration (cm-3) (4.3 ± 0.3) x 1012 (1.6 ± 0.1) x 1016
Phase structure tetragonal tetragonal cubic
Activation energy (eV) 0.28 ± 0.09 1.91 ± 0.06 0.94 ± 0.06
Diffusion coefficient
(0.2 ± 1.8) x 10-9 (5.3 ± 23.2) x 10-15 (1.7 ± 7.1) x 10-13
at 300 K (cm2 s-1)

2
Table S5. Characteristics of mobile ions in device #6 with a short-circuit current density of 16.1 mA/cm , an
open-circuit voltage of 0.97 V, a fill factor of 66%, and a power-conversion efficiency of 10.2%.

A1 C1

Migrating ion species I- MA+

Charge negative positive
Concentration (cm-3) (7.0 ± 1.4) x 1013 (1.8 ± 0.2) x 1016
Phase structure tetragonal tetragonal cubic
Activation energy (eV) 0.16 ± 0.05 0.96 ± 0.07 0.25 ± 0.03
Diffusion coefficient
(0.06 ± 0.26) x 10-9 (2.0 ± 11.3) x 10-13 (2.5 ± 5.9) x 10-12
at 300 K (cm2 s-1)
 2
Table S6. Characteristics of mobile ions in device #7 with a short-circuit current density of 17.2 mA/cm , an
open-circuit voltage of 1.02 V, a fill factor of 70%, and a power-conversion efficiency of 12.1%.

A1

Migrating ion species I-

Charge negative
Concentration (cm-3) (6.9 ± 0.8) x 1014
Phase structure tetragonal
Activation energy (eV) 0.32 ± 0.05
Diffusion coefficient
(1.2 ± 5.7) x 10-9
at 300 K (cm2 s-1)

Figure S10. (a) Current-voltage characteristic curve and (b) external quantum efficiency (EQE) spectrum of an
2
inverted perovskite solar cell fabricated in Konstanz with a short-circuit current density of 19.9 mA/cm , an

open-circuit voltage of 1.04 V, a fill factor of 63%, and a power-conversion efficiency of 13.0%. The integrated
2
current density of the EQE spectrum with the AM1.5G solar spectrum is 21.1 mA/cm , very close to the value

obtained from the current-voltage measurements.

Figure S11. Ion-drift-induced capacitance transient measured at (a) 320 and at (b) 340 K in addition to an initial
15
capacitance decrease related to the redistribution of ions inside the depletion layer. Grey points are

experimentally measured data and blue lines are obtained fits.

S10 DIODE FABRICATION

Device #1, device #2, and #3 were fabricated as described in the main text, with the

exception of device #2 which was fabricated using a NiOx precursor solution of 0.3M that was

spun on the cleaned ITO glass at 4000 rpm for 15 seconds.

Device #4, device #5, and #6 were fabricated as described hereafter: Indium tin oxide (ITO)

glass substrates were cleaned by ultrasonication for 20 minutes subsequently in detergent in

deionized water, deionized water, acetone, and isopropanol, followed by UV ozone

treatment for 15 minutes. Nickel oxide (NiOx) precursor solution (0.5 M nickel(II)

Acetylacetonate (Aldrich) in ethanol and Conc. HCl) filtered with PTFE 0.45 µm was spun on

the cleaned ITO glass at 5000 rpm for 30 seconds, dried at 100°C for 1 minute and annealed

at 320 °C for 45 minutes with a slow cooling rate.

The MAPbI3 perovskite precursor solution was prepared by mixing of total 1.5 M of

methylammonium iodide (MAI, Solaronix) and lead(II) iodide (PbI2, TCI) with 1:1 molar ratio

dissolved in N, N-dimethylformamide (anhydrous, Aldrich) and DMSO for 3 hours at 60 °C.

50 μL of the MAPbI3 precursor solution filtered through a 0.45 μm sized PTFE membrane

was spun onto NiOx coated substrates at 4000 rpm for 50 seconds in a nitrogen-filled glove
box. 10 seconds after the beginning of the rotation, 300 μL of Diethyl ether anti-solvent

(anhydrous, Aldrich) was quickly dropped onto the substrate. After the MAPbI3 spinning

process, the substrates were annealed at 100 °C for 3 minutes. 45 nm of C60 (0.2 Å/s rate)

was deposited on top of the MAPbI3 layer by thermal sublimation at pressures below

8 × 10-6 mbar. A thin layer of bathocuproine (99.99%, Aldrich) dissolved in ethanol

(0.5 mg/ml) was then spun on top of the C60 layer with 6000 rpm for 15 seconds. Finally,

100 nm of silver (1 Å/s) was deposited by thermal evaporation at pressures below

8 × 10-6 mbar.

References

(1) Garcia-Belmonte, G.; Bisquert, J. Distinction between Capacitive and Noncapacitive

Hysteretic Currents in Operation and Degradation of Perovskite Solar Cells. ACS
Energy Letters. American Chemical Society October 14, 2016, pp 683–688.
https://doi.org/10.1021/acsenergylett.6b00293.
(2) Deng, L. L.; Xie, S. Y.; Gao, F. Fullerene-Based Materials for Photovoltaic Applications:
Toward Efficient, Hysteresis-Free, and Stable Perovskite Solar Cells. Advanced
Electronic Materials. December 5, 2017, p 1700435.
https://doi.org/10.1002/aelm.201700435.
(3) Kim, H.-S.; Jang, I.-H.; Ahn, N.; Choi, M.; Guerrero, A.; Bisquert, J.; Park, N.-G. Control
of I-V Hysteresis in CH3NH3PbI3 Perovskite Solar Cell. J. Phys. Chem. Lett. 2015, 6 (22),
4633–4639. https://doi.org/10.1021/acs.jpclett.5b02273.
(4) Zhu, Z.; Bai, Y.; Zhang, T.; Liu, Z.; Long, X.; Wei, Z.; Wang, Z.; Zhang, L.; Wang, J.; Yan,
F.; et al. High-Performance Hole-Extraction Layer of Sol-Gel-Processed NiO
Nanocrystals for Inverted Planar Perovskite Solar Cells. Angew. Chemie - Int. Ed. 2014,
53 (46), 12571–12575. https://doi.org/10.1002/anie.201405176.
(5) Chen, W.; Wu, Y.; Yue, Y.; Liu, J.; Zhang, W.; Yang, X.; Chen, H.; Bi, E.; Ashraful, I.;
Grätzel, M.; et al. Efficient and Stable Large-Area Perovskite Solar Cells with Inorganic
Charge Extraction Layers. Science 2015, 350 (6263), 944–948.
https://doi.org/10.1126/science.aad1015.
(6) Mosconi, E.; Meggiolaro, D.; Snaith, H. J.; Stranks, S. D.; De Angelis, F. Light-Induced
Annihilation of Frenkel Defects in Organo-Lead Halide Perovskites. Energy Environ. Sci.
2016, 9 (10), 3180–3187. https://doi.org/10.1039/C6EE01504B.
(7) DeQuilettes, D. W.; Zhang, W.; Burlakov, V. M.; Graham, D. J.; Leijtens, T.; Osherov, A.;
Bulović, V.; Snaith, H. J.; Ginger, D. S.; Stranks, S. D. Photo-Induced Halide
Redistribution in Organic-Inorganic Perovskite Films. Nat. Commun. 2016, 7, 11683.
https://doi.org/10.1038/ncomms11683.
(8) Brivio, F.; Butler, K. T.; Walsh, A.; van Schilfgaarde, M. Relativistic Quasiparticle Self-
Consistent Electronic Structure of Hybrid Halide Perovskite Photovoltaic Absorbers.
 Phys. Rev. B 2014, 89 (15).
(9) Sze. Physics of Semiconductor Devices; Wiley-Interscience, 2014; Vol. 10.
https://doi.org/10.1007/978-3-319-03002-9.
(10) Frolova, L. A.; Dremova, N. N.; Troshin, P. A. The Chemical Origin of the P-Type and n-
Type Doping Effects in the Hybrid Methylammonium-Lead Iodide (MAPbI3) Perovskite
Solar Cells. Chem. Commun. 2015, 51 (80), 14917–14920.
https://doi.org/10.1039/c5cc05205j.
(11) Almora, O.; Aranda, C.; Mas-Marzá, E.; Garcia-Belmonte, G. On Mott-Schottky Analysis
Interpretation of Capacitance Measurements in Organometal Perovskite Solar Cells.
Appl. Phys. Lett. 2016, 109 (17), 173903. https://doi.org/10.1063/1.4966127.
(12) Heiser, T.; Mesli, A. Determination of the Copper Diffusion Coefficient in Silicon from
Transient Ion-Drift. Appl. Phys. A Solids Surfaces 1993, 57 (4), 325–328.
https://doi.org/10.1007/BF00332285.
(13) Yin, X.; Yao, Z.; Luo, Q.; Dai, X.; Zhou, Y.; Zhang, Y.; Zhou, Y.; Luo, S.; Li, J.; Wang, N.; et
al. High Efficiency Inverted Planar Perovskite Solar Cells with Solution-Processed NiOx
Hole Contact. ACS Appl. Mater. Interfaces 2017, 9 (3), 2439–2448.
https://doi.org/10.1021/acsami.6b13372.
(14) Guerrero, A.; You, J.; Aranda, C.; Kang, Y. S.; Garcia-Belmonte, G.; Zhou, H.; Bisquert,
J.; Yang, Y. Interfacial Degradation of Planar Lead Halide Perovskite Solar Cells. ACS
Nano 2016, 10, 218–224. https://doi.org/10.1021/acsnano.5b03687.
(15) Heiser, T.; Weber, E. Transient Ion-Drift-Induced Capacitance Signals in
Semiconductors. Phys. Rev. B 1998, 58 (7), 3893–3903.
https://doi.org/10.1103/PhysRevB.58.3893.