Migration of cations induces reversible performance losses over day/night cycling in perovskite solar cells

TL;DR: In this article, the authors demonstrate that ionic defects, migrating on timescales significantly longer (above 103 s) than what has so far been explored (from 10−1 to 102 s), abate the initial efficiency by 10−15% after several hours of operation at the maximum power point.

Abstract: Perovskites have been demonstrated in solar cells with a power conversion efficiency of well above 20%, which makes them one of the strongest contenders for next generation photovoltaics. While there are no concerns about their efficiency, very little is known about their stability under illumination and load. Ionic defects and their migration in the perovskite crystal lattice are some of the most alarming sources of degradation, which can potentially prevent the commercialization of perovskite solar cells (PSCs). In this work, we provide direct evidence of electric field-induced ionic defect migration and we isolate their effect on the long-term performance of state-of-the-art devices. Supported by modelling, we demonstrate that ionic defects, migrating on timescales significantly longer (above 103 s) than what has so far been explored (from 10−1 to 102 s), abate the initial efficiency by 10–15% after several hours of operation at the maximum power point. Though these losses are not negligible, we prove that the initial efficiency is fully recovered when leaving the device in the dark for a comparable amount of time. We verified this behaviour over several cycles resembling day/night phases, thus probing the stability of PSCs under native working conditions. This unusual behaviour reveals that research and industrial standards currently in use to assess the performance and the stability of solar cells need to be adjusted for PSCs. Our work paves the way for much needed new testing protocols and figures of merit specifically designed for PSCs.

Summary

Introduction

• Perovskite solar cells (PSCs) have the potential to become a new generation of photovoltaics with the shortest energy payback time and lowest CO2 emission factor among existing technologies.
• This impressive improvement of the PCE has not been matched by an equal advancement in the knowledge of the performance losses under standard working conditions (illumination and load).
• 10, 11 Conversely, prolonged exposure to solar cell operational temperatures (above 50 °C) can cause severe degradation, which cannot be avoided by sealing the PSCs.
• 25-27 Several studies indicated that, regardless of particular architecture and constituents within the PSCs, X defects migrate and reversibly accumulate within the perovskite lattice in narrow Debye layers at the interfaces with the charge selective contacts.
• The authors work paves the way towards developing specific testing protocols, definition of new figures of merits and calculation of energy payback time that are needed to characterize PSCs.

Results and discussion

• To study the impact of the long-term ion migration on device performance and stability, the authors prepared state-of-the-art PSCs, using the mixed halide-cation perovskite composition CH3NH3/CH(NH2)2 Pb Br/I and the antisolvent deposition method on mesoporous TiO2 substrates,56 which enabled the realization of power conversion efficiencies above 20% (see device characterization in SI).
• To isolate regime I from the subsequent degradation (regime II), the maximum power point tracking (MPPT) for a device C was stopped after only 5 hours and repeated periodically after leaving the device resting in dark for a varying number of hours (Figure 1b).
• The authors found that the experimental trend in Figure 3a was best reproduced by the model that accounts for halide vacancy migration, but does not directly account for the slowly moving cation vacancies even on the 300s settling timescale.
• To provide evidence that cation vacancies are effectively mobile the authors measured the current transient dynamics at short circuit after preconditioning the device at either forward (0.85 V) or reverse (-0.3 V) bias (Figure 4a).
• The authors have also proven, that both halide and cation vacancies are mobile (albeit the latter are considerably slower than the former) and their distribution in the perovskite layer can considerably affect charge extraction and, in consequence, PCE of the device.

Conclusions

• The authors investigated the impact of the ionic defect migration on the performance and stability of state-of-the-art perovskite solar cells (PSCs).
• The authors provide direct evidence of halide migration within the perovskite as a result of applied electric field.
• The authors show, that the accumulation of cation vacancies at the electron contact induces reversible performance losses that abate the initial efficiency of state-of-the-art PSCs by about 10-15% over several hours of operation at the maximum power point.
• The initial efficiency is fully recovered when leaving the device in dark for a comparable amount of time.
• The authors work provides indications for the much needed new testing protocols and figures of merits specifically designed for PSCs.

Figures

Table 1. Fitting parameters of the exponential decay functions used in Figure 1. Device A and B curves were fitted with a double exponential decay function J = A1 exp(-x/t1) + A2 exp(x/t2) + J0, where t is the time constant, Jo is the residual power output and A is the preexponential factor. Device C curves were fitted with a single exponential decay function J = A1 exp(-x/t1) + J0.

Figure 1. Maximum power output tracking for 3 identically prepared perovskite solar cells (device A, B and C) measured under UV-filtered 1 sun equivalent light. Devices were continuously kept at the maximum power point by the standard “perturb and observe” method. The efficiency of all freshly made devices was above 20% (see SI) and it dropped to around 17-18% after 2-3 weeks, when the stability data were recorded. Data were normalized to the maximum value, which was usually reached within several minutes of tracking. a Devices A and B were continuously tracked for over 100 hours. b Device C was cyclically tracked 4 times for 5 hours and it was left in dark at open circuit in between the consecutive measurements. c Experimental data were fitted to an exponential decay (single or double) and the fitting parameters are reported in Table 1.

Figure 3. a Imaginary component frequency spectra of the current response to the light modulated (10% of the stationary value) ~1 sun equivalent intensity under different applied voltage biases. Data were normalized to the maximum value corresponding to the peak between 10 and 100 kHz. The inset shows the J-V curve recorded for the same devices at 100 mV s-1 scan rate. b Model calculation of the capacitance per unit surface area at the edge of the perovskite layer as a function of the forward applied voltage bias, assuming the negative defects (cation vacancies) as non-mobile, slow or fast, compared to the timescale of the spectroscopic oscillations in the low frequency region (10-1 - 103 Hz).

Figure 5. Schematics of the evolution of the ion distribution within the perovskite layer sandwiched between the electron and hole selective contacts under solar cell working conditions: a initial condition, b non-stabilized condition on the timescale of minutes and c the stabilized condition on the timescale of hours.

Figure 4. a Current transient dynamics collected from the same device pre-condition for about 20 minutes under forward (0.85 V) or reverse (-0.3 V) bias and cooled to -20 °C before to abruptly switch to short circuit condition. b Modelled current transient dynamics as predicted by the evolution of the potential (left panel - considering only halide vacancy migration, and right panel - considering fast halide and slow cation vacancy migration). c Modelled maximum power conversion efficiency (with fast halide and slow cation vacancy migration) over 4 cycles of light and dark shows similar (non-quantitative) reversible performance losses to the experiment in Figure 1.

Figure 2. a Schematics of the sample prepared for the biasing experiment, consisting of two electronically-separated pixels on the same substrate. Both pixels were heated to 70 °C in N2 atmosphere. Only one pixel was cyclically biased at +2 and –2 V every 30 minutes for 16 hours (biased pixel). b Top view image of the sample after the biasing (scale in millimetre). Time of flight secondary ion mass spectrometry (ToF-SIMS) depth profiles of the control (c) and biased (d) pixels showing the relative secondary ions intensity of iodine and bromine clusters across the film depth. The SnO- signals coming from the fluorine doped SnO2 (FTO glass) are used as a reference, to which all traces in the respective graph are normalized.

Full text

!
1
!
Migration of cations induces reversible performance losses over
day/night cycling in perovskite solar cells
Konrad Domanski,
1
Bart Roose,
2
Taisuke Matsui,
3
Michael Saliba,
1
Silver-Hamill Turren-
Cruz,
4
Juan-Pablo Correa-Baena,
4
Cristina Roldan Carmona,
5
Giles Richardson,
6
Jamie M.
Foster,
6
Filippo De Angelis,
8,9
James M. Ball,
10
Annamaria Petrozza,
10
Nicolas Mine,
11
Mohammad K. Nazeeruddin,
5
Wolfgang Tress,
1
Michael Grätzel,
1
Ullrich Steiner,
2
Anders
Hagfeldt,
4
Antonio Abate
1,2
*
1
Laboratory for Photonics and Interfaces, Institute of Chemical Sciences and Engineering,
École Polytechnique Fédérale de Lausanne, CH-1015-Lausanne, Switzerland.
2
Adolphe Merkle Institute, University of Fribourg, Chemin des Verdiers 4, CH-1700
Fribourg, Switzerland
3
Advanced Research Division, Panasonic Corporation,1006, (Oaza Kadoma), Kadoma City,
Osaka 571-8501, Japan.
4
Laboratory of Photomolecular Science, Institute of Chemical Sciences and Engineering,
École Polytechnique Fédérale de Lausanne, CH-1015-Lausanne, Switzerland.
5
Group for Molecular Engineering of Functional Materials, Institute of Chemical Sciences
and Engineering, École Polytechnique Fédérale de Lausanne, CH-1015-Lausanne,
Switzerland.
6
Mathematical Sciences, University of Southampton, UK (SO17 1BJ)
7
Department of Mathematics, University of Portsmouth, Portsmouth, UK (PO1 2UP).
8
Computational Laboratory for Hybrid/Organic Photovoltaics (CLHYO), CNR-ISTM, Via
Elce di Sotto 8, 06123 Perugia, Italy.
9
CompuNet, Istituto Italiano di Tecnologia, Via Morego 30, 16163 Genova, Italy.
10
Center for Nano Science and Technology @Polimi, Istituto Italiano di Tecnologia, via
Giovanni Pascoli 70/3, 20133 Milan, Italy
11
Laboratory for Nanoscale Materials Science, Empa, Swiss Federal Laboratories for Material
Science and Technology, Ueberlandstr. 129,8600 Duebendorf, Switzerland
*Corresponding author: AA antonioabate83@gmail.com Antonio.abate@unifr.ch

!
2
!
Keywords: Perovskite solar cell, hysteresis, photovoltaics, ionic migration, perovskite
stability, ToF-SIMS, depth-profile, ageing perovskite solar cells

!
3
!
Abstract
Perovskites have been demonstrated in solar cells with power conversion efficiency well
above 20%, which makes them one of the strongest contenders for the next generation
photovoltaics. While there are no concerns about their efficiency, very little is known about
their stability under illumination and load. Ionic defects and their migration in the perovskite
crystal lattice are one of the most alarming sources of degradation, which can potentially
prevent the commercialization of perovskite solar cells (PSCs). In this work, we provide
direct evidence of electric field-induced ionic defect migration and we isolate their effect on
the long-term performance of state-of-the-art devices. Supported by modelling, we
demonstrate that ionic defects, migrating on timescales significantly longer (above 10
3
s) than
what has so far been explored (from 10
-1
to 10
2
s), abate the initial efficiency by 10-15% after
several hours of operation at the maximum power point. Though these losses are not
negligible, we prove that the initial efficiency is fully recovered when leaving the device in
the dark for a comparable amount of time. We verified this behaviour over several cycles
resembling day/night phases, thus probing the stability of PSCs under native working
conditions. This unusual behaviour reveals, that research and industrial standards currently in
use to assess the performance and the stability of solar cells need to be adjusted for PSCs.
Our work paves the way towards much needed new testing protocols and figures of merit
specifically designed for PSCs.

!
4
!
Introduction
Perovskite solar cells (PSCs) have the potential to become a new generation of photovoltaics
with the shortest energy payback time and lowest CO
2
emission factor among existing
technologies.
1
In only a few years, an unprecedented progress in preparation procedures and
material compositions has delivered lab-scale devices that have now reached power
conversion efficiencies (PCEs) of up to 22.1%.
2
However, this impressive improvement of
the PCE has not been matched by an equal advancement in the knowledge of the performance
losses under standard working conditions (illumination and load).
3-7
So far, discussion around PSCs stability has mainly focused on oxygen,
8
water
5
and UV light
exposure
9
as causes of rapid performance degradation in PSCs. These extrinsic factors have
been associated with a number of degradation mechanisms that can be retarded using the
sealing technologies industrialised for organic electronics, which provide oxygen and
humidity barriers and protection against UV light.
10, 11
Conversely, prolonged exposure to
solar cell operational temperatures (above 50 °C) can cause severe degradation, which cannot
be avoided by sealing the PSCs. These, so called intrinsic losses, have been mostly associated
with the degradation of organic materials and metal contacts within PSCs.
3, 12, 13
Indeed,
significant progress has been made by replacing the organic components with their inorganic
counterparts and passivating the interfaces between the different layers composing the
device.
13-18
Nonetheless, temperature activated formation and migration of ionic defects
within the organic-inorganic ABX
3
perovskite lattice remains a potential source of instability
for perovskite photovoltaics.
19-22
Halide anion (X) vacancies have been calculated to show
the lowest formation energies,
23
with bromide vacancies being favoured over iodide.
24
Correspondingly, X vacancies (together with interstitial X) have been shown to be the most
mobile defects, followed by cation A and B vacancies.
25-27
Several studies indicated that,
regardless of particular architecture and constituents within the PSCs, X defects migrate and
reversibly accumulate within the perovskite lattice in narrow Debye layers at the interfaces
with the charge selective contacts.
19, 28-34
Depending on voltage and light bias conditioning,
accumulation of ions (and their vacancies) partially screens the built-in electric field and
possibly creates interfacial electronic trap states, which reduce the charge extraction
efficiency.
25, 30, 31, 34-43
Ion migration on timescales from 10
-1
to 10
2
s has been widely
investigated to explain the hysteresis of current density-voltage (J-V) curves.
36, 37, 40, 44-48
However, the impact of X and potentially A and/or B defect formation and migration on PSC
performance on timescales above 10
3
s, which are indicative of long-term stability, remains
unknown.
49
Little experimental evidence exists on this subject since separating reversible ion

!
5
!
migration from any non-reversible long-term degradation is complex in real device working
conditions, i.e. prolonged exposure to continuous light and voltage bias.
29, 39, 50-55
In this work, we provide direct evidence of electric field-induced ion migration and its
effects on the long-term performance of perovskite solar cells working under different loads.
Cooling in situ the active area of working PSCs, we are able to inhibit thermally induced,
non-reversible degradation, thereby exposing fully reversible performance losses. Within
several hours of operation at the maximum power point (MPP), the reversible losses abate a
significant fraction of the initial PCE, which is followed be a period of stabilization.
Supported by modelling and elemental depth profiling, we correlate the reversible
performance losses in PSCs to the migration of ion vacancies on timescales (above 10
3
s),
which are significantly longer than those explored so far (from 10
-1
to 10
2
s). These unusually
slow dynamics reveal that academic and industrial standards currently in use to assess the
performance and stability of solar cells need to be adjusted for PSCs, which exhibit
phenomena previously unknown to the photovoltaics community. Importantly, we show that
over natural day/night cycles, PSCs that reversibly degrade during the day recover overnight
to “start fresh” every morning. Our work paves the way towards developing specific testing
protocols, definition of new figures of merits and calculation of energy payback time that are
needed to characterize PSCs.

Frequently asked questions

Q1. What are the contributions in "Migration of cations induces reversible performance losses over day/night cycling in perovskite solar cells" ?

In this paper, Pascoli et al. investigated the role of ionic migration in perovskite solar cells and found that the most mobile ionic defects are found in the debye layers at the interfaces with the charge selective contacts. 

Q2. What is the reason for the slow decay in efficiency observed in PSCs?

The strong asymmetry in the activation energy for migration of halide and cation vacancies implies that stabilization times on the order of hours (not minutes as widely believed) are required for PSCs to reach a true steady-state working conditions. 

Q3. What is the effect of ion migration on the built-in electric field?

28-34 Depending on voltage and light bias conditioning, accumulation of ions (and their vacancies) partially screens the built-in electric field and possibly creates interfacial electronic trap states, which reduce the charge extraction efficiency. 

Q4. What is the effect of the slow decay of the cation vacancies on the device?

(c) for timescales longer than 103 s (i.e. hours), cation vacancies form an additional Debye layer at the interface with the electron selective contact, which in turn inhibits charge extraction from the device. 

Q5. What is the alarming source of degradation in perovskite solar cells?

Ionic defects and their migration in the perovskite crystal lattice are one of the most alarming sources of degradation, which can potentially prevent the commercialization of perovskite solar cells (PSCs). 

Q6. What is the potential of perovskite solar cells?

Perovskite solar cells (PSCs) have the potential to become a new generation of photovoltaics with the shortest energy payback time and lowest CO2 emission factor among existing technologies. 

Q7. What is the effect of the slow cation migration on the efficiency of a PSC?

When PSCs are exposed to real operating conditions, the slow cation migration is responsible for the reversible losses in the device on the timescale of hours. 

Q8. What is the main reason for the ionic defects in the perovskite la?

13-18 Nonetheless, temperature activated formation and migration of ionic defects within the organic-inorganic ABX3 perovskite lattice remains a potential source of instability for perovskite photovoltaics. 

Q9. What causes rapid performance degradation in perovskite solar cells?

So far, discussion around PSCs stability has mainly focused on oxygen,8 water5 and UV light exposure9 as causes of rapid performance degradation in PSCs. 

Q10. How long did the authors wait for the current to stabilize?

Under each biasing condition, the authors waited for the current to stabilize for about 20 minutes at 20 °C, before cooling the device to –20 °C and switching it abruptly to short circuit condition. 

Q11. What is the reason why identically prepared devices age differently?

Since the long-term degradation is a convolution of several mechanisms that may abruptly impact the performance,57 it is not surprising that identically prepared devices age differently. 

Q12. How do the authors determine the efficiency of state-of-the-art PSCs?

The authors show, that the accumulation of cation vacancies at the electron contact induces reversible performance losses that abate the initial efficiency of state-of-the-art PSCs by about 10-15% over several hours of operation at the maximum power point. 

Q13. How was the device kept under the same conditions?

During the experiment, the devices were kept under 1 sun-equivalent white LED illumination at MPP (around 0.85 V) and under N2 atmosphere. 

Q14. What are the extrinsic factors associated with the degradation of organic materials?

These extrinsic factors have been associated with a number of degradation mechanisms that can be retarded using the sealing technologies industrialised for organic electronics, which provide oxygen and humidity barriers and protection against UV light. 

Q15. What is the use of time of flight secondary ion mass spectrometry?

The authors made use of time of flight secondary ion mass spectrometry (ToF-SIMS) to measure the effective elemental changes within the layers. 

Q16. What is the effect of cation vacancies on the efficiency of a PSC?

The authors have also proven, that both halide and cation vacancies are mobile (albeit the latter are considerably slower than the former) and their distribution in the perovskite layer can considerably affect charge extraction and, in consequence, PCE of the device. 

Q17. What is the effect of the long-term ion migration on device performance and stability?

To study the impact of the long-term ion migration on device performance and stability, the authors prepared state-of-the-art PSCs, using the mixed halide-cation perovskite composition CH3NH3/CH(NH2)2 Pb Br/I and the antisolvent deposition method on mesoporous TiO2 substrates,56 which enabled the realization of power conversion efficiencies above 20% (see device characterization in SI). 

Q18. What are the intrinsic losses associated with the degradation of organic materials and metal contacts?

so called intrinsic losses, have been mostly associated with the degradation of organic materials and metal contacts within PSCs.3, 12, 13 Indeed, significant progress has been made by replacing the organic components with their inorganic counterparts and passivating the interfaces between the different layers composing the device. 

Q19. What is the reason why the authors could not conclude if cations are mobile?

again the authors could not conclude if cations are mobile, largely due to the fact that the technique is not suited for tracing lightweight elements constituting CH3NH3+ ions (the fact that the devices had to be coated with C prior to the analysis, made the analysis even more challenging).