SEI - ACS Publications - American Chemical Society

Jul 25, 2016 - Consequences of Solid Electrolyte Interphase (SEI) Formation upon ... and found that the SEI induces an additional electron-transfer pr...
1 downloads 0 Views 2MB Size
Subscriber access provided by CORNELL UNIVERSITY LIBRARY

Article

Consequences of Solid Electrolyte Interphase (SEI) Formation Upon Ageing on Charge Transfer Processes in Dye-Sensitized Solar Cells Miguel Flasque, Albert Nguyen Van Nhien, Davide Moia, Piers R. F. Barnes, and Frédéric Sauvage J. Phys. Chem. C, Just Accepted Manuscript • DOI: 10.1021/acs.jpcc.6b05977 • Publication Date (Web): 25 Jul 2016 Downloaded from http://pubs.acs.org on July 26, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

The Journal of Physical Chemistry C is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Consequences Of Solid Electrolyte Interphase (SEI) Formation Upon Ageing On Charge Transfer Processes In Dye-sensitized Solar Cells Miguel Flasque,1,2 Albert Nguyen Van Nhien,2,3 Davide Moia,4 Piers R.F. Barnes,4 Frédéric Sauvage1,2* 1

Laboratoire de Réactivité et Chimie des Solides, Université de Picardie Jules Verne, CNRS UMR 7314, 33 rue Saint Leu, 80039 Amiens, France 2

3

Institut de Chimie de Picardie (ICP), CNRS FR 3085, 33 rue Saint Leu, 80039 Amiens, France

Laboratoire de Glycochimie, des Agroressources et des Anti-microbiens, Université de Picardie Jules Verne, CNRS UMR7378, 33 rue Saint Leu, 80039 Amiens, France 4

Physics Department, Imperial College, 1 Exhibition Road, London SW7 2AZ, U.K. *E-mail : [email protected], Tel : +33 3 2282 7971

Abstract Solid electrolyte interphase (SEI) layers form on sensitized-TiO2 photo-anodes and platinum counter-electrodes when dye-sensitized solar cells (DSSCs) are subjected to an accelerated ageing protocol (e.g. heating at 85 °C in the dark for 500 hours). To understand how this will impact the device operation, electrochemical impedance spectroscopy study showed that the SEI induces an additional electron transfer process from the TiO2 to the electrolyte. This is materialized by the onset of a new charge transfer semi-circle at higher frequencies, predominantly visible under bias voltage similar and above open circuit voltage. Our results emphasized on the detrimental role of the SEI formation on device performance and lifetime. Additionally, ns-transient absorption spectroscopy shows that SEI formation reduces the rate oxidised dye regeneration. We also show that a proportion of the photogenerated holes on the dyes are transferred to the SEI itself. Prolonged ageing duration leads to the electrode’s mesoporosity network entirely clogged by the SEI; thus impeding efficient transport of the electrolyte redox couple, also responsible for a further decline in photovoltaic performances.

1 ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 21

1. Introduction

Dye-sensitized solar cells (DSSC), which to some extent mimic the principle of natural photosynthesis, have attracted much attention as a photovoltaic which could be manufactured at low cost and deployed for unconventional applications. This potentially disruptive technology is underpinned by a robust photoelectrochemical process in which electron injection from a dye’s photoexcited state to electron acceptor levels in TiO2 is estimated to be stable for 107 – 108 turnovers.1-2 After years of intensive multi-disciplinal research, power conversion efficiency (PCE) of 14.3 % under A.M. 1.5G conditions (100 mW/cm2) has been reached.3 This accomplishment is credited to the use of two complementary organic dyes anchored to TiO2 in combination with the strongly oxidising one electron [Co(phen)3]3+/2+ redox mediator. In contrast, the performance of stable devices, which are the real indicator for potential commercialization of the technology, still lag far behind in terms of their photovoltaic performance and the understanding of the degradation pathways involved as they age under light and/or thermal stress. Currently the best PCE obtained for small area devices (< 1 cm2), which have the potential for long term stability, are limited to around 9 to 10 % PCE by using the benchmark iodine-based redox couple with low-volatility nitrile solvents and heteroleptic Ru+II polypyridil complexes.4,5 We have shown that the stability of DSSC devices is principally governed by the properties of the solid/liquid interface(s), in common with many electrochemical displays.6-8 Our results highlighted that a major degradation pathway stems from the unstable character of the TiO2/MPN-based electrolyte interface where the TiO2 surface exposed between adsorbed dye molecules plays a pivotal role in the degradation of the electrolyte and concentration/depletion of its components.8 These processes result in the growth of a solid polymeric layer on the surface of the mesoporous photo-anode network which we refer to as a solid electrolyte interphase (SEI), analogous to the terminology introduced by Peled and widely used in the study of battery systems.9 This SEI, for which the growth is promoted for ageing temperatures greater than 60°C, contains a high concentration of iodine/iodide, electrolyte additives and probably the underlying degradation products which are embedded in a cohesive polymerized acrylonitrile network.8 These first observations led us to the following question: how does this SEI influence the interfacial charge transfer processes and TiO2 surface properties? In this work we present electrochemical impedance spectroscopy and nanosecond transient

2 ACS Paragon Plus Environment

Page 3 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

absorption spectrometry measurements of both fresh and aged (SEI thickness up to 2-3 nm) photo-anodes and Pt counter electrodes to help resolve this question.

2. Experimental section

2.1.Synthesis Titanium(IV) isopropoxyde, iodomethane, 1-methylimidazole, iodine were purchased from Aldrich and used as received. The guanidinium thiocyanate and 3-methoxypropionitrile (3MPN) solvent was purchased from Alfa Aesar. Acetic acid was obtained from VWR and Nbutyl-1H-benzimidazole from Merck. All products used in this study were opened and stored within an Ar-filled glove box. Platinum nanoparticles were purchased from Sigma-Aldrich. The TiO2 nanoparticles composing the photo-anode were synthesized according to a procedure adapted from literature.10 An equimolar ratio of acetic acid (5.94 g) was added to 0.1 mol of titanium (IV) isopropoxyde (29.00 g) under constant stirring. This reaction was left for 15 minutes under stirring. Then, 175 mL of deionized (D.I.) water was added to hydrolyse the modified titanium alkoxyde precursor leading to an instantaneous formation of a white precipitate corresponding to amorphous hydrated titanium oxide. After 1 hour vigorous stirring (ca. 800 rpm), peptization process was carried out through the addition of 1 mL of concentrated nitric acid (48 %) before heating stepwise the solution to 78°C in 45 minutes. The peptization process was kept progressing for 90 minutes until the solution turns from a white to a transparent light blue color. Water and isopropanol byproducts were removed by means of a rotavap at 50°C under 60 mbar vacuum until the amorphous colloidal TiO2.1.6H2O in solution was concentrated to give a final mass of 60.00 g. This concentrated solution was then transferred into a 100 mL Parr reactor. The crystallization of the anatase structure was reached after a thermal treatment at 250°C for 12 hours (ramp 4°C/min). The powder was retrieved by centrifugation, washed one time with D.I. water and three times with ethanol before being dried at 60°C overnight.

The synthesis of 1,3-dimethylimidazolium iodide (DMII) was performed according to the procedure in reference.11 The electrolyte was prepared using 3-methoxypropionitrile (3-MPN) as solvent, 0.15 mol/L of iodine (I2), 0.5 mol/L of N-butyl-1H-benzimidazole (NBB), 0.1 mol/L of guanidium thiocyanate (GuNCS) and 1 mol/L of DMII.

3 ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 21

2.2 Techniques

The BET surface area was evaluated by nitrogen adsorption at 77 K using a Micromeritics ASAP 2020. Prior to analysis, the powder was degassed at 150°C for 3 hours under vacuum (< 1µm Hg). High resolution transmission electron microscopy was carried out on 200 kV FEI Tecnai F20 S-TWIN microscope.

The photo-anode was prepared by screen-printing on FTO-type TCO glass (NSG10) through a 90T mesh size screen. The films were heated at 500 °C for 30 minutes using a heat gun before sensitization. 300 µM of cheno-3a,7a-dihydroxy-5b-cholic acid was dissolved with an equimolar of C106 complex in a mixture of tert-butanol and acetonitrile solvent (1:1 by volume). After being washed by acetonitrile and dried in air, the overnight sensitized electrodes were sealed using a 50 µm thick Bynel gasket, melted by heating with the Ptmodified TEC7 Pilkington TCO counter electrode (described below). The device was then double sealed with Dyesol two part hermetic sealing to avoid oxygen and water ingress. The counter electrode was prepared by spreading out a drop of 5 mM H2PtCl6 ethanol solution onto the counter electrode before treating it at 400 °C for 15 min under air. A hole was introduced in the counter electrode by sand-blasting, allowing the internal space between the two electrodes to be filled with electrolyte by vacuum back filling. Photovoltaic performances were measured with a 5×1 cm2 rectangular photo-anode geometry. The cells were masked with a black tape with an aperture of 5 cm2. A 450 W xenon light source equipped with A.M.1.5G filter (Newport SOL3A) was used to provide an incident irradiance of 100 mW/cm² at the surface of the solar cells. The (J-V) measurements were performed using a Keithley model 2400 digital source meter (Keithley, USA) by applying an external voltage to the cell and measuring the photo-generated current. Light soaking experiments were performed at open circuit using an Atlas XLS+II chamber equipped with xenon lamp source and UV cut-off glass filters (λ < 360 nm) (light power of 76 mW/cm2). Alternatively, devices were aged at 85°C in darkness with a relative ambient humidity of ca. 50-55 %. Electrochemical impedance spectroscopy was performed under dark conditions to monitor charge transfer processes on a ms – s time scale. Ohmic drop differences between cells and before / after ageing was corrected to the applied bias voltage. Transient absorption spectroscopy (TAS) was used to probe both dye regeneration kinetic and charge 4 ACS Paragon Plus Environment

Page 5 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

recombination with the dye cation S+ on a nanosecond timescale. A Nd:YAG pumped OPO (Opotek Opolette 355) was used as pump. 530 nm was set as pump wavelength in order to excite the dye in its absorption window (MLCT band). A laser pulse repetition rate of 20 Hz was used. The fluence used was 60 µJ.cm-2 per pulse unless stated otherwise. The devices were masked and placed perpendicularly to the pump beam, which was directed from the laser output through a beam-guide. The sample was set at an angle of about 30° with respect to the optical axis of the probe beam. A quartz halogen lamp (Bentham IL1) driven by constant current power supply (Bentham 605) was used as probe. Neutral density and long pass filters were used to control the intensity of the probe and limit its spectral range to wavelengths above the absorption window of the dye. After the sample, the probe beam was passed through a monochromator where we selected the probe wavelength which was focused on a silicon photodetector. The electrical signal from the detector was amplified via a Costronics preamplifier and amplifier and recorded with an oscilloscope. The background level was also measured and its value was used to calculate the variation in optical density upon laser excitation, ∆OD, using the formula ∆OD(t) = -log10(1 + V(t)/VBG) ≈ - V(t)/(VBG × 2.3), where V(t) is the transient signal and VBG the background level. To obtain V(t) for each measurement, a second transient measurement was performed while blocking the probe beam. This was subtracted from the signal obtained with the probe beam passing through the sample and reaching the photodetector in order to correct for eventual electronic interference and photoluminescence from the sample. Transient absorption signals at 780 nm were measured to monitor the oxidized dye population decay upon laser excitation.

3. Results and discussion

The growth of a SEI has been observed on the bare surface of TiO2 as well as on sensitized TiO2 with C106 as a dye.8 The development of this interfacial polymeric layer is triggered exclusively by the external action of temperature rather than by a photo-induced stress.8 After ca. 500 hours of ageing at 85°C in dark conditions, SEI forms a relatively conformal covering with a thickness in the range of 3 nm (Fig. 1a). This is similar or slightly greater than the typical thickness of the dye monolayer. The high content of heavy iodide species in this layer provides a darker contrast on transmission electron micrographs compared to the original TiO2 nanoparticles sensitized with C106. The high iodine content of the SEI was confirmed previously by XPS and ToF-SIMS measurements and demonstrated the idea that SEI growth 5 ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 21

is responsible for the well-established depletion of iodine/iodide in electrolytes. The SEI also contains sulfur, cyano and imidazolium units coming from the 1,3-di-methylimidazolium iodide or/and the N-butylimididazole used in the electrolyte composition.8 Post-mortem analysis on devices aged at 85°C in darkness with the transmission electron microscope show that this SEI layer can completely envelope clusters of TiO2 particles in the photo-anode. The SEI in these clusters forms a very cohesive polymer made of polyacrylonitrile whose thickness can reach 50 nm (Fig. 1b). Indeed, based on GC/MS/FT-IR analysis of the aged electrolyte, we found that the 3-MPN molecules react with TiO2 (sensitized or not) at 85°C to yield two distinct gasses: methanol and acrylonitrile. The latter is known to polymerize in presence of halides and/or mild temperatures; both these conditions are met in dye-sensitized solar cells. We consider that clogging of the mesopores by the SEI growth (especially at temperatures greater than 60°C) is a major mechanism for the degradation in device performance. Varying degrees of clogging dependent on TiO2 film morphology are likely to be the primary explanation for the lack of consensus in the literature on achieving stability at 85°C in the dark when the electrolyte is based on 3-MPN. Indeed, while the growth of SEI is comparatively slow at 60°C (not conformal coverage with < 1nm thickness after 500 hours ageing), its formation is drastically accelerated when devices are aged at temperatures of and above 80°C.

Figure 1. Transmission Electron Micrographs of (a) C106-sensitized TiO2 nanoparticles used as photo-anode after 500 hours ageing at 85°C in dark (b) Photo-anode particles on post mortem devices aged at 85°C. Selected Area Electron Diffraction pattern of (b) is reported (SAED). Fig. 2 shows that a similar SEI also forms on the platinum nanoparticles whose role is to electro-catalyze I-I bond cleavage at the counter electrode. However, by contrast to what has been observed on TiO2, its formation has little impact on cell electrolyte composition. This is not surprising if we consider that the surface area of the platinum nanoparticles on the counter 6 ACS Paragon Plus Environment

Page 7 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

electrode have a far lower area in contact with the electrolyte than the TiO2 (ca. 0.03 m2/mL versus 28 m2/mL for the photo-anode). This again stresses the role of the electrode’s surface on iodine depletion. Chemical analysis of the SEI on platinum revealed the same elements as those deposited on the TiO2.8 However, in contrast to the TiO2, the SEI grown on the platinum electrode contains a lower concentration of imidazolium and iodide and a higher concentration of sulfur. The layer is also thinner and more conformal (Fig. 2).

With such drastic modification of the interfaces during the accelerating ageing tests (a part of IEC 61646)12, electrochemical impedance spectroscopy (EIS) was used to examine the evolution of charge transfer processes in the ms-s time scale, and nanosecond pump-probe transient absorption spectroscopy enabled the dye regeneration and recombination of photogenerated holes on dyes to be monitored on µs-ms time scales.

Figure 2. Evolution of the charge transfer resistance as a function of ageing time at 60°C/100 mW.cm2 or 85°C/dark conditions determined by electrochemical impedance spectroscopy on FTO/Pt symmetric cells. Related transmission electron micrographs are reported in a) for initial time and b) after 500 hours ageing at 85°C / dark showing the formation of SEI. To investigating the electrochemical properties of the SEI formed on platinum, symmetric FTO-Pt devices were aged either at 60°C with 100 mW.cm-2 illumination, or at 85°C in the dark to study the evolution of the charge transfer resistance of the counter electrode without complications from the photo-anode (Fig. 2). The impedance spectra in Nyquist representation systematically showed two semi-arcs. The first, at higher frequencies (e.g. the arc labeled 16 kHz in the Fig. 2 inset), corresponds to the global charge transfer resistance between Pt and I3-/I- redox couple whereas the low frequency semi-circle, ca. 1 Hz, is ascribed to the semi-finite I3- diffusion process in electrolyte. The apparent tri-iodide diffusion coefficient ( D I 3− ) does not vary significantly during ageing, from 2.2×10-5 cm2/s to 1.2×10-5 cm2/s after 83 days ageing at 60°C/100 mW.cm-2 and down to 9.30×10-6 cm2/s after 14 days ageing at 85°C/dark (note: assuming no evolution of iodine concentration in electrolyte – in 7 ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 21

reality we estimate that ca. 10 % is consumed at 85°C and almost nothing at 60°C). This diffusion coefficient was extracted using the following equation: DI − = δ2 × K N (eq. 1) 3

where δ represents half the distance between the two electrodes13 and KN is deduced from the ion diffusion impedance (ZN)14: ZN =

 iω   (eq.2) tanh  KN  iω  

W

where W=

kT 2 2

n e

[I3− ]A

DI −

(eq. 3)

3

-1

ω represents the angular frequency (in rad.s ), W the Warburg parameter, n the number of electron transferred (n = 2), k the Boltzmann constant, T the temperature, e the elementary charge and A the electrode’s surface.

In comparison to this “bulk” electrolyte property, the charge transfer resistance value drops drastically, from 12.5 Ω.cm2 to less than 3 Ω.cm2 after only 7 days ageing at 60°C/100 mW.cm-2 and even to less than 2 Ω.cm2 when ageing at 85°C/dark. After this phase, no further evolution is experienced. The charge transfer resistance then continues to decrease marginally. Such evolution has been noted previously in the literature on complete devices but without extensive explanation.15-17 We attribute this interfacial electrochemical modification of Pt/electrolyte characteristic to the formation of SEI surrounding the platinum nanoparticles during ageing. We conclude that its formation on platinum has a beneficial impact on the device performance since it improves the electro-catalytic properties of the platinum nanoparticles itself.

The device performances under standard illumination condition (A.M. 1.5G) before and after 100 hours and 500 hours ageing at 85°C in dark is reported in Table 1. After ageing, all (J-V) characteristics show decrease in performance. Jsc (mA/cm2)

Voc (mV)

ff (%)

η (%)

Fresh

12.3

676

65.7

5.5

100 hours ageing

12.5

649

65.1

5.3

500 hours ageing

11.6

592

62.6

4.3

8 ACS Paragon Plus Environment

Page 9 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Table 1. PV characteristics of C106-based dye sensitized solar cells (5 cm2) before and after 100 hours and 500 hours ageing at 85°C in darkness.

In particular, after the 500 hours ageing, the photo-voltage drops by 13 % from 676 mV to 592 mV leading to a power conversion efficiency decrease from 5.5 to 4.3 % on 1 x 5 cm2 device. Fig. 3a shows impedance measurements of complete devices, and indicates the onset of a clearly visible additional semi-arc in the high frequency region (relaxation at 3.79 kHz when 840 mV is applied) after the device has been aged (500 hours at 85°C in dark). We ascribe this new semi-circle to an additional contribution from the SEI to electron transfer between the TiO2 and electrolyte. The attribution of this feature is supported by the occurrence of a very similar feature in the kHz region observed for the electrode materials of battery’s, for instance on graphite, where the SEI formation stems from the carbonate-based solvent reduction of the electrolyte typically used in lithium-ion batteries.9,18-24 Our observations suggest that the SEI formed in DSSCs contributes to electron capture by triiodide. Interestingly, the greater the magnitude of the applied forward bias voltage beyond the open circuit photo-voltage, the clearer the separation between the relaxation time of SEI/electrolyte and TiO2/electrolyte charge transfer becomes (Fig. S1). Conversely, near short circuit, the two contributions have more similar relaxation times yielding a ‘flattened’ semiarc. When we attempted to fit the impedance data of aged devices using only a single RREC//CPE(Cµ) distributed parallel element (upper circuit in Fig. 3b), corresponding to only one recombination pathway, an unphysically low value for the exponent of the constant-phase element (CPE) was required. This also resulted in unreasonable parameters for electron diffusion length and charge collection efficiency. Thus, we have added this new recombination pathway to the standard equivalent circuit model of the cell as an additional distributed parallel element, RSEI//CPE(CSEI), where RSEI is the resistance to charge transfer through the SEI and CPE(CSEI) is a constant phase element corresponding to the chemical capacitance of the layer (Fig. 3b). The values for RREC and RSEI obtained by fitting this model to the spectra are reported in Fig. 3c. This allows comparison of the relative contributions to recombination in the device from the SEI/electrolyte interface and the well-established sensitized-TiO2/electrolyte interface.

9 ACS Paragon Plus Environment

The Journal of Physical Chemistry

Before Avantageing vieillissement

a)

-20

Z'' (ohm)

840 mV

After ageing Après vieillissement

Before ageing After ageing

-15

-10

Rec. Recomb 114 114 Hz Hz

Rec. SEI SEI 3.79 3,79 kHz kHz

-5

Recomb 9.99 Hz 9,99 Hz

0 0

5

10

15

20

Z' (ohm) Rs

rtr

rtr

rtr

rtr

Electrolyte

.. c μ rrec

rrec

rrec



TCO / Pt

cμ Rct

..

TCO

Zd(sol)

SEI

SEI Rs

cPt

TiO2

b) rtr

rtr

rtr

rtr

Electrolyte

.. Rrec

cμ Rrec,SEI

cμ,SEIRrec

cμ Rrec,SEI

cμ,SEI Rrec

cμ Rrec,SEI

TCO / Pt

cμ,SEI

Rct ..

TCO

Zd(sol)

cPt

TiO2 SEI

100

c) 2

Recombination resistance (Ω Ω .cm )

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 21

10

SEI 1 TiO

2

Recombination dominated by SEI

0.1 450

500

550

600

Recombination dominated by TiO2

650

700

750

800

Bias voltage (mV)

10 ACS Paragon Plus Environment

Page 11 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Figure 3. (a) Comparison of Nyquist plot before and after 500 hours ageing at 85°C/dark of C106-sensitized devices at a bias voltage of 840 mV. (b) Representation of the equivalent electrical model used to account for the device functioning when TiO2 is conductive for fresh (top circuit) and aged devices (bottom circuit) including the contribution of the SEI. (c) Evolution of the global recombination resistance including the relative contribution of TiO2 and SEI as a function of the bias voltage. Near and above the open circuit photovoltage value, recombination with tri-iodide is dominated by charge transfer from the TiO2 surface since this has the lowest recombination resistance of the two processes. Closer to short-circuit recombination via the SEI layer appears to be the dominant process with the lower resistance, signifying that the SEI acts as an electron “leaking” pathway when the device is under operation. Between 540 mV to 800 mV the interfacial charge transfer resistance between SEI/electrolyte appears relatively independent of the external bias voltage. This is something that can also be saw in the literature without any further description.25 On the other hand, the TiO2/electrolyte contribution shows an exponential dependence on voltage (as does the SEI/electrolyte component below 540 mV). Although speculative at this stage, we hypothesize that the threshold voltage close to Voc where SEI’s resistance turns from independent to dependent of the external bias voltage could stem from the ionic species trapped in the SEI which are moving from an immobile to a mobile state when going towards short-circuit, i.e. as a result of a drift-diffusion process to compensate the build-up of free charges in TiO2’s surface below Voc.26 Fig. 4 shows the chemical capacitance of the TiO2, Cµ, as a function of measurement voltage, derived from fits of the equivalent circuits in Fig. 3b. This parameter is typically ascribed to a distribution of localized trap states below the TiO2 conduction band. The results suggest the TiO2 conduction band has shifted up in energy relative to the electrolyte redox couple and/or that the density of trap states has reduced, in agreement with previous observations of aged devices.17 Because the SEI contains a relatively high concentration of charged and polar species (eg. I-, I3-, NCS-, Gu+, polyacrylonitrile and probably other degradation byproducts from electrolyte) neither of these processes would be surprising.27 However, given the reduction in the Voc of the aged device, an increase in the conduction band seems less likely than the possibility that some of the TiO2 surface traps have been passivated by the SEI.

11 ACS Paragon Plus Environment

The Journal of Physical Chemistry

900 After ageing

800

Bias voltage (mV)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 21

700

Before ageing

600 500 400 300 0.002 0.004 0.006 0.008 0.01 2

Cell capacitance (F/cm )

Figure 4. Comparison of bias voltage as a function of cell capacitance for C106-based devices before and after 500 hours ageing at 85°C/dark. Regardless, the electron lifetime after ageing decreases by up one order of magnitude when plotted either against voltage or capacitance (Fig. 5) demonstrating the harmful influence of the SEI on recombination. The exponential dependency of the electron lifetime as a function of bias voltage typically observed is modified after aging owing to the intermediate role played by the SEI on electron – tri-iodide (or iodine) charge transfer reaction. Indeed, its evolution can be divided into two clearly-visible portions, one roughly above Voc for which the value is independent of the external bias voltage, and one below tending to an exponential evolution. This crossover depends on whether the charge transfer kinetic is controlled by TiO2 surface or mediated through the SEI.

12 ACS Paragon Plus Environment

Page 13 of 21

10 Before ageing

10

a)

1

1 Electron lifetime (s)

Electron lifetime (s)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

After ageing

0.1 V V

0.01

OC

OC

450

500

550

600

b)

After ageing

0.1

0.01

0.001 400

Before ageing

650

700

750

800

0.001 0.001

Bias voltage (mV)

0.01 2

Cell capacitance (mF/cm )

Figure 5. Comparison of the evolution of electron lifetime before and after 500 hours ageing at 85°C/dark as a function of (a) bias voltage (b) cell capacitance. Finally, we monitored the influence of the SEI on the process of dye regeneration. For this, a series of transient absorption measurements were carried out on 2 µm thick fresh or aged electrodes (85°C in darkness for 100 hours or 500 hours). To avoid additional contributions resulting from the iodine depletion in the electrolyte (and variation of other electrolyte components), the aged photo-anodes were re-assembled into a new device including a fresh electrolyte either with or without an iodine/iodide redox couple. Figure 6a and b show the measured transient absorption signal probed at 780 nm upon pulsed laser excitation of the samples at 530 nm. At 780 nm probe wavelength, we expect a significant contribution to the signal from the oxidized dye absorption. The absorption window of electrons in the TiO2 also extends to this wavelength as we discuss below. The blue lines in figure 6a and b show that, in absence of redox couple in the electrolyte, a slow dynamics of the transient absorption signal is observed for both the (a) fresh and (b) the sample aged for 500 hours at 85°C in the dark. The transient absorption signal does not completely decay for both the blue traces, so that only a lower limit in the order of milliseconds to the half-life can be estimated. For the samples with redox couple (red traces) we recognize a biphasic behavior. This has been previously discussed as the signature of oxidized dye regeneration to the redox couple (fast phase) followed by recombination of electrons in the TiO2 with the electrolyte (second phase).28 Fits to the transient absorption decays were obtained using stretched exponential functions to account for the inhomogeneity of the regeneration rate within the device and the ionic diffusion control of iodide when the electrolyte is composed of a redox mediator.29,30 By fitting the red traces with a sum of two stretched exponential to account for both phases, we extract estimates of the half-life for the oxidized dye. Values of τ1/2 < 1.5 µs result from the fit 13 ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 21

for both the fresh, and the 100 hours aged samples. This is close to our transient absorption measurement resolution, so an accurate estimate of the lifetime is not possible. On the other hand, the samples that were aged for 500 hours at 85°C in the dark showed significantly longer values of τ1/2, between 10 and 20 µs. This observation suggests that the formation of the SEI results in a slowdown in the dye regeneration process. This is expected to degrade the power conversion efficiency in aged solar cell devices due to the competition between the dye regeneration process with back electron transfer from TiO2 to the oxidized dye, thus reducing the dye regeneration efficiency. We also investigate the dynamics of the nominally redox-free samples at different probe wavelength (data are shown in the supporting information). In figure S2 we show that the transient absorption signals of a fresh sample without redox couple probed at 780 nm (where oxidized dye absorption dominates) and at 980 nm (where the absorption of electrons in the TiO2 dominates) follow identical dynamics. This suggests that electron hole recombination is the dominant process occurring in the film upon photoexcitation. On the other hand, the same set of data collected for a sample with SEI (aged for 500 hours at 85°C in the dark) show that the TAS signal attributed to oxidized dyes decays faster than the one attributed to electrons in the TiO2. This difference is likely to be related to a regenerative effect of the SEI layer. We expect the presence of iodide-based species in the SEI surrounding the dyes to allow hole transfer from the dye to the SEI.

Figure 6. Transient absorption spectroscopy measurements performed for devices with and without redox couple for (a) a fresh device and (b) a device measured after 500 hours ageing at 85°C/dark. Pump and probe wavelength were respectively 530 and 780 nm. The measurements were carried out using a repetition rate of 20 Hz and a fluence of 60 µJ.cm-2 per pulse. The solid lines correspond to fits to the raw data (also shown in the background) using one (blue lines) or the sum of two (red lines) stretched exponentials. Note that in contrast to what has been reported on N719 and Z907Na using more drastic ageing conditions than ours, we have no evidence for any ligand exchange reaction of the 14 ACS Paragon Plus Environment

Page 15 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

monodentate thiocyanate with electrolyte’s components, or decarboxylation reaction which could have modified the dye properties on the photo-anode upon ageing.31-34 Nevertheless, in addition to the intrinsic loss of performances triggered by the SEI, a slower rate for dye regeneration can prompt important issues for degradation since dyes are typically less stable in their oxidized form.34 Finally, in addition to the electrochemical impedance study, the transient absorption study strengthen the conclusion that the SEI is not simply an additional interfacial media in the device. For redox-free electrolyte, we found that the SEI itself plays a role of regenerative media for the dye cation to some degree (Fig. S2). This is not totally surprising since this latter is embedding, among other components of the electrolyte, iodine/iodide species as previously demonstrated.8 We speculated above that these ions may display some mobility within the polymeric network of the SEI. This new regenerative path appears much slower than the regeneration through an electrolyte and is expected to operate solely until all iodide captured in the SEI is consumed (non-regenerative process).

4. Conclusion

This work sheds light on the role of the solid electrolyte interphase layer (SEI) which forms on sensitized-TiO2 photo-anodes and platinum counter electrodes when devices are subjected to accelerated ageing protocols involving high external temperature stress (≥ 60°C). We found that SEI noticeably accelerates the electron transfer rate at counter electrode, but also with triiodide species in the electrolyte. Additionally, it also allows hole transfer from dye cation to SEI because of its high content of iodide-based species, while it impedes the hole transfer from the dye cation to iodine. Thus, we have presented for first time evidence for regenerative behavior of the SEI itself in aged and redox free solar cells. Our results stress the detrimental role of the SEI to maintain the high performance in these devices. These observations explain the general trends in Jsc, Voc and ff observed in the literature when devices are exposed to ageing at 60°C under light or 85°C in dark. In addition, prolonged ageing will cause the mesopores to become clogged leading to potentially sharp loss in performances and the superficial perception that electrolyte has suddenly leaked. We believe that the SEI formation is a/or the major issue which limits the device life-in service. The SEI could also play a role in some well-established but unexplained phenomena in DSSCs, such as the light soaking recovery of the performances when devices are subjected to prolonged ageing in dark (at 80°C or 85°C).35

15 ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 21

Acknowledgements MF is indebted to the Conseil Regional de Picardie and FEDER for the Ph-D grant. MF, ANVN and FS acknowledge Dr. Carine Davoisne (Ass. Prof. at LRCS) for electron microscopy investigations and the Region de Picardie and FEDER for funding ROBUST and CONCERTO research programs in which this work has been elaborated. PRFB and DM are grateful to the UK Engineering and Physical Sciences Research Council for financial support (grants EP/J002305/1, EP/M023532/1 and EP/I019278/1).

Supporting Information Example of Nyquist plot recorded at different bias voltage and transient absorption decays with an electrolyte containing no redox couple, before and after ageing are reported as Supporting Information.

16 ACS Paragon Plus Environment

Page 17 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

References (1) Graetzel M. Solar Energy Conversion by Dye-sensitized Photovoltaic Cells, Inorg. Chem. 2005, 44(20), 6841-6851. (2) Hinsch A.; Kroon J.M.; Kern R.; Uhlendorf I.; Holzbock J.; Meyer A.; Ferber J. Longterm Stability of Dye-sensitized Solar Cells, Prog. Photovolt. Res. Appl. 2001, 9(6), 425-438.

(3) Kakiage K.; Aoyama Y.; Yano T.; Oya K.; Fujisawa J.; Hanaya M. Highly-efficient Syesensitized Solar Cells with Collaborative Sensitization by Silyl-anchor and Carboxy-anchor Dyes, Chem. Commun. 2015, 51(88), 15894-15897.

(4) Sauvage F.; Chhor S.; Marchioro A.; Moser J.-E.; Graetzel M. Butyronitrile-based Electrolyte for Dye-sensitized Solar Cells, J. Am. Chem. Soc. 2011, 133(33), 13103-13109.

(5) Yu Q.; Zhou D.; Shi Y.; Si X.; Wang Y.; Wang P. Stable and Efficient Dye-sensitized Solar Cells: Photophysical and Electrical Characterizations,

Energy Environ. Sci. 2010,

3(11), 1722-1725.

(6) Tarascon J.-M.; Armand M. Issues and Challenges Facing Rechargeable Lithium Batteries, Nature 2001, 414, 359-367.

(7) Irvine J.T.S.; Neagu D.; Verbraeken M. C.; Chatzichristodoulou C.; Graves C.; Mogensen M. B. Evolution of the Electrochemical Interface in High-temperature Fuel Cells and Electrolysers, Nat. Energy 2016, 1, 1-13.

(8) Flasque M.; Nguyen Van Nhien A.; Swiatowska J.; Seyeux A.; Davoisne C.; Sauvage F. Interface Stability of TiO2 / 3-methoxypropionitrile-based Electrolyte: First Evidence for SEI Formation and Implications, ChemPhysChem. 2014, 15(6), 1126-1137.

(9) Peled E. The Electrochemical Behavior of Alkali and Alkaline Earth Metals in Nonaqueous Battery Systems—The Solid Electrolyte Interphase Model, J. Electrochem. Soc.

1979, 126(12), 2047-2051. (10) Ito S.; Murakami T.N.; Comte P., Liska P.; Grätzel C.; Nazeeruddin M.K.; Grätzel M. Fabrication of Thin Film Dye Sensitized Solar Cells with Solar to Electric Power Conversion Efficiency over 10 %, Thin Solid Films 2008, 516(14), 4613-4619.

(11) Oertel A. M.; Ritleng V.; Burr L.; Chetcuti M. J. Synthesis and Structural Characterization of Half-Sandwich Nickel Complexes Bearing Two Different N-Heterocyclic Carbene Ligands, Organometallics 2011, 30(24), 6685-6691.

(12) https://webstore.iec.ch/publication/5697, access on May 27th, 2016

17 ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 21

(13) Haush A.; Georg A. Diffusion in the Electrolyte and Charge-transfer Reaction at the Platinum Electrode in Dye-sensitized Solar Cells, Electrochimica Acta 2001, 46(22), 34573466.

(14) MacDonald J.-R. Impedance Spectroscopy, Wiley, New-York, USA, 1987. (15) Jiang N.; Sumitomo T.; Lee T.; Pellaroque A.; Bellon O.; Milliken D.; Desilvestro H. High Temperature Stability of Dye Solar Cells, Sol. Energy Materials Sol. Cells 2013, 119, 36-50.

(16) Harikisun R.; Desilvestro H. Long-term Stability of Dye Solar Cells, Solar Energy 2011, 85(6), 1179–1188.

(17) Kontos A. G.; Stergiopoulos T.; Likodimos V.; Milliken D.; Desilvesto H.; Tulloch G.; Falaras P. Long-term Thermal Stability of Liquid Dye Solar Cells, J. Phys. Chem. C 2013, 117(17), 8636−8646.

(18) Besenhard J.O.; Winter M.; Yang J.; Biberacher W. Filming Mechanism of Lithiumcarbon Anodes in Organic and Inorganic Electrolytes, J. Power Sources 1995, 54(2), 228-231.

(19) Ein-Eli Y.; Markovsky B.; Aurbach D.; Carmeli Y.; Yamin H.; Luski S. The Dependence of the Performance of Li-C Intercalation Anodes for Li-ion Secondary Batteries on the Electrolyte Solution Composition, Electrochimica Acta 1994, 39(17), 2559-2569.

(20) Ein-Eli Y.; Koch V.R. Chemical Oxidation: A Route to Enhanced Capacity in Li‐Ion Graphite Anodes, J. Electrochem. Soc. 1997, 144(9), 2968-2973.

(21) Ohzuku T.; Iwakoshi Y.; Sawai K. Formation of Lithium‐Graphite Intercalation Compounds in Nonaqueous Electrolytes and Their Application as a Negative Electrode for a Lithium Ion (Shuttlecock) Cell, J. Electrochem. Soc. 1993, 140(9), 2490-2498.

(22) Peled E.; Golodnitsky D.; Ulus A.; Yufit V. Effect of Carbon Substrate on SEI Composition and Morphology, Electrochimica Acta 2004, 50(2-3), 391-395.

(23) Scott M.G.; Whitehead A.H.; Owen J.R. Chemical Formation of a Solid Electrolyte Interface on the Carbon Electrode of a Li-ion Cell, J. Electrochem. Soc. 1998, 145(5), 15061510.

(24) Liu Z.; Yu A.; Lee J.Y. Modifications of Synthetic Graphite for Secondary Lithium-ion Battery Applications, J. Power Sources 1999, 81–82, 187-191.

(25) Stergiopoulos T.; Kontos A.G.; Jiang N.; Milliken D.; Desilvestro H.; Likodimos V.; Falaras P. High Boiling Point Solvent-based Dye Solar Cells Pass a Harsh Thermal Ageing Test, Sol. Energy Materials Sol. Cells 2016,144, 457–466.

18 ACS Paragon Plus Environment

Page 19 of 21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

(26) Gentilini D.; Gagliardi A.; Franco A.A.; Sauvage F.; Di Carlo A. A Drift-Diffusion Study on Charge Unbalancing Effects in Dye-Sensitized Solar Cells, J. Electrochem. Soc.

2015, 162(10), H753-H758. (27) De Angelis F.; Fantacci S.; Selloni A.; Grätzel M.; Nazeeruddin Md. K. Influence of the Sensitizer Adsorption Mode on the Open-circuit Potential of Dye-sensitized Solar Cells, Nano Lett. 2007, 7(10), 3189-3195.

(28) Anderson A.Y.; Barnes P.R.; Durrant J.R.; O’Regan B.C. Simultaneous Transient Absorption and Transient Electrical Measurements on Operating Dye-sensitized Solar Cells: Elucidating the Intermediates in Iodide Oxidation, J. Phys. Chem. C 2010, 114(4), 1953-1958.

(29) Anderson A.Y.; Barnes P.R.; Durrant J.R.; O’Regan B.C. Quantifying Regeneration in Dye-sensitized Solar Cells, J. Phys. Chem. C 2011, 115(5), 2439−2447.

(30) Ansari-Rad M. Dye Regeneration Kinetics in Dye-Sensitized Solar Cell: Long-Range Charge-Transfer Effects, J. Phys. Chem. C 2016, 120(16), 9000-9006.

(31) Hansen G.; Gervang B.; Lund T. Products of the Electrochemical Oxidation of cisL2Ru(II)(NCS)2 in Dimethylformamide and Acetonitrile Determined by LC-UV/Vis-MS, Inorg. Chem. 2003, 42(18), 5545-5550.

(32) Nour-Mohammadi F.; Nguyen H.T.; Boschloo G.; Lund T. An Investigation of the Photosubstitution Reaction Between N719-dyed Nanocrystalline TiO2 particles and 4-tertbutylpyridine, J. Photochem. Photobiol. A:Chem. 2007, 187(2-3), 348-355.

(33) Nguyen P.T.; Deng R.; Nguyen H.T.; Lund T. Thiocyanate Ligand Substitution Kinetics of the Solar Cell Dye Z-907 by 3-methoxypropionitrile and 4-tert-butylpyridine at Elevated Temperatures, Sol. Energy Mater. Sol. Cells 2009, 93(11), 1939-1945.

(34) Grünwald R.; Tributsch H. Mechanisms of Instability in Ru-based Dye Sensitization Solar Cells, J. Phys. Chem. B 1997, 101(14), 2564-2575.

(35) Sommeling P.M; Späth M.; Smit H.J.P.; Bakker N.J.; Kroon J.M. Long-term Stability Testing of Dye-sensitized Solar Cells, J. Photochem. Photobiol. A: Chem. 2004, 164(1-3), 137-144.

19 ACS Paragon Plus Environment

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 21

Table of Contents

20 ACS Paragon Plus Environment

Page 21 of 21

The Journal of Physical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

21 ACS Paragon Plus Environment