Volume 16, Issue 12 e202202320
Research Article
Open Access

Imparting Stability to Organic Photovoltaic Components through Molecular Engineering: Mitigating Reactions with Singlet Oxygen

Dr. Petr Henke

Dr. Petr Henke

Department of Chemistry, University of Aarhus, Langelandsgade 140, DK-8000 Aarhus C, Denmark

Search for more papers by this author
Cecilie Rindom

Cecilie Rindom

Department of Chemistry, University of Copenhagen, Universitetsparken 5, DK-2100 Copenhagen Ø, Denmark

Search for more papers by this author
Dr. Um Kanta Aryal

Dr. Um Kanta Aryal

Centre for Advanced Photovoltaics and Thin Film Energy Devices (SDU CAPE), Mads Clausen Institute, University of Southern Denmark, Alsion 2, DK-6400 Sønderborg, Denmark

Search for more papers by this author
Dr. Malte Frydenlund Jespersen

Dr. Malte Frydenlund Jespersen

Department of Chemistry, University of Copenhagen, Universitetsparken 5, DK-2100 Copenhagen Ø, Denmark

Search for more papers by this author
Dr. Line Broløs

Dr. Line Broløs

Department of Chemistry, University of Copenhagen, Universitetsparken 5, DK-2100 Copenhagen Ø, Denmark

Search for more papers by this author
Dr. Mads Mansø

Dr. Mads Mansø

Department of Chemistry, University of Copenhagen, Universitetsparken 5, DK-2100 Copenhagen Ø, Denmark

Search for more papers by this author
Prof. Dr. Vida Turkovic

Corresponding Author

Prof. Dr. Vida Turkovic

Centre for Advanced Photovoltaics and Thin Film Energy Devices (SDU CAPE), Mads Clausen Institute, University of Southern Denmark, Alsion 2, DK-6400 Sønderborg, Denmark

Search for more papers by this author
Prof. Dr. Morten Madsen

Corresponding Author

Prof. Dr. Morten Madsen

Centre for Advanced Photovoltaics and Thin Film Energy Devices (SDU CAPE), Mads Clausen Institute, University of Southern Denmark, Alsion 2, DK-6400 Sønderborg, Denmark

Search for more papers by this author
Prof. Dr. Kurt V. Mikkelsen

Corresponding Author

Prof. Dr. Kurt V. Mikkelsen

Department of Chemistry, University of Copenhagen, Universitetsparken 5, DK-2100 Copenhagen Ø, Denmark

Search for more papers by this author
Prof. Dr. Peter R. Ogilby

Corresponding Author

Prof. Dr. Peter R. Ogilby

Department of Chemistry, University of Aarhus, Langelandsgade 140, DK-8000 Aarhus C, Denmark

Search for more papers by this author
Prof. Dr. Mogens Brøndsted Nielsen

Corresponding Author

Prof. Dr. Mogens Brøndsted Nielsen

Department of Chemistry, University of Copenhagen, Universitetsparken 5, DK-2100 Copenhagen Ø, Denmark

Search for more papers by this author
First published: 10 March 2023
Citations: 2

Graphical Abstract

Stable organic photovoltaics: Incorporation of cyano substituents at the core of indenofluorene-extended tetrathiafulvalenes using Pd-catalyzed cyanation reactions reduces the reactivity of the exocyclic fulvene carbon-carbon double bonds towards singlet oxygen. The molecules were employed for the construction of more stable organic photovoltaic devices in combination with non-fullerene acceptors.

Abstract

One key challenge in the development of viable organic photovoltaic devices is to design component molecules that do not degrade during combined exposure to oxygen and light. Such molecules should thus remain comparatively unreactive towards singlet molecular oxygen and not act as photosensitizers for the generation of this undesirable species. Here, novel redox-active chromophores that combine these two properties are presented. By functionalizing indenofluorene-extended tetrathiafulvalenes (IF-TTFs) with cyano groups at the indenofluorene core using Pd-catalyzed cyanation reactions, we find that the reactivity of the exocyclic fulvene carbon-carbon double bonds towards singlet oxygen is considerably reduced. The new cyano-functionalized IF-TTFs were tested in non-fullerene acceptor based organic photovoltaic proof-of-principle devices, revealing enhanced device stability.

Introduction

Singlet molecular oxygen, O2(a1Δg), is known to play a key role in processes that contribute to the degradation of photoresponsive organic materials used in electroluminescent devices and solar cells.1-4 Many organic chromophores can photosensitize the production of O2(a1Δg),5, 6 and this is a relevant source of O2(a1Δg) upon irradiation of photoresponsive materials exposed to the ambient atmosphere containing oxygen. Degradation can be prevented or attenuated by addition of different types of antioxidants in the active layers of organic solar cells7-11 and by introduction of sputtered metal-oxide interlayers.12 Attention is also put on development of efficient encapsulation.13-16 However, because encapsulation layers are not perfect oxygen barriers, resistance to oxygenating/oxidative reactions in the presence of light and oxygen is, in fact, a key limiting feature that defines the lifetime of many devices that use “high performing” organic semiconductors. In lieu of encapsulating the photoresponsive device to exclude oxygen, the most desirable solution to this problem is to develop organic materials (i. e., donor, acceptor, or electron/hole transport interfacial layer components) that are not adversely influenced by oxygen, particularly O2(a1Δg).

Organic molecules can be oxygenated by O2(a1Δg) through π2+π2 and π2+π4 cycloaddition reactions, as well as through the ene reaction, to yield peroxides.17-20 These peroxides can undergo subsequent photochemical and/or thermal decomposition to yield radicals (e. g., alkoxyl radicals) that contribute to the decomposition of the material.1

Rate constants for reaction with O2(a1Δg) are largest for electron-rich substrates, and adding electron-withdrawing functional groups to olefins, for example, makes them less susceptible to reaction with O2(a1Δg).19 In this regard, the incorporation of nitriles into polyunsaturated molecules used in electroluminescent devices has been shown to increase the stability of the system towards O2(a1Δg)-mediated photooxidative degradation.21

π-Extended derivatives of tetrathiafulvalene (TTF; see Figure 1) have recently shown promise as light-harvesting donor molecules for organic photovoltaics.22-25 TTF can be either expanded at the periphery or by introducing a π-conjugated spacer between the two dithiafulvene rings. The advantage of such expansions is to redshift the absorption so that a better match with the solar spectrum is achieved. One possible modification is to introduce an indenofluorene (IF) scaffold as a π-conjugated spacer, generating indenofluorene-extended TTFs (IF-TTFs).25-29 Such compounds with alkylthio substituents at the dithiafulvene rings (like the IF-TTF shown in Figure 1) exhibit a longest-wavelength absorption maximum around 472–475 nm, and they undergo two reversible oxidations (like the parent TTF30).

Details are in the caption following the image

Structures of indeno[1,2-b]fluorene (IF), tetrathiafulvalene (TTF), indenofluorene-extended TTF (IF-TTF) with hexylthio substituents at the dithiole rings, and mono- and dicyano-substituted IF-TTFs, cIF-TTF and dcIF-TTF.

By combining the IF-TTF motif with a benzodithiazole electron acceptor at the IF-TTF core via an ethynediyl bridge and linking this acceptor to a benzoic acid unit, a donor-acceptor dyad for dye-sensitized solar cells has been reported; one such cell was found to exhibit a power conversion efficiency of 6.4 %.25 Although there is still room for optimization of the efficiency by HOMO–LUMO tuning of the structural motifs, we now focus instead on how to render the IF-TTF system stable for organic photovoltaic devices.

IF-TTF derivatives may potentially act as photosensitizers for the generation of O2(a1Δg) that, in turn, could react with the IF-TTF. Indeed, tetrathienyl-substituted TTFs in oxygen-saturated solution were shown to produce O2(a1Δg) that would subsequently undergo cycloaddition with TTF at its central fulvalene bond to give dioxetanes that rearranged to tetrathienyl-substituted 1,2,5,8-tetrathiecine-6,7-diones (10-membered heterocycles).31 Thus, to progress with these promising materials as components in a photovoltaic device, it is crucial to develop compounds that are photostable upon exposure to the ambient atmosphere.

In this regard, we pose three questions: 1) To what extent does the IF-TTF act as a photosensitizer for O2(a1Δg)? 2) How reactive is IF-TTF towards O2(a1Δg) and what products are formed? 3) How can we reduce a potential instability of the IF-TTF system to O2(a1Δg)-mediated photooxidative degradation by structural modifications?

To address these questions, and based on some of our previous work,21 we hypothesize that by introducing cyano substituents on the IF-TTF, an enhanced stability will be observed. Therefore, we decided to target two new derivatives, cyano-IF-TTF (cIF-TTF) and dicyano-IF-TTF (dcIF-TTF) shown in Figure 1, containing one and two cyano groups at the IF-TTF core, respectively. For this study, we set out to examine the relevant properties of these two compounds in comparison to the parent IF-TTF, both in solution and when incorporated in devices.

Results and Discussion

Computational modelling–attack of singlet oxygen

Based on (1) a general understanding of O2(a1Δg) behavior,19 and (2) the results of experiments on substituted phenylene vinylenes,21 we hypothesized that adding cyano groups to IF-TTF should decrease the rate of the reaction between O2(a1Δg) and the IF-TTF. However, before embarking on potentially tedious synthetic and kinetic procedures to assess this hypothesis, we felt it judicious to first use available computational procedures32-34 to model the reactivity of the IF-TTF with O2(a1Δg) and assess the effect of the added cyano groups.

Because the length of the peripheral alkyl substituents has no significant influence on the electronic properties of IF-TTFs,26-29 the hexyl groups were replaced with methyl groups to facilitate computational ease. Activation energies for the reaction of the IF-TTF with O2(a1Δg) (Figure 2) were found to increase by 0.7-0.9 kcal mol−1 and 1.0–2.2 kcal mol−1 upon introducing one and two cyano groups, respectively, at the IF-TTF core (more details are provided below and in the Supporting Information, SI). The scene is thereby set for studying these derivatives experimentally.

Details are in the caption following the image

Transition state structure (TS1) for the first step in the reaction between IF-TTFs and singlet oxygen; formation of this structure becomes more energetic upon introducing cyano groups according to calculations, AP-M06-2X/6-31+G(d,p).

Synthesis

For introducing one cyano group, we chose the known25, 29 iodo-functionalized IF-TTF 1 as precursor (Scheme 1). Compound 1 was successfully subjected to a Pd-catalyzed cyanation reaction upon treatment with Zn(CN)2, generating cIF-TTF in a yield of 55 %.

Details are in the caption following the image

Synthesis of cIF-TTF and dcIF-TTF by Pd-catalyzed cyanation reactions. See SI for details.

To add two cyano groups using the same procedure, we first had to introduce two iodine functionalities at the IF-TTF core. We prepared the diiodo-IF-TTF, compound 2, by subjecting 2,8-diiodoindeno[1,2-b]fluorene-6,12-dione (synthesis shown in SI) to olefination upon treatment with a 1,3-dithiol-2-thione in a reaction mediated by Lawesson's reagent (recently shown to be a convenient method for preparing IF-TTFs from such precursors35). Details are provided in the SI. To accomplish two-fold cyanation of 2, we had to change the cyanation conditions from those used for cyanating 1. By replacing the triphenylphosphine ligands with bulky and electron rich tris(tert-butyl)phosphine ligands (inspired by conditions reported by Maddaford and co-workers36) and adding zinc powder to avoid cyanide-induced catalyst deactivation,37, 38 we were able to obtain dcIF-TTF in a yield of 60 % using Zn(CN)2 as the cyanation agent.

Electrochemistry

Cyclic voltammetry (CV) and differential pulse voltammetry (DPV) measurements of cIF-TTF and dcIF-TTF were performed in CH2Cl2 with Bu4NPF6 as supporting electrolyte, and voltammograms are shown in Figure 3. Redox potentials obtained from the DPV measurements are listed in Table 1. Both compounds underwent two reversible one-electron oxidations with a rather broad first wave for the first oxidation on account of cation associations (radical cation/neutral and radical cation/radical cation dimers) as previously established26, 27 for IF-TTFs. They also underwent an irreversible oxidation at higher potential. The electron-withdrawing influence of the cyano substituents is clearly evident by the progressively more positive oxidation potentials upon proceeding from IF-TTF to cIF-TTF and then to dcIF-TTF. Moreover, the irreversible reduction of the compounds occurs at less negative potential with increasing number of cyano groups (by 0.2–0.3 V per cyano group): −2.31 V (IF-TTF), −2.04 V (cIF-TTF), −1.84 V (dcIF-TTF). Subtracting the first reduction potential from the first oxidation potential provides HOMO–LUMO gaps of 2.6 eV, 2.4 eV, and 2.3 eV for IF-TTF, cIF-TTF, and dcIF-TTF, respectively. These HOMO–LUMO gaps complement the bathochromic shift observed for the longest-wavelength absorption maxima along the same sequence of compounds (Table 2).

Details are in the caption following the image

Cyclic and differential pulse voltammograms for the newly synthesized cIF-TTF (top; 0.5 m and dcIF-TTF (bottom; 0.5 mm); potentials vs. Fc/Fc+. Solvent: CH2Cl2; supporting electrolyte: 0.1 m Bu4NPF6; scan rate 0.1 V s−1.

Table 1. Oxidation and reduction potentials derived from DPVs.

Compound

Eox[a]

[V]

Ered[a]

[V]

IF-TTF[b]

+0.24 (1e), +0.42 (1e), +1.09[c]

−2.31[c]

cIF-TTF

+0.36 (1e), +0.49 (1e), +1.09[c]

−2.04[c]

dcIF-TTF

+0.43 (1e), +0.55 (1e), +1.15[c]

−1.84[c]

  • [a] Potentials (vs Fc/Fc+) derived from DPV recorded in CH2Cl2 with 0.1 m Bu4NPF6 as supporting electrolyte. [b] IF-TTF with peripheral butyl groups instead of hexyl groups; Ref. [28, 29]. [c] Irreversible oxidation or reduction.
Table 2. Absorption (lowest-energy) and emission properties for the IF-TTF compounds.

Compound

λabs,max

[nm]

λems,max

[nm]

Stokes shift

[nm]

ϵabs,max

[M−1 cm−1]

IF-TTF

472

500

28

77 000

cIF-TTF

489

520

31

64 000

dcIF-TTF

504

531

27

63 000

Absorption and emission spectra

Absorption and emission spectra were recorded for the IF-TTF compounds dissolved in toluene (Figure 4). The addition of one cyano substituent, and then a second, leads to a systematic bathochromic shift in the respective spectra. The associated Stokes shifts are not appreciably altered by cyano substitution (Table 2, Figure S1). The molar absorption coefficient at the band maximum decreases slightly upon successive addition of the cyano group. The quantum yield of fluorescence measured for these IF-TTF molecules is small (≤0.001), which is consistent with the fact that the parent TTF shows no fluorescence at all.39

Details are in the caption following the image

Scaled absorption (A) and emission (B) spectra for the IF-TTF compounds recorded in toluene solutions. Values of the band maxima and Stokes shifts are shown in Table 2. For recording emission spectra, IF-TTF was excited at 350 nm and both cIF-TTF and dcIF-TTF were excited at 360 nm.

Quantum yield of IF-TTF-sensitized O2(a1Δg) production

Quantum yields of O2(a1Δg) production, ΦΔ, sensitized by the IF-TTF molecules were determined by monitoring the intensity of the time-resolved O2(a1Δg)→O2(X3Σg) phosphorescence signal at 1275 nm produced upon pulsed laser irradiation of the IF-TTF molecule, and then comparing this intensity to that obtained upon irradiation of a standard sensitizer for which ΦΔ is known.

For experiments performed in air-saturated toluene-h8, the IF-TTF-sensitized O2(a1Δg) phosphorescence signal was very weak (see Figure S2). Nevertheless, a signal was observed, and the lifetime obtained from the decay rate of this signal corresponded to what is expected for the lifetime of O2(a1Δg) in toluene-h8. Comparison of these IF-TTF O2(a1Δg) data to those obtained from a standard sensitizer indicated that, for all three IF-TTF compounds, the quantum yield of O2(a1Δg) production was ≤0.001 (see Figure S3).

To corroborate the data obtained in toluene-h8, we repeated all O2(a1Δg) phosphorescence experiments in toluene-d8. The advantage of this approach is that we exploit the established H/D solvent isotope effect on the lifetime of O2(a1Δg). This effect results in an appreciable increase in the quantum efficiency of O2(a1Δg) phosphorescence at 1275 nm.40 In short, in toluene-d8, increased signal-to-noise levels provide a more accurate assessment of the IF-TTF-sensitized O2(a1Δg) yields. The data thus obtained indicate that the quantum yield of O2(a1Δg) production for these IF-TTF molecules is indeed ≤0.001.

A small quantum yield of O2(a1Δg) production is desirable for a molecule designed to be used in an organic solar cell. The combination of the small quantum yields for O2(a1Δg) production and fluorescence implies that, in the absence of any kinetically competing bimolecular process (e. g., intermolecular electron transfer), almost all the excitation energy imparted upon irradiation of these IF-TTF molecules is dissipated via nonradiative deactivation

Rate constant for O2(a1Δg) removal mediated by the IF-TTF molecules

In the context of assessing the photooxidative stability of our IF-TTF compounds, it is useful to quantify the total rate constant for O2(a1Δg) removal, ktotal, mediated by the IF-TTF compound. When the given IF-TTF is dissolved in a liquid solvent, the reciprocal lifetime of O2(a1Δg), τΔ−1, is expressed as shown in Equation (1), where ksolv is the pseudo first order rate constant for solvent-mediated O2(a1Δg) deactivation and ktotal is the bimolecular rate constant for IF-TTF-mediated O2(a1Δg) deactivation.
urn:x-wiley:18645631:media:cssc202202320:cssc202202320-math-0001(1)
In turn, as shown in Equation (2), ktotal can be expressed as the sum of the rate constant for reaction with O2(a1Δg), krxn, and a rate constant for physical deactivation of O2(a1Δg), kphys, that is the IF-TTF-mediated O2(a1Δg)→O2(X3Σg) transition.
urn:x-wiley:18645631:media:cssc202202320:cssc202202320-math-0002(2)

As seen through Equation (1), ktotal is most readily obtained by quantifying the O2(a1Δg) lifetime as a function of the IF-TTF concentration in solution. The most accurate way of quantifying the O2(a1Δg) lifetime is to monitor the O2(a1Δg)→O2(X3Σg) 1275 nm phosphorescence in time-resolved experiments.41 For such experiments to quantify ktotal, it is generally desirable to have an independent source of O2(a1Δg) that would not interfere with the interaction between O2(a1Δg) and the added solute (i. e., IF-TTF in this case). Over the past ≈40 years in which the very weak O2(a1Δg) phosphorescence has been used to quantify ktotal,5, 19 O2(a1Δg) has been produced by energy transfer from a photosensitizer. This approach clearly has the disadvantage that the added photosensitizer can itself quench O2(a1Δg) and/or interfere with the solute for which the value of ktotal is desired. We recently resolved the disadvantages of using a photosensitizer to produce O2(a1Δg) under such conditions by demonstrating that O2(a1Δg) can be produced by irradiating oxygen itself at 765 nm in a sensitizer-free solvent.40, 42, 43 In this case, we pump the O2(X3Σg)→O2(b1Σg+) transition, and O2(b1Σg+) rapidly decays to form O2(a1Δg) with essentially unit quantum efficiency. The amount of O2(a1Δg) thus produced is sufficient to be detected in a 1275 nm phosphorescence experiment with a good signal-to-noise ratio.

Solutions of each IF-TTF molecule were prepared in toluene-d8, and O2(a1Δg) was produced in these systems upon irradiation at 765 nm. The 1275 nm phosphorescence thus obtained in time-resolved experiments had an acceptable signal-to-noise ratio to yield an accurate value for the O2(a1Δg) lifetime from the associated single exponential fit to the data (e. g., see Figure 5). As noted in the previous section, we used toluene-d8 to exploit the established H/D solvent isotope effect on the lifetime of O2(a1Δg) that, in turn, results in an increase in the quantum efficiency of O2(a1Δg) phosphorescence at 1275 nm.40 The longer lifetime of O2(a1Δg) in toluene-d8 (i.e., a smaller value of ksolv in Equation (1)) also facilitates access to the kinetically competing process in which the IF-TTF deactivates O2(a1Δg) (i.e., the ktotal[IF-TTF] term in Equation (1)). To our knowledge, based on data from a variety of organic substrates, there is no evidence to indicate that H/D isotopic substitution in the solvent affects values of ktotal thus obtained.

Details are in the caption following the image

Example of the time-resolved 1275 nm O2(a1Δg) phosphorescence signal obtained upon irradiating oxygen at 765 nm in air-saturated toluene-d8. These data were recorded for a solution containing 0.2 mm IF-TTF and used in a plot such as that shown in Figure 6 to obtain the rate constant for O2(a1Δg) removal by the IF-TTF. The red line shows a single exponential fit to the data that yields the O2(a1Δg) lifetime, τΔ.

The data thus obtained were plotted according to Equation (1) (e. g., Figure 6 for dcIF-TTF), and the resultant values of ktotal are listed in Table 3. As shown in Figure 6, it is important to note that the intercepts in these plots of Equation (1) are consistent with the lifetime of O2(a1Δg) determined in an independent experiment performed in the absence of the IF-TTF (τΔ in toluene-d8=314±6 μs).40

Details are in the caption following the image

Plot of the reciprocal O2(a1Δg) lifetime, τΔ−1, against the concentration of dcIF-TTF in toluene-d8. O2(a1Δg) was produced by direct pumping of oxygen at 765 nm in a sensitizer-free system and detected in a time-resolved 1275 nm phosphorescence experiment (see Figure 5). Errors on the individual data points are reflected in the size of the dot used as a symbol. The line shows the result of a linear fitting function. The data point at the intercept was determined in toluene-d8 without added dcIF-TTF; as such, it is a real data point used in the fitting function. Similar plots were obtained for IF-TTF and cIF-TTF (not shown). The slopes of these plots yield the total rate constant for O2(a1Δg) removal/deactivation by the IF-TTF (see Table 3).

Table 3. Rate constants for the IF-TTF-mediated removal of O2(a1Δg) in toluene.

Compound

ktotal[a]

[106 s−1m−1]

krxn[b]

[106 s−1m−1]

krxn[c]

[106 s−1m−1]

krxn/ktotal[b]

IF-TTF

7.2±0.1

3.0

3.7

0.42

cIF-TTF

5.2±0.1

1.8

2.3

0.35

dcIF-TTF

3.2±0.1

0.83

1.1

0.26

  • [a] Obtained from plots such as that shown in Figure 6. [b] Obtained in a C60-photosensitized experiment (see text and SI; Figure S4). [c] Obtained in an experiment using a zinc phthalocyanine as the O2(a1Δg) sensitizer. In this case, up to 20 % of the phthalocyanine “photobleached” upon irradiation (see text and SI; Figure S5).

Our data reveal a systematic and incremental decrease in ktotal upon increasing the number of cyano substituents on the IF-TTF framework. The magnitude of ktotal for the unsubstituted IF-TTF, ≈7×106 s−1m−1, indicates that this compound removes O2(a1Δg) with moderate reactivity; many compounds with larger values of ktotal are known.19 Most importantly, however, the decrease in ktotal with an increase in the number of cyano substituents is consistent with expectation. Considering Equation (2), this cyano-dependent decrease could reflect a decrease in the magnitude of kphys. In this case, the cyano groups could decrease the ability of the IF-TFF to donate charge to O2(a1Δg) and thereby mitigate the role of a charge-transfer mediated deactivation channel, a well-established phenomenon that facilitates the O2(a1Δg)→O2(X3Σg) nonradiative transition.5 Indeed, our electrochemical data reveal a significantly weakened donor strength upon introducing cyano substituents (Table 1). The cyano-dependent decrease in the magnitude of ktotal could also reflect a decrease in the magnitude of krxn, discussed in the Introduction as a motivation for this study.

Rate constant for O2(a1Δg) removal by reaction with the IF-TTF molecules

Using a previously-established protocol,43 we quantified krxn by monitoring the decrease in the absorbance of a given IF-TTF molecule in the presence of a known amount of O2(a1Δg). Of course, implicit in this approach is the assumption that the only process that results in the change of the IF-TTF absorbance is reaction with O2(a1Δg), a point that we discuss further below.

The sensitizer-free approach of generating O2(a1Δg) by irradiation of oxygen at 765 nm (see above) did not generate enough O2(a1Δg) to result in measurable changes in the IF-TTF concentration over practical periods of time. Likewise, we have established that the IF-TTF molecules themselves do not sensitize an appreciable amount of O2(a1Δg). Thus, to achieve an appreciable amount of O2(a1Δg)-mediated IF-TTF bleaching over an experimentally practical period, we had to irradiate a co-dissolved O2(a1Δg) sensitizer. In this regard, and from a device-oriented perspective, this likely mimics the case where other molecules in a multi-component system would be the source of O2(a1Δg) that could oxygenate the IF-TTF molecules.

We performed independent experiments using two different O2(a1Δg) sensitizers: C60 and zinc phthalocyanine (ZnPc).6, 43, 44 One advantage of using these molecules is that they absorb light at wavelengths where the absorbance of the IF-TTF molecules is small or zero. Examples of irradiation-dependent changes in IF-TTF concentration used to obtain krxn are provided in the SI (Figures S4 and S5).

As shown in Figures S4 and S5, our protocol for obtaining krxn from the IF-TTF bleaching data involves extrapolating changes in the IF-TTF absorbance to zero irradiation time. In this way, we mitigate the influence of absorbance from IF-TTF degradation products. Indeed, values of krxn thus obtained were the same, within our margin of error, whether we accounted for the slight absorbance due to degradation products, or not. In the ZnPc experiment, the data showed that up to 20 % of the ZnPc bleached upon irradiation (see Figure S5), and this could influence the values of krxn obtained. In contrast, there was no evidence of sensitizer bleaching in the C60 experiment.

Despite the caveat raised in the preceding paragraph, values of krxn obtained from the experiments with the different O2(a1Δg) sensitizers are similar (Table 3). Irrespective of whether we ascribe confidence in the absolute values of krxn, the relative cyano-dependent changes in krxn are consistent with both our original expectation and with the corresponding changes in ktotal.

The sensitizer-dependent differences in krxn shown in Table 3 could also reflect interactions between the IF-TTF and the sensitizer. Suspicions about such interactions are justified by experiments performed as a function of oxygen concentration. For air-saturated samples, irradiation-dependent changes in the IF-TTF concentration show curvature when plotted as a function of the elapsed photolysis time (see SI; Figures S4 and S5). As an aside, it is important to note that the extent of such curvature decreased with an increase in the number of added CN groups. In N2-saturated samples, the IF-TTF concentration likewise decreased upon irradiation of the co-dissolved photosensitizer, but these changes are linear as a function of the elapsed photolysis time. There is precedence for such observations, which imply kinetic competition between different mechanisms of solute degradation.45, 46 In this case, a O2(a1Δg)-mediated channel for IF-TTF degradation appears to compete with an oxygen-independent process that may involve a bimolecular reaction of the IF-TTF with the sensitizer.

Keeping these caveats in mind, it is important to note that the fraction krxn/ktotal decreases with an increase in the number of cyano groups added (Table 3). This points to a key goal in the construction of an air-stable photoresponsive device: photoactive molecules should facilitate the physical deactivation of O2(a1Δg) (kphys) at the expense of reaction with O2(a1Δg) (krxn).

Degradation products analysis

To identify the degradation products, we treated a more concentrated sample of IF-TTF (2.2 mm) with O2(a1Δg) in C6D6, photosensitized by co-dissolved C60. From 1H NMR spectroscopic studies (see SI), the products were identified as the ketone 3 and 1,3-dithiol-2-one 447 (Figure 7). These products are consistent with an apparent cycloaddition (not concerted, see below) of one fulvene C=C with O2(a1Δg) followed by ring-opening of the resultant dioxetane. Product characterization was further supported by mass spectrometry, TLC inspections, and IR spectroscopic studies. For the latter, the IR spectrum of the degradation products showed characteristic absorbances at 1706, 1529, and 1481 cm−1 (see SI; Figure S7).

Details are in the caption following the image

Products formed upon reaction between IF-TTF with O2(a1Δg).

For comparison, Bryce and co-workers48 found that photolysis of aerated solutions of a 9,10-bis(1,3-dithiol-2-ylidene)-9,10-dihydroanthracene compound (“anthraquinone-extended“ TTF) gave a photodegradation product assigned to the corresponding ketone, with one dithiafulvene unit remaining.

Computational study on reaction mechanism

To shed further light on the reaction with O2(a1Δg), we examined the compounds shown in Scheme 2 (i. e., the IF-TTF analogues with peripheral Me substituents) in a computational study using Gaussian 16.49 The energy of O2(a1Δg) was obtained using AP-UM06-2X/6-31+G(d,p).

Details are in the caption following the image

Illustration of the lowest-energy path for the reaction between the IF-TTFs and O2(a1Δg), with calculated energies (kcal mol−1) for the stationary points (using AP−M06-2X/6-31+G(d,p), with toluene as the solvent). All energies obtained through both the restricted and unrestricted calculations are provided in the SI, as are the results of calculations for the geometric isomer in which the peroxide is formed on the outer side of the bond between the IF and TTF moieties. Although the numerical values of the energies shown in this figure accurately represent our computations, the illustration itself is not drawn to represent the appropriate scale.

The computed reaction sequence for IF-TTF-Me and the two related cyano derivatives cIF-TTF-Me and dcIF-TTF-Me is shown in Scheme 2. Formation of the dioxetane intermediate (DOI) was found to occur in a stepwise sequence via a peroxy-biradical intermediate (POI; the transition state structure towards this intermediate is shown in Figure 2). Two routes for these two steps were found, which differed in the orientation of the O in the C−O−O of the peroxy-biradical, pointing either towards the center of the molecule (energies included in Scheme 2) or away from the center (energies included in SI). Finally, the dioxetane intermediate opens to form the carbonyl product, 3-Me, c-3-Me, or dc-3-Me, and the 1,3-dithiol-2-one 4-Me.

Most importantly, and as shown in both Figure 2 and Scheme 2, we found that the energies of TS1 increased incrementally with the addition of the cyano substituents. Thus, the calculations confirm the reduced reactivity of IF-TTFs towards O2(a1Δg) as more cyano groups are introduced. We also see that the overall oxidation reaction is more energetically favorable for IF-TTF-Me than it is for the cyano derivatives (−114.18 vs −113.61 vs −111.79 kcal mol−1).

A similar computational study was performed for the reaction of O2(a1Δg) with 3-Me (See SI). Here the related structures for TS1 were found to be 0.5–1.0 kcal mol−1 more energetic than those of IF-TTF-Me. Thus, the product of the first oxygenation reaction, 3-Me, is expected to be less reactive towards O2(a1Δg) in line with the fact that we did not detect experimentally the diketone resulting from further reaction of 3.

Organic photovoltaic device studies

To study the effect of the cyano-functionalized IF-TTFs on the stability of organic photovoltaic (OPV) devices, the IF-TTFs were integrated as hole interfacial layers in non-fullerene acceptor based OPV cells. The rationale behind this was the HOMO level of the dcIF-TTF at around −5.2 eV, as extracted from the voltammetry measurements (first oxidation at 0.43 V; HOMO energy of Fc=−4.8 eV against vacuum50), is consistent with (a) the HOMO level of other high performing OPV electron donor systems,51 and (b) the reported use of other TTF derivatives as the hole transport material in solar cells.52 In addition, the relatively large HOMO–LUMO gaps (2.3 to 2.6 eV, see above) of the IF-TTFs provide efficient electron/exciton blocking at the anode interface.

OPV devices were prepared in an inverted device configuration, having a structure of ITO/ZnO/PBDB-T:N2200/IF-TTF or dcIF-TTF/MoO3/Ag, using ultrathin interfacial layers of IF-TTF and dcIF-TTF to study the effect of the cyano functionalization in these devices. Thickness optimization of the IF-TTF interfacial layers was conducted and is shown in SI, Tables S2 and S3. The device configuration with schematic energy level diagram of each component is depicted in Figure 8a, and the photovoltaic parameters are shown in Table 4.

Table 4. Photovoltaic performance of organic photovoltaic devices with different hole interfacial materials.

Interfacial layer

VOC

[V]

JSC

[mA cm−2]

Integrated JSC

[mA cm−2]

FF

[%]

PCE

[%]

No IF-TTF

0.88

12.9

11.5

61.6

7.0

IF-TTF

0.87

10.5

10.1

59.5

5.4

dcIF-TTF

0.87

11.1

10.2

59.5

5.7

The reference OPV device without hole interfacial materials exhibited a power conversion efficiency (PCE) of 7.0 %, whereas the optimized devices with IF-TTF and dcIF-TTF yielded PCEs of 5.4 % and 5.7 %, respectively. The full photovoltaic parameters are shown in Table 4. The current density-voltage (J-V) characteristics measured under simulated 1 Sun AM1.5G illumination at 100 mW cm−2, and external quantum efficiency (EQE) curves of the optimized devices, are shown in Figure 8b and c. These indicate slightly lower short-circuit current density JSC for the IF-TTF and dcIF-TTF devices, but almost similar open-circuit voltage VOC and fill factor FF. The integrated JSC values (from EQE measurements) for all the devices align well with the values obtained from the J-V measurements, as summarized in Table 4. The slightly higher JSC (with similar FF values) in the OPV cells without IF-TTF could arise due to interference mediated shifts in the electromagnetic field distribution in the thin film solar cells. Further analysis would be needed to address this issue. A further comparison of common device lifetime performances is provided in Table 5.

Details are in the caption following the image

(a) Device configuration with energy level diagram of each component. (b) J-V curves measured under AM 1.5 G irradiation (100 mW cm−2) illumination, and (c) external quantum efficiency (EQE) measurements.

Table 5. Comparison of common device lifetime performances.

Interfacial

layer

tburn-in[a]

[h]

PCEburn-in[b]

[%]

PCE80[c]

[%]

t80[d]

[h]

tlifetime[e]

[h]

PCEinitial[f]

[%]

(PCEinitialPCEburn-in)/

PCEinitial[g]

[%]

APGlifetime[h]

[Wh m−2]

No IF-TTF

7.2

2.6

2.1

11.4

4.1

6.5

60.0

189.0

dcIF-TTF

9.6

2.7

2.2

22.8

13.2

5.6

51.5

431.6

IF-TTF

8.6

2.4

1.9

16.5

7.9

5.2

53.9

266.4

  • [a] Burn-in time (tburn-in). [b] Power conversion efficiency (PCE) at the end of the burn-in period (PCEburn-in). [c] PCE reduced to 80 % compared to the burn-in (PCE80). [d] Time at which PCE is reduced to 80 % compared to the burn-in (tT80). [e] Period between burn-in and T80, that is the lifetime of the device (tlifetime). [f] Extracted initial PCE (PCEinitial). [g] Magnitude of burn-in: (PCEinitialPCEburn-in)/PCEinitial. [h] Accumulated power generated over the lifetime of the device (APGlifetime).

The device containing dcIF-TTF shows a slightly improved PCE compared to the device with IF-TTF. Most importantly, however, the stability of the solar cells containing the cyano functionalized compound is appreciably better, as illustrated in Figure 9.

Details are in the caption following the image

Device stability of OPV devices measured under continuous 1 Sun AM1.5G illumination in ambient air (ISOS-L-1, no encapsulation).

Specifically, upon continuous 1 Sun AM1.5G light illumination in ambient conditions, following the ISOS-L-1 degradation protocol (air stability without encapsulation), the PCE of the OPV of reference devices (no IF-TTF interlayer) drops to 8 % of its initial PCE after 52 h (Figure 9). Meanwhile, devices with IF-TTF or dcIF-TTF display an improved stability retaining 17 % and 26 % of its initial PCEs within the same duration of ISOS-L-1 degradation, respectively. The lifetime curves were fitted using biexponential decay function to extract the stability parameters. The superior stability of dcIF-TTF and IF-TTF devices is reflected in both a strong slowing down of their burn-in period and a decrease of the burn-in magnitude, resulting in a strong increase in the overall accumulated power generated over their lifetime, APGlifetime (for fitting parameters, SI; Table S4). The evolution of the photovoltaic parameters, including VOC, JSC, and FF, upon the ISOS-L-1 degradation is shown in the SI (Figure S9). It is seen that it is mainly the JSC and FF that worsen in these experiments, and that the cyano group tends to mitigate especially rapid drops in JSC and thus in PCE. This leads to an overall stabilization effect of the OPV cells with the inclusion of the cyano-functionalized IF-TTF, in comparison to the non-functionalized IF-TTF molecule. This correlates well to the observed reduced reactivity of the dcIF-TTF molecule to O2(a1Δg).

The improved device stability upon cyano addition to the IF-TTF interlayer molecule can, as noted, be directly linked to the IF-TTF's improved photostability upon this molecular engineering approach. Because the cyano addition promotes more physical deactivation of O2(a1Δg) at the expense of reaction with O2(a1Δg), this could also assist in stabilizing the active layer molecules in the solar cells, i. e. unwanted O2(a1Δg) that would otherwise react with active layer molecules gets deactivated at the IF-TTF interlayer (and more efficiently at the dcIF-TTF interlayer). This is a point upon which further studies should be based.

Conclusion

New redox-active chromophores that absorb visible light were conveniently prepared by Pd-catalyzed cyanation reactions of iodo-functionalized indenofluorene-extended tetrathiafulvalenes (IF-TTFs). We ascertained that these cyano-substituted IF-TTFs are poor photosensitizers for the generation of O2(a1Δg), which is a particularly attractive property for exploiting such molecules as components of photovoltaic devices. Moreover, the reactivity of the IF-TTFs towards O2(a1Δg) was significantly reduced by introducing the cyano substituents at the indenofluorene core. This latter feature of the cyano-functionalized IF-TTFs is likewise a significant attribute for use in a photovoltaic device where other components may generate O2(a1Δg) in greater yield. The reduced reactivity towards O2(a1Δg) is consistent with a reduced nucleophilicity of the cyano-functionalized compounds, which is, in turn, consistent with cyclic and differential pulse voltammetry studies of the IF-TTFs. Computations support the hypothesized formation of a dioxetane in the reaction between the IF-TTF and O2(a1Δg) and indicate that this occurs in a stepwise process. The computations also indicate that introduction of a cyano substituent on the indenofluorene core increases the activation energy for the reaction with O2(a1Δg). In future work, it could be interesting to incorporate cyano groups at other positions of the core or to introduce more than two groups. It could also be relevant to study other electron-withdrawing groups (as long as the bonds are not readily photo-cleaved; for example carbon-halogen bonds that would generate radicals).

With this example of molecular engineering of organic compounds and their use in photovoltaic devices, we can appreciably decrease the extent of component photooxidation and, thereby, mitigate the need for encapsulation protocols to prevent degradation and device failure. There is clearly room for further improvement of the devices; for example, in regard to power conversion efficiency. However, the aim of this work has instead been to show how our molecular engineering approach for imparting stability is not only valid in the solution phase, but also in the first heterojunction devices based on the IF-TTF components. Here we focused on using the IF-TTF within hole transport interfacial layers, while previous work25 focused on using IF-TTF as donor component in dye-sensitized solar cells.

Experimental Section

Synthesis of cIF-TTF

An Ar-degassed suspension of compound 1 (101 mg, 86 μmol) in DMF (40 mL) was heated to 110 °C until all the solids were dissolved. Pd(PPh3)4 (32 mg, 280 μmol, 32 mol %) and Zn(CN)2 (53 mg, 451 μmol, 5.00 equiv.) were added to the solution, and it was allowed to stir at 110 °C overnight upon which a color change from orange to red was observed. The reaction progress was monitored by TLC, which still showed starting material after 20 h of stirring. Additional Pd(PPh3)4 (35 mg, 30 μmol, 35 mol %) and Zn(CN)2 (55 mg, 468 μmol, 5.00 equiv.) were then added to the reaction mixture. After stirring at 110 °C for an additional 4 h, no more starting material was visible on TLC, and the reaction was quenched with H2O (50 mL) and cooled to room temperature. The aqueous phase was extracted with CH2Cl2 (3×40 mL), and the combined organic phases were washed with H2O (3×100 mL), dried over MgSO4 and concentrated in vacuo. The residue was purified by flash column chromatography (SiO2, 40–60 % CH2Cl2/heptane), which yielded compound cIF-TTF (45 mg, 48 μmol, 55 %) as an orange solid. Rf=0.25 (50 % CH2Cl2/heptane). M.p.: 94–97 °C. NMR spectra recorded at a concentration of 5.78 mm. 1H NMR (500 MHz, CDCl3) urn:x-wiley:18645631:media:cssc202202320:cssc202202320-math-0003 7.93 (s, 2H), 7.86 (s, 1H), 7.83 (d, J=7.3 Hz, 1H), 7.82 (d, J=7.8 Hz, 1H), 7.68 (d, J=7.9 Hz, 1H), 7.51 (d, J=7.8 Hz, 1H), 7.42 (td, J=7.9, 1.1 Hz, 1H), 7.33 (td, J=7.3, 1.1 Hz, 1H), 3.05–2.98 (m, 8H), 1.79–1.72 (m, 8H), 1.53–1.46 (m, 8H), 1.36–1.32 (m, 16H), 0.92–0.89 (m, 12H) ppm. 13C NMR (126 MHz, CDCl3) δ 141.75, 140.85, 138.83, 137.95, 137.87, 137.69, 137.24, 135.53, 135.18, 134.93, 130.14, 129.94, 128.87, 128.69, 128.04, 127.36, 126.05, 125.70, 123.10, 120.60, 120.56, 119.82, 119.65, 119.06, 115.01, 113.99, 109.03, 36.99, 36.95, 36.86, 36.74, 31.56, 31.54, 31.52, 30.00, 29.97, 29.94, 29.88, 28.49, 28.45, 22.73, 22.71, 14.21, 14.19 ppm. 6 aliphatic signals missing, presumably due to overlap. HRMS (MALDI+ FT-ICR, dithranol): m/z=943.2583 [M]⋅+, calcd. for [C51H61NS8+] m/z=943.2564.

Synthesis of dcIF-TTF

To an Ar-flushed flask were added compound 2 (101 mg, 86.0 μmol; for synthesis of compound 2, see Supporting Information, SI), Pd2dba3 (45.1 mg, 49.0 μmol, 0.58 equiv.), Zn(CN)2 (86.2 mg, 734 μmol, 8.50 equiv.) and Zn powder (8.90 mg, 136 μmol, 1.58 equiv.). DMF (20 mL) was added, and the suspension was sonicated for 5 min before being degassed with Ar for 10 min. P(tBu)3 (0.102 mL, 102 μmol, 1 m in toluene, 1.18 equiv.) was added dropwise, and the suspension was heated to 135 °C under an inert atmosphere upon which all the solids were dissolved, and a color change from yellow to red was observed. The reaction mixture was stirred overnight, after which it was cooled to room temperature and filtered through a plug of SiO2 (CH2Cl2). The organic phase was washed with H2O (3×70 mL), dried over MgSO4 and concentrated in vacuo. The product was purified by flash column chromatography (SiO2, toluene) followed by precipitation from CH2Cl2 with MeOH and subsequent trituration of the solid with MeOH (3×7 mL). This yielded compound dcIF-TTF (50 mg, 55.0 μmol, 60 %) as a red solid. Rf=0.66 (toluene). M.p.: 183–185 °C. NMR spectra recorded at a concentration of 5.28 mm. 1H NMR (500 MHz, CDCl3) δ 7.86 (s, 2H), 7.83 (s, 2H), 7.80 (d, J=7.8 Hz, 2H), 7.54 (d, J=7.8 Hz, 2H), 3.06–3.03 (m, 4H), 3.01–2.98 (m, 4H), 1.78–1.72 (m, 8H), 1.52–1.47 (m, 8H), 1.37–1.32 (m, 16H), 0.93–0.89 (d, 12H) ppm. 13C NMR (126 MHz, CDCl3) δ 141.98, 141.00, 137.42, 136.44, 135.34, 130.93, 128.80, 128.37, 126.10, 120.22, 119.97, 118.43, 114.73, 109.78, 37.08, 36.90, 31.56, 31.54, 29.94, 29.88, 28.48, 28.46, 22.73, 22.70, 14.21, 14.19 ppm. HRMS (MALDI+ FT-ICR, dithranol): m/z=968.25529 [M]⋅+, calcd. for [C52H60N2S8+] m/z=968.25167.

Absorption spectroscopy

Absorption spectra were recorded using a Shimadzu model UV3600 spectrometer and fluorescence spectra using a Fluoromax P spectrometer (Horiba Jobin Yvon). Fluorescence quantum yields were determined using a Quantaurus-QY Plus spectrofluorometer (Hamamatsu C13534-33) with a high-power xenon light source (L13685-01) and A13686 optical filter.

Singlet oxygen studies

The equipment and procedures used to monitor and quantify O2(a1Δg) formation and decay have been described elsewhere.11, 53, 54 Briefly, samples were excited using the output of an amplified femtosecond pulsed laser operating at 0.5 kHz. Depending on the experiment, we tuned the output of the amplifier to obtain 765 nm or we frequency-doubled the amplifier output at 840 nm to obtain the excitation wavelength of 420 nm. The characteristic 1275 nm O2(a1Δg) phosphorescence was isolated with a 1064 nm long-pass filter and a 1270 nm (20 nm FWHM) band-pass filter and detected using a cooled NIR-sensitive PMT (Hamamatsu). For measurements of the O2(a1Δg) quantum yield, ΦΔ, 1H-phenalenone (PN) in toluene (ΦΔ=0.97±0.03)55, 56 served as the reference photosensitizer. All O2(a1Δg) experiments were performed in 1-cm path length cuvettes using samples with an absorbance of 0.1, or less, at the excitation wavelength.

For the photosensitized reactions, zinc phthalocyanine and C60 were obtained from Sigma Aldrich and used as received.

Computational methods

Structures were optimized using RM06-2X/6-31+G(d,p), while transition states (TS) were also optimized with the broken symmetry UM06-2X/6-31+G(d,p) method using Gaussian 16.48 Energy refinement was done with the approximate spin projection method. Geometries of singlet and triplet oxygen were obtained with the (R/RO)M06-2X/6-31+G(d,p) methods. Solvation effects (toluene) were included using the integral equation formalism of a polarizable continuum model (IEF-PCM). Further details on the methods can be found in the SI, including the geometries of intermediates and TSs.

Device fabrication and characterization

OPV devices were prepared with a device structure of ITO/ZnO/PBDB-T:N2200/IF-TTF or dcIF-TTF/MoO3/Ag. The ITO-coated glass substrates were firstly cleaned with detergent, and then ultrasonically washed with deionized water, acetone, and isopropanol sequentially. The cleaned and dried substrates were treated in a UV-Ozone cleaner for 20 min and transferred into a glovebox. A ZnO solution was spin-coated on top of ITO substrates at 3000 rpm for 60 s, for 35 nm thickness. Then the samples were baked at 130 °C for 10 min on a hot plate. The solution of PBDB-T/N2200 (2 : 1, 14 mg/mL in chlorobenzene solvent) was spin-coated on top of the ZnO, following thermal annealing at 100 °C for 10 min. The ultrathin interfacial layers; IF-TTF and dcIF-TTF (0.2 mg/mL in toluene) were spin coated on top of the active layer, and then the substrates were transferred to a glovebox-integrated thermal evaporator for deposition of MoO3 and Ag. 10 nm of MoO3 and 100 nm Ag were thermally deposited under high vacuum with the device area of 2.8 mm2 defined by the aperture of the deposition mask. The photovoltaic performance of the OPV devices were measured using a calibrated air mass (AM) 1.5 G solar simulator (Oriel Sol3A Class AAA solar simulator, Sun 3000, Abet Technologies Inc., USA) with a light intensity of 100 mW cm−2 adjusted using a standard PV reference cell (2 cm×2 cm monocrystalline silicon solar cell, calibrated at NREL, Golden, CO) and a computer controlled Keithley 2400 (Keithley Instruments Inc., USA) source measure unit. The external quantum efficiency spectrum was determined by using the Bentham PVE300 (Bentham Instruments Ltd.), equipped with a dual xenon/quartz halogen as light source, a monochromator, an optical chopper, a lock-in amplifier, and a calibrated silicon photodetector. The lifetime data were extracted by periodically measuring J-V characteristics under accelerated conditions according to the ISOS-L-1 degradation protocol standards. The devices were continuously illuminated in ambient air at room temperature, using an InfinityPV ISOSun solar simulator consisting of an Osram metal halide lamp delivering 1 kW m−2, and characterized using a Keithley 2602 A source-measure unit mounted on a Keithley 3706A multiplexer unit, all of which were controlled by home-built Labview software.

Acknowledgments

The Independent Research Fund Denmark, Technology and Production Sciences (0136-00081B) and The Novo Nordisk Foundation (NNF20OC0061574) are acknowledged for financial support.

    Conflict of interest

    The authors declare no conflict of interest.

    Data Availability Statement

    The data that support the findings of this study are available from the corresponding author upon reasonable request.