Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Origin, Nature, and Location of Defects in PM6:Y6 Organic Solar Cells

Origin, Nature, and Location of Defects in PM6:Y6 Organic Solar Cells IntroductionOrganic semiconductors can provide added functionality compared to conventional semiconductors in electronic applications owing to their lightweight, biocompatible, and flexible nature combined with easy large‐scale processing at room temperature and ambient pressures.[1–3] A large variety of devices have emerged ranging from organic field‐effect transistors (OFETs), organic electrochemical transistors, organic memristors, organic light‐emitting diodes (OLEDs) to organic photodetectors and organic photovoltaics (OPVs).[4–12] Despite intense research interest and potential advantages, applications on large scale remain primarily limited to OLEDs.To further mature technologies based on organic semiconductors, limiting factors should be identified and eliminated. A universal obstacle for organic semiconductors is the presence of defects, acting as a source of charge‐trapping and of (non‐radiative) recombination.[13–15] Knowledge on defect formation is vital to formulate a targeted approach to lower their number and impact. Given their scarce nature, however, defects are notoriously difficult to characterize and their occurrence and energy levels are often deduced from device measurements combined with physical simulation models.[16–20]The origin of defects is a longstanding debate in the field. Defect formation has been tentatively ascribed to (synthetic) contaminants, conformational changes, or molecular defects resulting from imperfect synthesis, or by the influence of ambient atmosphere sometimes accelerated by ultraviolet (UV) or visible light.[21,22,31–35,23–30] Particularly for OFETs, it has been demonstrated that ambient air with or without light has a direct influence on the defect density.[28–30,36] Oftentimes, oxygen or ozone are identified as the culprit in defect formation.[27,37,38] However, recently, the introduction of an iso‐energetic defect by water‐containing nanovoids was proposed.[39–42] In general, the role of the different constituents of ambient air (e.g., oxygen, ozone, and water) remains elusive.In this work, we use highly sensitive external quantum efficiency (EQE) measurements to directly record the photogeneration of charges by defects in an OPV device architecture. It has been shown that EQE measurements up to 100 dB dynamic range are feasible by carefully eliminating noise in the measurement.[43] With this technique it is possible to measure very small photocurrents that originate from excitation of charges trapped in defect states.[44,45] The magnitude of the EQE response is proportional to the defect density, and by following the response we can thus track the defect concentration.Here, we investigate the origin and nature of defect states in PM6:Y6 blends (PM6 is poly[(2,6‐(4,8‐bis(5‐(2‐ethylhexyl‐3‐fluoro)thiophen‐2‐yl)‐benzo[1,2‐b:4,5‐b′]dithiophene))‐alt‐(5,5‐(1′,3′‐di‐2‐thienyl‐5′,7′‐bis(2‐ethylhexyl)benzo[1′,2′‐c:4′,5′‐c′]dithiophene‐4,8‐dione)] and Y6 is 2,2′‐((2Z,2′Z)‐((12,13‐bis(2‐ethylhexyl)‐3,9‐diundecyl‐12,13‐dihydro‐[1,2,5]thiadiazolo[3,4‐e]thieno‐[2″,3″:4′,5′]thieno[2′,3′:4,5]pyrrolo[3,2‐g]thieno‐[2′,3′:4,5]thieno[3,2‐b]indole‐2,10‐diyl)bis(methanylylidene))‐bis(5,6‐difluoro‐3‐oxo‐2,3‐dihydro‐1H‐indene‐2,1‐diylidene))dimalononitril). Sensitive EQE spectroscopy is used to study the change in defect response upon exposing either the active layer or complete devices to different components that are present in ambient air while keeping the sample in the dark. Rather than oxygen and water, we find that trace compounds in ambient air cause a distinct increase in defect response coupled to a decrease in device performance. By controlled exposure to O3 and N2O, we determine that oxidation by O3 is the source of defect formation in PM6:Y6 blends when exposed to ambient air. By studying semitransparent devices comprising either PM6:Y6, PM6, or Y6 active layers in inverted (n–i–p) and regular (p–i–n) cell architectures, we show that the defect response is dominated by PM6 and is indifferent to the device polarity. Additionally, we perform energy resolved‐electrochemical impedance spectroscopy (ER‐EIS) on films of PM6 and Y6 to determine the density of states (DOS). For PM6 we record the increase of a sub‐bandgap defect state upon O3 exposure and aging in ambient air. In contrast, the DOS of Y6 remains unaltered. By comparing optical simulations with the ratio of experimental EQE spectra of semitransparent cells with and without an additional optical spacer−mirror stack we determine that these oxidation‐induced defects are located at the back interface.Results and DiscussionVarious AtmospheresWe fabricated PM6:Y6 organic solar cells with an inverted (n–i–p) configuration (glass | indium tin oxide (ITO) | ZnO | PM6:Y6 | MoO3 | Ag) under inert conditions.[46] After spin coating the PM6:Y6 layer in a N2‐filled glove box, the semi‐finished solar cells were subjected to ambient atmosphere, to purified compressed air saturated with H2O for ≈48 h in the dark, or kept under N2 as a reference. Compressed air consists of N2 (≈78%), O2 (≈21%), Ar (≈1%), negligible amounts of H2O before saturating with H2O by bubbling through water, and is free of O3 (see Experimental Section for details). Ambient atmosphere contains potentially reactive trace gasses such as CH4, H2, N2O, CO, and O3 (listed most‐to‐less‐abundant).[47] After the optional exposure of the active layer to ambient air or to H2O‐saturated compressed air, the solar cells were completed by thermal evaporation of MoO3 and Ag as top contact in high vacuum.The current density – voltage (J–V) characteristics of these cells are presented in Figure 1a. As expected, pristine PM6:Y6 cells provide high efficiency,[48,49] and a similar high performance is found after aging the PM6:Y6 layer in H2O‐saturated compressed air prior to evaporating the top contact. Exposure of the PM6:Y6 layer to ambient atmosphere before completing the device, however, resulted in a distinct decrease in device performance with the short‐circuit current density (JSC), open‐circuit voltage (VOC), and fill factor (FF) being affected. The corresponding EQE spectra confirm the trend, with similar spectra for pristine films and films exposed to H2O‐saturated compressed air, while films exposed to ambient atmosphere resulted in a significantly lower EQE (Figure 1b).1Figurea) J–V measurements recorded with simulated AM1.5G light of ITO | ZnO | PM6:Y6 | MoO3 | Ag solar cells with a pristine active layer (red) or with active layers aged for ≈48 h in H2O‐saturated compressed air (green) or in ambient atmosphere (blue) before completing the solar cell with a MoO3 | Ag top contact. b) Corresponding EQE spectra versus wavelength. c) Semilogarithmic plot of the corresponding sensitive EQE spectra versus photon energy.Figure 1c shows a semilogarithmic plot of the sensitive EQE spectra of these cells versus photon energy. Focusing on the pristine cell, the above‐bandgap (>1.4 eV) EQE transitions to an Urbach tail region, characterized by an exponentially decreasing EQE from the bandgap at ≈1.4 eV until ≈1.05 eV with an Urbach energy of 23 meV.[50] Below 1.05 eV the onset of a defect response is visible, as a deviation from the exponential behavior, just before the spectrum becomes noise limited. Comparing the pristine sample with the sample aged in H2O‐saturated compressed air we find that the sensitive EQEs are identical. For layers exposed to ambient atmosphere, however, the EQE is lowered above the bandgap and also reveals an increase of the defect response by approximately one order of magnitude at photon energies less than 1.15 eV (Figure 1c). Thus, through a direct measurement of the defect response, we infer that exposure of the ITO | ZnO | PM6:Y6 stack to ambient atmosphere gives rise to an increase in defects. Further, since H2O‐saturated compressed air does not give a detectable change in the sensitive EQE spectrum, the cause must be a trace constituent of ambient atmosphere that is not present in H2O‐saturated compressed air. We also note that the defects observed in the sensitive EQE spectra correlate with a loss in performance of the solar cell. Aging the ZnO layer before PM6:Y6 deposition does not lead to a measurable change in device performance or defect response (Figure S1, Supporting Information). Therefore the reduced solar cell performance and increased defect response shown in Figure 1 are ascribed to the PM6:Y6 active layer.The significant losses in JSC, VOC, and FF for films exposed to ambient air are ascribed to extensive electron‐hole recombination primarily caused by hindered charge transport, rather than poor charge generation because at −2 V the photocurrent of the layer exposed to ambient air is very similar to that of the other two samples (Figure S2, Supporting Information). Hence, a reverse electric field allows charges, that otherwise recombine at short‐circuit, to be extracted. The fact that EQE is reduced over the entire spectral range is consistent with this explanation.We then studied the impact of exposure of a complete solar cell to different atmospheres and report the J–V characteristics and sensitive EQE spectra in Figure S3a,b (Supporting Information). For a complete device, H2O‐saturated compressed air does decrease device performance and increase the defect EQE, while dry compressed air does not. The top electrode (MoO3 | Ag) in conjunction with H2O must underlie these observations as the defect density of the ITO | ZnO | PM6:Y6 stack was not affected under these conditions. It is known that when MoO3 is exposed to H2O, the band bending that enables a selective hole contact vanishes as the work function of MoO3 is reduced by H2O.[51–53] This effect is commonly ascribed to the adsorption of H2O.[52,53] We postulate that for the pristine cell the band bending decreased the defect response by lowering the filled‐defect density near the interface. A schematic diagram is included in Figure S4 (Supporting Information). Upon exposure of MoO3 to H2O, the band bending is reduced resulting in more frequent occupation of the defect states and thus more states available for photogeneration of charges. Implicit to the proposed mechanism is the charge generation by defects occurring through the excitation of filled‐defect states.The change in energy levels of MoO3 due to H2O adsorption are reported to be reversible.[53] Indeed, we find that upon exposure to vacuum (≈5 × 10−7 mbar) or prolonged aging in N2 atmosphere, the defect response in sensitive EQE is again similar to the pristine cell (Figure S3c,d, Supporting Information). We conclude that the H2O adsorbed to MoO3 can be removed in vacuum, thereby restoring the band bending at the MoO3 interface and yielding a defect response comparable to a pristine cell (Figure S3c, Supporting Information). Similarly, prolonged storage in dry N2 also removes adsorbed H2O (Figure S3d, Supporting Information). In contrast, for an air‐exposed complete solar cell the defect response increases further rather than decreases over time in N2 atmosphere (Figure S3d, Supporting Information) and the increase therefore has a different underlying cause for aging in ambient atmosphere. These results strengthen the proof that the decreased band bending due to H2O‐exposed MoO3 underlies the observed increase in defect response for these devices exposed to H2O‐saturated compressed air as more filled defect states become available for photo‐excitation. Concurrently, since band bending occurs primarily near the PM6:Y6 | MoO3 interface, the defect contribution could be dominated by this interface.The increased, and reversible, defect response upon H2O exposure of finished cells closely resembles the defect response upon ambient air aging (Figure S3b, Supporting Information). Therefore, we attribute both defect responses as coming from the same defect. Additionally, this provides evidence that the defects present in pristine PM6:Y6 cells are the same as those formed upon exposure to ambient air.The reversible increase in defect response of a finished device upon aging in H2O‐saturated compressed air could alternatively be explained by the presence of water‐induced defects, as often discussed in literature.[41,54] The reversibility of this process would then be ascribed to H2O leaving the active layer. However, the same defect EQE is observed when aging a finished cell in ambient atmosphere, but in that case the effect is irreversible. Therefore these defects cannot simply be ascribed to the uptake of water because this mechanism should be reversible (or irreversible) regardless the aging in ambient air or H2O‐saturated compressed air. In fact, the results shown in Figure 1c give no indication for the presence of water‐induced defects. Of course, it is possible that a change in water‐induced defect‐density is not recorded in the EQE when these defects do not generate a photocurrent when excited.Trace Constituents of Ambient AirHaving found that a trace constituent of ambient air likely causes the formation of defects in PM6:Y6 films, the question arises which compound is responsible for this effect. Considering the various trace compounds in ambient air, ozone (O3) is known for its reactivity toward organic substrates,[47] especially in oxidizing unsaturated bonds. To investigate its effect, we exposed semi‐finished inverted PM6:Y6 devices to O3‐enriched air for times between 30 s and 5 min (see Experimental Section for details), followed by completing the device with a MoO3 | Ag top contact. The J–V and EQE characteristics (Figure 2a,b) reveal that such short exposures to O3‐enriched air result in a strong decrease in device performance, similar to exposure to ambient air for prolonged periods (48 h). The VOC, FF, and JSC (verified by EQE) suffer from O3 exposure and after 5 min the photovoltaic behavior is completely lost.2Figurea) J–V measurements recorded under simulated AM1.5G light of ITO | ZnO | PM6:Y6 | MoO3 | Ag solar cells with a pristine active layer (dark green) or with active layers exposed for different times to O3‐enriched air before completing the solar cell with a MoO3 | Ag top contact. b) Corresponding EQE spectra versus wavelength. c) Semilogarithmic plot of the corresponding sensitive EQE spectra versus photon energy. d) The difference in optical density of a PM6:Y6 layer on glass exposed to 15 min ambient atmosphere or up to 15 min of O3‐enriched air compared to the pristine film.The sensitive EQE spectra (Figure 2c) of these samples show a substantial increase in defect signal after only 30 s of exposure to O3‐enriched air. The result is similar to 48 h aging in ambient atmosphere as is shown in Figure S5a (Supporting Information). Exposure to O3 for up to 5 min results in only a small additional increase in defect signal, but does decrease the above‐bandgap EQE significantly, consistent with the changes in the J–V characteristics (Figure 2a). Because the measured defect response is the product of defect density and charge collection efficiency from photo‐excited defects, it is possible that the intensity of the defect signal remains similar for prolonged O3 exposure due to a decrease in collection efficiency and a concurrent increasing defect density. Overall, these experiments indicate that aging in ambient atmosphere and exposure to O3‐enriched air present the same effects to device performance and defect density (Figure S5a, Supporting Information).O3 is highly reactive to unsaturated carbon–carbon bonds forming an initial ozonide adduct followed by cleavage (ozonolysis) resulting in a variety of oxidation products.[55,56] Exposure of polythiophene to O3 enriched air (<1 ppm) results in the formation of carboxyl (−COOH) and sulfone (−SO2−) groups.[57] At higher concentrations, O3 fully bleaches the π‐π* absorption of conjugated polymers as a result of chemical oxidation to carbonyls, carboxylic acids, hydroperoxides, and sulfones (for polythiophenes).[27,58–61]The thermodynamically favorable oxidation by nitrous oxide (N2O) can yield similar reaction products as oxidation by O3. N2O, however, is much less reactive due to unfavorable kinetics.[62,63] To rule out oxidation of PM6:Y6 by N2O, we exposed a semi‐finished inverted PM6:Y6 device to 10% N2O in He for ≈30 min and recorded the J–V, and sensitive EQE (Figure S6, Supporting Information). We find that, even at these high concentrations, N2O had a negligible influence on the defects and device performance and thus we conclude that O3 is the leading cause of defect formation in PM6:Y6 in ambient air. These results imply that stable solar cells need to be encapsulated to protect them from ingress of O3, and additionally from H2O in case MoO3 is used as hole transport layer.To further study the oxidative degradation by O3, we recorded the ultraviolet‐visible‐near infrared (UV–vis–NIR) absorptance spectra of PM6:Y6 layers on glass before and after exposure to O3‐enriched air for up to 15 min (Figure 2d). As a control, we included the difference after 15 min exposure to ambient atmosphere. We find that exposure to O3‐enriched air causes a bleaching of the absorption of both PM6 (centered ≈600 nm) and Y6 (centered ≈800 nm), similar to the O3‐induced bleaching of the p‐p* absorption reported for poly(3‐hexylthiophene) and poly[2‐methoxy‐5‐(3″,7″‐dimethyloctyloxy)‐1,4‐phenylenevinylene].[27] We note that after 30 s the decrease in absorption was small while the sensitive EQE spectrum showed a large increase in defect response, which is consistent with an interface effect on the sensitive EQE. After 15 min of exposure to O3‐enriched air, the loss in absorbance (or optical density (OD)) is more substantial for PM6 (ΔOD = −0.21) than for Y6 (ΔOD = −0.056), while their original optical densities are similar (0.82 vs 076 as shown in Figure S7, Supporting Information). The circa 3.5 times higher susceptibility of PM6 than of Y6 to O3 is possibly related to the slightly higher (by 0.09 eV) HOMO energy of PM6 in combination with its lower crystallinity that enables easier diffusion of ozone.[48] The strong effect of crystallinity on the susceptibility to photostability was recently demonstrated for isomeric, amorphous and crystalline, non‐fullerene acceptors.[64] The higher susceptibility of PM6 to O3 raises the question whether the defects observed in sensitive EQE are also primarily related to PM6.Resolving the Defect ResponseTo differentiate between defects states originating from either PM6 or Y6, we compare devices with PM6:Y6, PM6, and Y6 active layers. In general, comparing sensitive EQE spectra of devices with different active layers and different layer thickness is hampered by interference effects that distort measured spectra.[44,45,65] By using semitransparent devices, the effects of interference can be reduced substantially. Hence, we fabricated semitransparent devices with an active layer of PM6:Y6, PM6, and Y6 in which we replaced the Ag top electrode by a sputtered ITO electrode and recorded their sensitive EQE spectra (Figure 3a). Compared to the opaque PM6:Y6 cell with an Ag top electrode (Figure 1c), the semitransparent cell with a sputtered ITO electrode yields a substantial increase in the defect response (Figure 3a), which we ascribe to the presence of reactive oxygen species during ITO sputtering. Apparently, these reactive oxygen species oxidize the PM6:Y6 blend similar to O3.3Figurea) Semilogarithmic plot of sensitive EQE spectra versus photon energy of semitransparent ITO | ZnO | active layer | MoO3 | ITO devices with an active layer of PM6:Y6 (purple), PM6 (dark red), or Y6 (green). b,c) Semilogarithmic plot of the DOS versus energy obtained from ER‐EIS for films of b) PM6 and c) Y6. Films are either pristine or exposed to O3 for 30 s or 2 min prior to the measurements. d) Semilogarithmic plot of sensitive EQE spectra versus photon energy of ITO | ZnO | PM6:Y6 | MoO3 | Ag cells (dark green), of ITO | PEDOT:PSS | PM6:Y6 | PFN‐Br | Ag cells (yellow), and of semitransparent ITO | PEDOT:PSS | PM6:Y6 | AZO | ITO cells (red).By comparing the EQE spectra of the semitransparent devices, the lowest‐energy defect in PM6:Y6 (≈0.7 eV onset) can be assigned to a defect originating from PM6. The Y6 device shows a defect onset near 1.0 eV and the same signal can also be discerned in the EQE spectrum of PM6:Y6 as a slight deflection. Comparing the defect response of the semitransparent devices to those of opaque devices shown in Figure 1 and Figure 2, we find that the defect response for the opaque devices extends to at least 0.8 eV and thus beyond the defect state present in Y6. For the opaque devices, the Urbach tail extends to almost 1.0 eV and overshadows possible Y6‐defect contributions. We conclude that the defect EQE observed for opaque devices is dominated by the contribution of PM6 but that an additional (small) defect contribution originating from Y6 cannot be fully excluded.We use ER‐EIS to investigate the changes in the DOS of PM6 and Y6 films after O3 exposure. ER‐EIS is able to provide information on (defect) states in the bandgap and relate these to the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energies. Details on the measurements can be found in the Experimental Section. The DOS of PM6 and Y6 determined via ER‐EIS are shown in Figure 3b,c, respectively. For pristine PM6, there are distinct onsets of the HOMO and LUMO close to −6.2 and −3.8 eV versus vacuum, respectively. The HOMO‐LUMO difference in ER‐EIS corresponds to the electrochemical bandgap, which is known to often be higher than the optical bandgap.[66,67] Further, a minor defect contribution (2 × 1017 eV−1 cm−3) appears near the HOMO of PM6 at ≈−5.6 eV. The plateau in the bandgap is likely related to leakage currents. Upon exposure to O3 the defect density near the HOMO is significantly increased, and after 2 min the defect density is enlarged by two orders of magnitude. The LUMO level also shifts after O3 exposure, but the electrochemical bandgap remains larger than the optical bandgap. For pristine Y6, the DOS is markedly different compared to PM6. The HOMO (−6.5 eV) and LUMO (−4.7 eV) of Y6 are peaked much sharper and no obvious defect signal can be discerned in the bandgap. Exposure to O3 for up to 2 min did not alter the DOS significantly, in contrast to the observations for PM6. The DOS of PM6 and Y6 determined via ER‐EIS for other exposure times is included in the Figure S8 (Supporting Information).For reference, ER‐EIS spectra were also recorded after exposure to ambient air (Figures S5b and S9, Supporting Information). We find that the spectra are extremely similar to those recorded after O3 exposure (Figure S5b, Supporting Information). Again the Y6 spectrum remains constant while the DOS of PM6 shows an increased defect density near the HOMO. This is further confirmation that O3 exposure and ambient aging amount to the same defect formation in PM6.Thus, we conclude from ER‐EIS that the density of defects present in pristine PM6 increases upon O3 exposure or aging in ambient air, while Y6 is much less affected. This corroborates the results from the sensitive EQE experiments where the defects in pristine layers also increase in ambient aged and O3 exposed layers and originate from PM6. Additionally, ER‐EIS indicates that the relevant defect is located near the HOMO of PM6.With defects predominantly stemming from PM6 and energetically located near the HOMO of PM6, one might expect the defect response to be altered when the polarity of the device is switched. Nevertheless, after comparing inverted (n–i–p) and regular (p–i–n) devices, we find that the device polarity does not influence the defect response (Figure 3d). Further, a regular device in opaque and semitransparent configuration shows a practically identical defect response, ruling out the effect of interference on the defect response (Figure 3d). The oxygen scavenging nature of Al‐doped ZnO (AZO) and the larger thickness of AZO (≈60 nm) compared to MoO3 (15 nm) might underly its protective capability against ITO sputter damage.Location of the DefectsInformation on the spatial location of defects observed in sensitive EQE spectra can be obtained by changing the optical interference in a semitransparent device by evaporating an optical spacer−mirror stack (MgF2 | Ag) on top of the ITO top contact, as recently shown for perovskite solar cells.[65] Because the device itself is not changed in this experiment, any change in the EQE upon including the optical spacer−mirror stack must stem from the altered optical interference. The optical interference can be accurately modeled using the transfer matrix method,[68] and by comparing the modeling results with measured EQE spectra the defect location can be assessed.Here we apply this method to semitransparent PM6:Y6 devices to uncover the location of the observed defects within the active layer. Sensitive EQE spectra were recorded for three inverted semitransparent PM6:Y6 devices before and after evaporation of 100, 126, or 185 nm MgF2, and 100 nm Ag. Figure 4a reveals significant changes in the defect region as a consequence of changing the interference via different optical spacers. In Figure 4b–d we plot the normalized ratios of the normalized EQE spectra versus wavelength of semitransparent cells with and without the optical spacer−mirror stack as solid lines that are offset vertically for clarity. As further detailed below, the dashed lines correspond to the results of the optical simulations and represent the ratio of the photon densities with and without the optical spacer−mirror stack (expressed as the ratio of the normalized modulus squared of the optical electric field, |E|2), experienced by the absorbing species, being the PM6:Y6 active layer or the defect.[44,65]4Figurea) Semilogarithmic plot of the normalized sensitive EQE spectra versus wavelength of inverted PM6:Y6 devices in a semitransparent configuration or with an optical spacer−mirror stack using either 100, 126, or 185 nm MgF2. b) The ratio of the sensitive EQEs of semitransparent cells with and without an optical spacer−mirror stack versus wavelength for the three MgF2 thicknesses (continuous lines). The dashed lines represent the optically‐simulated ratio of the |E|2 spectra for semitransparent cells with and without the optical spacer−mirror stack assuming absorption occurs throughout the active layer. The measured and simulated ratios are offset per MgF2 thickness and normalized for clarity. The defect fraction versus wavelength is included to visualize the transition between above‐bandgap (PM6:Y6) dominated EQE response and below‐bandgap (defect) dominated EQE response (thick black line). c) Same as b) assuming absorption near the back (PM6:Y6 | MoO3) interface. d) Same as b) assuming absorption near the front (ZnO | PM6:Y6) interface.For the optical modeling it is imperative to first define the defect region in the EQE spectra. To this end, we fit the absorption edge with an Urbach tail (Figure S10a, Supporting Information) and define the onset of the defect region as the wavelength where the experimental EQE starts to deviate from the Urbach tail. The defect contribution can then be expressed as a fraction of the total EQE response. For the inverted semitransparent PM6:Y6 devices, we find that the defect contribution dominates the EQE spectrum from ≈1200 nm onwards (Figure S10b, Supporting Information).To rationalize the observed changes in the EQE spectra we employ transfer‐matrix modelling, using the refractive index − extinction coefficient (n‐k) spectra and thickness of all layers involved in the device stack as input (Figure S11, Supporting Information). Then we average the calculated |E|2 either over the entire active layer (throughout), or for the first 10 nm near the back interface (back, PM6:Y6 |MoO3), or near the front interface (front, ZnO | PM6:Y6).Finally, we calculate the ratio of the |E|2 spectra for semitransparent devices without and with an optical spacer−mirror stack. In Figure 4b,c,d we compare the ratios of the normalized EQEs with the calculated |E|2 ratios. The defect fraction is included as a thick black line for reference. Comparing the measured EQE ratio with the calculated |E|2 ratios assuming charge generation throughout, we find very good agreement for wavelengths below 1000 nm corresponding to above‐bandgap absorption. More specifically, the minima and maxima are correctly predicted at their corresponding wavelengths (Figure 4b). This is to be expected because the above‐bandgap absorption and photocurrent generation are bulk phenomena.Beyond 1000 nm, the defect response starts to dominate the EQE spectrum and the agreement between measurement and simulation is lost when assuming absorption throughout the active layer. If, however, the absorption is assumed to be located near the back interface (Figure 4c) there is excellent agreement between measurement and simulation in the defect region (>1200 nm). Assuming absorption near the front interface (Figure 4d) does not give simulated ratios in agreement with the experimental ratios in any wavelength region. From these results, we therefore conclude that the defect response recorded in Figure 4a stems from the back interface.We repeated the same procedure for three regular architecture semitransparent devices employing AZO and a MgF2 thickness of 100, 230, or 380 nm (Figure S12, Supporting Information). Again, we find that the defect response must be originating from the back interface, being in this case the PM6:Y6 | AZO interface. The defects in PM6:Y6 are thus located near the back interface regardless the device polarity. This could point towards defective molecules or (reactive) contaminants accumulating near the solvent‐air interface during spin coating. Alternatively, defect formation could occur during the thermal evaporation or sputtering of the back contact. During these high energy processes (temperature/plasma), adsorbed species could react with the PM6:Y6 blend.ConclusionIn this work, we used sensitive EQE measurements to analyze the defects present in a layer of PM6:Y6 in an OPV architecture. By exposing the active layer to ambient atmosphere and H2O‐saturated compressed air, we find that a trace constituent in air increases the defect density and reduces device performance. O3 was identified as responsible for the defect formation because it yielded similar results in exposure experiments as ambient atmosphere. From UV–vis–NIR absorption spectroscopy, we find that defect formation involves reaction of O3 with the PM6:Y6 blend. To explain the observed (reversible) increase in defect response upon aging complete inverted‐configuration cells in H2O‐saturated compressed air, we postulate the change in work function of MoO3 to influence the band bending at the PM6:Y6 | MoO3 interface, resulting in an increased density of occupied defect states and consequently higher defect response. Semitransparent devices were fabricated to circumvent spectral distortions by interference effects and the sensitive EQE spectra of these devices showed the low energy defect to originate from PM6. The results from sensitive EQE are supported by ER‐EIS experiments on the DOS of PM6 and Y6 which indicate that O3 causes an increased defect signal near the HOMO of PM6 but did not change the DOS of Y6. Finally, by comparing the EQE ratios of semitransparent devices with and without an optical spacer−mirror stack added, we could attribute the defect response to an absorptive site located near the back interface regardless the n–i–p or p–i–n polarity of the cell architecture.Given the reactivity of O3 and the common design motives in organic semiconductors, it can be assumed that defect formation by O3 is common to many organic semiconductors. In general, during synthesis, purification, or processing π‐conjugated polymers and molecules are exposed to ambient air for shorter or longer time. Oxidation by trace O3 concentrations could be responsible for defects observed in organic semiconductor based devices.Experimental SectionMaterials2‐Methoxyethanol (99+%, Acros Organics), ethanolamine (>99.5%, Aldrich), zinc acetate dihydrate (98+%, Acros Organics), MoO3 (99.97%, Aldrich), 1‐chloronaphthalene (>85% TCI), chloroform, acetonitrile (>99.9%, extra dry over mol. sieves, Thermo Scientific), and TBAPF6 (>99.0%, electrochemical grade, Merck) were employed without further purification. PM6, Y6, and poly(9,9‐bis(3′‐(N,N‐dimethyl)‐N‐ethylammoinium‐propyl‐2,7‐fluorene)‐alt‐2,7‐(9,9‐dioctylfluorene))dibromide (PFN‐Br) were acquired from Solarmer Materials. AZO (N‐21X) and poly(3,4‐ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) (Clevios P, VP Al 4083) were obtained from Avantama and Heraeus, respectively. ITO substrates were obtained from Naranjo Substrates and the ITO sputtering target was obtained from Angstrom Engineering.Fabrication and AgingPre‐patterned and fully covered ITO‐covered glass substrates were rinsed by sonication in acetone, scrubbing with a soap solution in water, flushing with deionized water and sonication in isopropanol. Prior to spin coating the PEDOT:PSS, ZnO, or PM6:Y6 solution, the substrate was subjected to 30 min UV/O3 cleaning (UVP PR‐100). A ZnO colloidal solution was obtained by dissolving zinc acetate dihydrate (109.7 mg) in 2‐methoxyethanol (1 mL) and adding ethanolamine (30.2 µL). The ZnO solution was stirred for >1 h before use. For the inverted device architecture, the patterned ITO was spin coated with the ZnO solution at 4000 rpm for 60 s and subsequently annealed at 150 °C for 5 min. The PM6:Y6 (1:1.2) active layer was cast from chloroform containing 0.5 vol.‐% 1‐chloronaphthalene at a concentration of 16 mg mL−1. Spin coating proceeded at 3000 rpm for 30 s in nitrogen atmosphere followed by annealing at 110 °C for 10 min. Inverted devices were finalized by thermal evaporation of 15 nm MoO3 and 100 nm Ag.For regular devices, PEDOT:PSS was spin coated at 4000 rpm for 60 s on top of pre‐patterned ITO‐covered glass substrates followed by annealing at 150 °C for 5 min. After depositing the active layer following the procedure described above, PFN‐Br (0.5 mg mL−1 in methanol) was deposited on a rotating substrate at 3000 rpm in N2. Alternatively, AZO was spin coated at 2000 rpm for 60 s, followed by annealing at 75 °C for 1 min, another spin coating at 2000 rpm for 60 s and a final annealing at 75 °C for 5 min. The cells were completed by depositing Ag (100 nm) via thermal evaporation. For semitransparent devices, an ITO (180 nm) electrode was deposited by radiofrequency magnetron sputtering (Angstrom Engineering).PM6 (≈70 nm) and Y6 (≈120 nm) films for ER‐EIS were cast on fully ITO covered glass substrates from chloroform containing 0.5 vol.‐% 1‐chloronaphthalene at a concentration of 8 mg mL−1 and 16 mg mL−1 respectively. Spin coating of these films was done at 1200 rpm for 60 s in nitrogen atmosphere, followed by annealing at 110 °C for 10 min. The films were exposed to high vacuum (<5 × 10−7 mbar) overnight before exposure to different atmospheres and measurement with ER‐EIS.Samples for absorption measurements were produced by casting the PM6:Y6 active layer on a cleaned glass substrate following the procedure for the inverted/regular devices.Exposure to different atmospheres was performed in an airtight flow box with inlet and outlet. For other atmospheres than ambient, the flow box was kept at a slight overpressure relative to ambient pressure. Samples were kept in the dark during aging. Intentional O3 exposure was conducted by guiding the O3 enriched internal atmosphere of an UV/O3 photoreactor (UVP PR‐100) in operation over the semi‐finished devices or PM6:Y6 layers on (ITO covered) glass. During intentional O3 exposure, samples were kept in the dark.Compressed air was taken from a central facility consisting of a large tank, kept at constant pressure. The tank was refilled by ambient air that was filtered, dried, and passed over active coal to remove reactive trace gasses. The air was further dried by a H2O‐absorber upon exiting the tank and before entering the central distribution system. Following initial intake, the air was never exposed to light and could remain inside the tank and central facility for extended periods of time. Owing to the short half‐life of O3 (indicated as several tens of minutes to several hours)[69–71] and the filtering with active coal, the compressed air was virtually free of O3 at the point of use.CharacterizationThe current density–voltage (J–V) characterization was performed using a Keithley 2400 sourcemeter unit while the device was exposed to simulated AM1.5G illumination. The incident spectrum was achieved by using a Schott GG385 UV filter and a Hoya LB120 daylight filter in combination with a tungsten halogen lamp light source.EQE spectra were obtained by illuminating the device with chopped (Stanford Research Systems SR540) light of a tungsten‐halogen lamp that passed a monochromator (Oriel Cornerstone 130) and aperture. The generated current was measured by a low‐noise current preamplifier (Stanford Research Systems SR570) and a lock‐in amplifier (Stanford Research Systems SR830). The devices were simultaneously exposed to bias light (730 nm light emitting diode, Thorlabs M730L4) to better simulate operating conditions under AM1.5G illumination. The device response was quantified by comparing with a calibrated Si solar cell.Sensitive EQE in the sub‐bandgap region was measured using an Oriel 3502 light chopper, a Cornerstone 260 monochromator (CS260‐USB‐3‐MC‐A) with appropriate sorting filters, a Stanford Research SR 570 low‐noise current preamplifier, a Stanford Research SR830 lock‐in amplifier, and a 250 W tungsten‐halogen lamp. A detailed description of the set‐up and measurement can be found in Ref. [65].A PerkinElmer Lambda 1050 UV–vis‐NIR spectrophotometer was used to measure the absorption spectra. n–k spectra were obtained by variable angle spectroscopic ellipsometry using a WVASE32 ellipsometer (J.A. Woollam Co.). Thicknesses were also obtained from VASE and validated by profilometry (Veeco Dektak 150). Simulations were performed using an in‐house adapted version of the open‐source script provided by McGehee et al.[72]The energy resolved‐electrochemical impedance spectroscopy (ER‐EIS) method was adapted from the work of Nádaždy et al.[73] In a three‐electrode electrochemical cell, a glass | ITO working electrode was covered with either PM6 or Y6 of which 28.3 mm2 was exposed to the electrolyte (200 µL, 200 mm TBAPF6 in dry acetonitrile). The counter electrode was a platinum wire (0.5 mm, >99.99%) cleaned in a roaring blue flame. The pseudo‐reference electrode was a silver wire (0.5 mm, >99.9%) cleaned in a roaring blue flame and dipped in aqua regia before being rinsed with water and acetonitrile. The electrodes were connected to a Biologic VSP potentiostat with impedance capabilities and a “p” low‐current option. Before each measurement a voltaic hold of 20 s was applied with the potential set to the starting potential of the ER‐EIS measurement. The potential was increased/decreased with steps of 40 mV to measure the HOMO/LUMO of the organic films, starting at a point in the bandgap. The frequency was fixed at 0.5 Hz with an amplitude of 60 mV. At each potential step an average of six impedance measurements was taken and later converted to the DOS using the equations described by Schauer et al.[74] Measuring HOMO and LUMO was done in separate measurements on different parts of the same film.AcknowledgementsThe authors would like to thank Guus J. W. Aalbers for measuring the sensitive EQE spectrum of cells employing aged ZnO layers. The authors acknowledge funding from the Ministry of Education, Culture, and Science (Gravity program 024.001.035). The work is further part of the Advanced Research Center for Chemical Building Blocks, ARC CBBC, which is co‐founded and co‐financed by Netherlands Organisation for Scientific Research (NWO) and the Netherlands Ministry of Economic Affairs (project 2016.03.Tue). The authors acknowledge the financial support by NWO via a Spinoza grant.Conflict of InterestThe authors declare no conflict of interest.Data Availability StatementThe data that support the findings of this study are available from the corresponding author upon reasonable request.J. M. Shaw, P. F. Seidler, IBM J. Res. Dev. 2001, 45, 3.R. M. Owens, G. G. Malliaras, MRS Bull. 2010, 35, 449.Y. van de Burgt, A. Melianas, S. T. Keene, G. Malliaras, A. Salleo, Nat. Electron. 2018, 1, 386.J. Rivnay, S. Inal, A. Salleo, R. M. Owens, M. Berggren, G. G. Malliaras, Nat. Rev. Mater. 2018, 3, 17086.J. T. Mabeck, G. G. Malliaras, Anal. Bioanal. Chem. 2006, 384, 343.Y. van de Burgt, E. Lubberman, E. J. Fuller, S. T. Keene, G. C. Faria, S. Agarwal, M. J. Marinella, A. Alec Talin, A. Salleo, Nat. Mater. 2017, 16, 414.P. Cheng, G. Li, X. Zhan, Y. Yang, Nat. Photonics 2018, 12, 131.Z. Hu, J. Wang, X. Ma, J. Gao, C. Xu, K. Yang, Z. Wang, J. Zhang, F. Zhang, Nano Energy 2020, 78, 105376.G. Horowitz, Adv. Mater. 1998, 10, 365.G. Simone, M. J. Dyson, S. C. J. Meskers, R. A. J. Janssen, G. H. Gelinck, Adv. Funct. Mater. 2020, 30, 1904205.J. Bauri, R. B. Choudhary, G. Mandal, J. Mater. Sci. 2021, 56, 18837.A. Salehi, X. Fu, D. Shin, F. So, Adv. Funct. Mater. 2019, 29, 1808803.H. F. Haneef, A. M. Zeidell, O. D. Jurchescu, J. Mater. Chem. C 2020, 8, 759.S. Zeiske, O. J. Sandberg, N. Zarrabi, W. Li, P. Meredith, A. Armin, Nat. Commun. 2021, 12, 3603.P. H. Nguyen, S. Scheinert, S. Berleb, W. Brüiting, G. Paasch, Org. Electron. 2001, 2, 105.C. Krellner, S. Haas, C. Goldmann, K. P. Pernstich, D. J. Gundlach, B. Batlogg, Phys. Rev. B 2007, 75, 245115.M. Grünewald, P. Thomas, D. Würtz, Phys. Status Solidi 1980, 100, K139.W. L. Kalb, F. Meier, K. Mattenberger, B. Batlogg, Phys. Rev. B 2007, 76, 184112.M. Meier, S. Karg, K. Zuleeg, W. Brütting, M. Schwoerer, J. Appl. Phys. 1998, 84, 87.D. V. Lang, J. Appl. Phys. 1974, 45, 3023.J. J. Dittmer, E. A. Marseglia, R. H. Friend, Adv. Mater. 2000, 12, 1270.D. Hu, J. Yu, G. Padmanaban, S. Ramakrishnan, P. F. Barbara, Nano Lett. 2002, 2, 1121.M. P. Nikiforov, B. Lai, W. Chen, S. Chen, R. D. Schaller, J. Strzalka, J. Maser, S. B. Darling, Energy Environ. Sci. 2013, 6, 1513.K. Müllen, Nat. Rev. Mater. 2016, 1, 15013.K. H. Hendriks, W. Li, G. H. L. Heintges, G. W. P. van Pruissen, M. M. Wienk, R. A. J. Janssen, J. Am. Chem. Soc. 2014, 136, 11128.M. Streiter, D. Beer, F. Meier, C. Göhler, C. Lienert, F. Lombeck, M. Sommer, C. Deibel, Adv. Funct. Mater. 2019, 29, 1903936.A. Früh, H. J. Egelhaaf, H. Hintz, D. Quinones, C. J. Brabec, H. Peisert, T. Chassé, J. Mater. Res. 2018, 33, 1891.E. J. Meijer, C. Detcheverry, P. J. Baesjou, E. van Veenendaal, D. M. De Leeuw, T. M. Klapwijk, J. Appl. Phys. 2003, 93, 4831.M. L. Chabinyc, R. A. Street, J. E. Northrup, Appl. Phys. Lett. 2007, 90, 123508.H. F. Iqbal, Q. Ai, K. J. Thorley, H. Chen, I. McCulloch, C. Risko, J. E. Anthony, O. D. Jurchescu, Nat. Commun. 2021, 12, 2352.N. B. Kotadiya, A. Mondal, P. W. M. Blom, D. Andrienko, G. J. A. H. Wetzelaer, Nat. Mater. 2019, 18, 1182.J.‐M. Zhuo, L.‐H. Zhao, R.‐Q. Png, L.‐Y. Wong, P.‐J. Chia, J.‐C. Tang, S. Sivaramakrishnan, M. Zhou, E. C. W. Ou, S.‐J. Chua, W.‐S. Sim, L.‐L. Chua, P. K. H. Ho, Adv. Mater. 2009, 21, 4747.P. K. Nayak, R. Rosenberg, L. Barnea‐Nehoshtan, D. Cahen, Org. Electron. 2013, 14, 966.S. N. Yaliraki, R. J. Silbey, J. Chem. Phys. 1996, 104, 1245.T. Huser, M. Yan, L. J. Rothberg, Proc Natl Acad Sci U S A 2000, 97, 11187.Y.‐Y. Noh, D.‐Y. Kim, Y. Yoshida, K. Yase, B.‐J. Jung, E. Lim, H.‐K. Shim, R. Azumi, J. Appl. Phys. 2005, 97, 104504.A. Seemann, T. Sauermann, C. Lungenschmied, O. Armbruster, S. Bauer, H. J. Egelhaaf, J. Hauch, Sol. Energy 2011, 85, 1238.J. Schafferhans, A. Baumann, A. Wagenpfahl, C. Deibel, V. Dyakonov, Org. Electron. 2010, 11, 1693.H. T. Nicolai, M. Kuik, G. A. H. Wetzelaer, B. de Boer, C. Campbell, C. Risko, J. L. Brédas, P. W. M. Blom, Nat. Mater. 2012, 11, 882.M. Nikolka, I. Nasrallah, B. Rose, M. K. Ravva, K. Broch, A. Sadhanala, D. Harkin, J. Charmet, M. Hurhangee, A. Brown, S. Illig, P. Too, J. Jongman, I. McCulloch, J. L. Bredas, H. Sirringhaus, Nat. Mater. 2017, 16, 356.G. Zuo, M. Linares, T. Upreti, M. Kemerink, Nat. Mater. 2019, 18, 588.H. F. Iqbal, M. Waldrip, H. Chen, I. McCulloch, O. D. Jurchescu, Adv. Electron. Mater. 2021, 7, 2100393.S. Zeiske, C. Kaiser, P. Meredith, A. Armin, ACS Photonics 2019, 7, 256.C. Kaiser, S. Zeiske, P. Meredith, A. Armin, Adv. Opt. Mater. 2020, 8, 1901542.J. Melskens, M. Schouten, R. Santbergen, M. Fischer, R. Vasudevan, D. J. Van Der Vlies, R. J. V. Quax, S. G. M. Heirman, K. Jäger, V. Demontis, M. Zeman, A. H. M. Smets, Sol. Energy Mater. Sol. Cells 2014, 129, 70.G. Li, C.‐W. Chu, V. Shrotriya, J. Huang, Y. Yang, Appl. Phys. Lett. 2006, 88, 253503.J. M. Wallace, P. V. Hobbs, in Atmos. Sci., Elsevier, New York, 2006, pp. 153‐207.J. Yuan, Y. Zhang, L. Zhou, G. Zhang, H. L. Yip, T. K. Lau, X. Lu, C. Zhu, H. Peng, P. A. Johnson, M. Leclerc, Y. Cao, J. Ulanski, Y. Li, Y. Zou, Joule 2019, 3, 1140.J. Yuan, T. Huang, P. Cheng, Y. Zou, H. Zhang, J. L. Yang, S.‐Y. Chang, Z. Zhang, W. Huang, R. Wang, D. Meng, F. Gao, Y. Yang, Nat. Commun. 2019, 10, 570.F. Urbach, Phys. Rev. 1953, 92, 1324.D. Y. Kim, J. Subbiah, G. Sarasqueta, F. So, H. Ding, Y. G. Irfan, Appl. Phys. Lett. 2009, 95, 093304.Irfan, H. D., Y. Gao, C. Small, D. Y. Kim, J. Subbiah, F. So, Appl. Phys. Lett. 2010, 96, 243307.M. C. Gwinner, R. Di Pietro, Y. Vaynzof, K. J. Greenberg, P. K. H. Ho, R. H. Friend, H. Sirringhaus, Adv. Funct. Mater. 2011, 21, 1432.M. Shi, T. Wang, R. Sun, Q. Wu, D. Pei, H. Wang, W. Yang, W. Wang, Y. Wu, G. Xie, T. Wang, L. Ye, J. Min, Sci. China Mater. 2021, 64, 2629.P. S. Bailey, Chem. Rev. 1958, 58, 925.R. W. Murray, Acc. Chem. Res. 1968, 1, 313.J. Heeg, C. Kramer, M. Wolter, S. Michaelis, W. Plieth, W.‐J. Fischer, Appl. Surf. Sci. 2001, 180, 36.P. S. Bailey, H. H. Hwang, J. Org. Chem. 1985, 50, 1778.J. Nowaczyk, W. Czerwiński, E. Olewnik, Polym. Degrad. Stab. 2006, 91, 2022.J. Nowaczyk, P. Olszowy, P. Cysewski, A. Nowaczyk, W. Czerwiński, Polym. Degrad. Stab. 2008, 93, 1275.H. Hintz, H.‐J. Egelhaaf, H. Peisert, T. Chassé, Polym. Degrad. Stab. 2010, 95, 818.F. S. Bridson‐Jones, G. D. Buckley, L. H. Cross, A. P. Driver, J Chem Soc 1951, 2999.A. V. Leont'ev, O. A. Fomicheva, M. V. Proskurnina, N. S. Zefirov, Russ. Chem. Rev. 2001, 70, 91.J. Guo, Y. Wu, R. Sun, W. Wang, J. Li, E. Zhou, J. Guo, T. Wang, Q. Wu, Z. Luo, W. Gao, Y. Pan, C. Yang, J. Min, Sol. RRL 2021, 5, 2000704.B. T. van Gorkom, T. P. A. van der Pol, K. Datta, M. M. Wienk, R. A. J. Janssen, Nat. Commun. 2022, 13, 349.R. E. M. Willems, C. H. L. Weijtens, X. de Vries, R. Coehoorn, R. A. J. Janssen, Adv. Energy Mater. 2019, 9, 1803677.S. Admassie, O. Inganäs, W. Mammo, E. Perzon, M. R. Andersson, Synth. Met. 2006, 156, 614.L. A. A. Pettersson, L. S. Roman, O. Inganäs, J. Appl. Phys. 1999, 86, 487.C. J. Weschler, Indoor Air 2000, 10, 269.J. G. Waller, G. McTurk, J. Appl. Chem. 1965, 15, 363.R. Pandiselvam, S. Sunoj, M. R. Manikantan, A. Kothakota, K. B. Hebbar, Ozone: Sci. Eng. 2017, 39, 115.G. F. Burkhard, E. T. Hoke, M. D. McGehee, Adv. Mater. 2010, 22, 3293.V. Nádaždy, F. Schauer, K. Gmucová, Appl. Phys. Lett. 2014, 105, 142109.F. Schauer, J. Appl. Phys. 2020, 128, 150902. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Advanced Energy Materials Wiley

Origin, Nature, and Location of Defects in PM6:Y6 Organic Solar Cells

Loading next page...
 
/lp/wiley/origin-nature-and-location-of-defects-in-pm6-y6-organic-solar-cells-14JQMO0kGU

References (138)

Publisher
Wiley
Copyright
© 2023 Wiley‐VCH GmbH
ISSN
1614-6832
eISSN
1614-6840
DOI
10.1002/aenm.202300003
Publisher site
See Article on Publisher Site

Abstract

IntroductionOrganic semiconductors can provide added functionality compared to conventional semiconductors in electronic applications owing to their lightweight, biocompatible, and flexible nature combined with easy large‐scale processing at room temperature and ambient pressures.[1–3] A large variety of devices have emerged ranging from organic field‐effect transistors (OFETs), organic electrochemical transistors, organic memristors, organic light‐emitting diodes (OLEDs) to organic photodetectors and organic photovoltaics (OPVs).[4–12] Despite intense research interest and potential advantages, applications on large scale remain primarily limited to OLEDs.To further mature technologies based on organic semiconductors, limiting factors should be identified and eliminated. A universal obstacle for organic semiconductors is the presence of defects, acting as a source of charge‐trapping and of (non‐radiative) recombination.[13–15] Knowledge on defect formation is vital to formulate a targeted approach to lower their number and impact. Given their scarce nature, however, defects are notoriously difficult to characterize and their occurrence and energy levels are often deduced from device measurements combined with physical simulation models.[16–20]The origin of defects is a longstanding debate in the field. Defect formation has been tentatively ascribed to (synthetic) contaminants, conformational changes, or molecular defects resulting from imperfect synthesis, or by the influence of ambient atmosphere sometimes accelerated by ultraviolet (UV) or visible light.[21,22,31–35,23–30] Particularly for OFETs, it has been demonstrated that ambient air with or without light has a direct influence on the defect density.[28–30,36] Oftentimes, oxygen or ozone are identified as the culprit in defect formation.[27,37,38] However, recently, the introduction of an iso‐energetic defect by water‐containing nanovoids was proposed.[39–42] In general, the role of the different constituents of ambient air (e.g., oxygen, ozone, and water) remains elusive.In this work, we use highly sensitive external quantum efficiency (EQE) measurements to directly record the photogeneration of charges by defects in an OPV device architecture. It has been shown that EQE measurements up to 100 dB dynamic range are feasible by carefully eliminating noise in the measurement.[43] With this technique it is possible to measure very small photocurrents that originate from excitation of charges trapped in defect states.[44,45] The magnitude of the EQE response is proportional to the defect density, and by following the response we can thus track the defect concentration.Here, we investigate the origin and nature of defect states in PM6:Y6 blends (PM6 is poly[(2,6‐(4,8‐bis(5‐(2‐ethylhexyl‐3‐fluoro)thiophen‐2‐yl)‐benzo[1,2‐b:4,5‐b′]dithiophene))‐alt‐(5,5‐(1′,3′‐di‐2‐thienyl‐5′,7′‐bis(2‐ethylhexyl)benzo[1′,2′‐c:4′,5′‐c′]dithiophene‐4,8‐dione)] and Y6 is 2,2′‐((2Z,2′Z)‐((12,13‐bis(2‐ethylhexyl)‐3,9‐diundecyl‐12,13‐dihydro‐[1,2,5]thiadiazolo[3,4‐e]thieno‐[2″,3″:4′,5′]thieno[2′,3′:4,5]pyrrolo[3,2‐g]thieno‐[2′,3′:4,5]thieno[3,2‐b]indole‐2,10‐diyl)bis(methanylylidene))‐bis(5,6‐difluoro‐3‐oxo‐2,3‐dihydro‐1H‐indene‐2,1‐diylidene))dimalononitril). Sensitive EQE spectroscopy is used to study the change in defect response upon exposing either the active layer or complete devices to different components that are present in ambient air while keeping the sample in the dark. Rather than oxygen and water, we find that trace compounds in ambient air cause a distinct increase in defect response coupled to a decrease in device performance. By controlled exposure to O3 and N2O, we determine that oxidation by O3 is the source of defect formation in PM6:Y6 blends when exposed to ambient air. By studying semitransparent devices comprising either PM6:Y6, PM6, or Y6 active layers in inverted (n–i–p) and regular (p–i–n) cell architectures, we show that the defect response is dominated by PM6 and is indifferent to the device polarity. Additionally, we perform energy resolved‐electrochemical impedance spectroscopy (ER‐EIS) on films of PM6 and Y6 to determine the density of states (DOS). For PM6 we record the increase of a sub‐bandgap defect state upon O3 exposure and aging in ambient air. In contrast, the DOS of Y6 remains unaltered. By comparing optical simulations with the ratio of experimental EQE spectra of semitransparent cells with and without an additional optical spacer−mirror stack we determine that these oxidation‐induced defects are located at the back interface.Results and DiscussionVarious AtmospheresWe fabricated PM6:Y6 organic solar cells with an inverted (n–i–p) configuration (glass | indium tin oxide (ITO) | ZnO | PM6:Y6 | MoO3 | Ag) under inert conditions.[46] After spin coating the PM6:Y6 layer in a N2‐filled glove box, the semi‐finished solar cells were subjected to ambient atmosphere, to purified compressed air saturated with H2O for ≈48 h in the dark, or kept under N2 as a reference. Compressed air consists of N2 (≈78%), O2 (≈21%), Ar (≈1%), negligible amounts of H2O before saturating with H2O by bubbling through water, and is free of O3 (see Experimental Section for details). Ambient atmosphere contains potentially reactive trace gasses such as CH4, H2, N2O, CO, and O3 (listed most‐to‐less‐abundant).[47] After the optional exposure of the active layer to ambient air or to H2O‐saturated compressed air, the solar cells were completed by thermal evaporation of MoO3 and Ag as top contact in high vacuum.The current density – voltage (J–V) characteristics of these cells are presented in Figure 1a. As expected, pristine PM6:Y6 cells provide high efficiency,[48,49] and a similar high performance is found after aging the PM6:Y6 layer in H2O‐saturated compressed air prior to evaporating the top contact. Exposure of the PM6:Y6 layer to ambient atmosphere before completing the device, however, resulted in a distinct decrease in device performance with the short‐circuit current density (JSC), open‐circuit voltage (VOC), and fill factor (FF) being affected. The corresponding EQE spectra confirm the trend, with similar spectra for pristine films and films exposed to H2O‐saturated compressed air, while films exposed to ambient atmosphere resulted in a significantly lower EQE (Figure 1b).1Figurea) J–V measurements recorded with simulated AM1.5G light of ITO | ZnO | PM6:Y6 | MoO3 | Ag solar cells with a pristine active layer (red) or with active layers aged for ≈48 h in H2O‐saturated compressed air (green) or in ambient atmosphere (blue) before completing the solar cell with a MoO3 | Ag top contact. b) Corresponding EQE spectra versus wavelength. c) Semilogarithmic plot of the corresponding sensitive EQE spectra versus photon energy.Figure 1c shows a semilogarithmic plot of the sensitive EQE spectra of these cells versus photon energy. Focusing on the pristine cell, the above‐bandgap (>1.4 eV) EQE transitions to an Urbach tail region, characterized by an exponentially decreasing EQE from the bandgap at ≈1.4 eV until ≈1.05 eV with an Urbach energy of 23 meV.[50] Below 1.05 eV the onset of a defect response is visible, as a deviation from the exponential behavior, just before the spectrum becomes noise limited. Comparing the pristine sample with the sample aged in H2O‐saturated compressed air we find that the sensitive EQEs are identical. For layers exposed to ambient atmosphere, however, the EQE is lowered above the bandgap and also reveals an increase of the defect response by approximately one order of magnitude at photon energies less than 1.15 eV (Figure 1c). Thus, through a direct measurement of the defect response, we infer that exposure of the ITO | ZnO | PM6:Y6 stack to ambient atmosphere gives rise to an increase in defects. Further, since H2O‐saturated compressed air does not give a detectable change in the sensitive EQE spectrum, the cause must be a trace constituent of ambient atmosphere that is not present in H2O‐saturated compressed air. We also note that the defects observed in the sensitive EQE spectra correlate with a loss in performance of the solar cell. Aging the ZnO layer before PM6:Y6 deposition does not lead to a measurable change in device performance or defect response (Figure S1, Supporting Information). Therefore the reduced solar cell performance and increased defect response shown in Figure 1 are ascribed to the PM6:Y6 active layer.The significant losses in JSC, VOC, and FF for films exposed to ambient air are ascribed to extensive electron‐hole recombination primarily caused by hindered charge transport, rather than poor charge generation because at −2 V the photocurrent of the layer exposed to ambient air is very similar to that of the other two samples (Figure S2, Supporting Information). Hence, a reverse electric field allows charges, that otherwise recombine at short‐circuit, to be extracted. The fact that EQE is reduced over the entire spectral range is consistent with this explanation.We then studied the impact of exposure of a complete solar cell to different atmospheres and report the J–V characteristics and sensitive EQE spectra in Figure S3a,b (Supporting Information). For a complete device, H2O‐saturated compressed air does decrease device performance and increase the defect EQE, while dry compressed air does not. The top electrode (MoO3 | Ag) in conjunction with H2O must underlie these observations as the defect density of the ITO | ZnO | PM6:Y6 stack was not affected under these conditions. It is known that when MoO3 is exposed to H2O, the band bending that enables a selective hole contact vanishes as the work function of MoO3 is reduced by H2O.[51–53] This effect is commonly ascribed to the adsorption of H2O.[52,53] We postulate that for the pristine cell the band bending decreased the defect response by lowering the filled‐defect density near the interface. A schematic diagram is included in Figure S4 (Supporting Information). Upon exposure of MoO3 to H2O, the band bending is reduced resulting in more frequent occupation of the defect states and thus more states available for photogeneration of charges. Implicit to the proposed mechanism is the charge generation by defects occurring through the excitation of filled‐defect states.The change in energy levels of MoO3 due to H2O adsorption are reported to be reversible.[53] Indeed, we find that upon exposure to vacuum (≈5 × 10−7 mbar) or prolonged aging in N2 atmosphere, the defect response in sensitive EQE is again similar to the pristine cell (Figure S3c,d, Supporting Information). We conclude that the H2O adsorbed to MoO3 can be removed in vacuum, thereby restoring the band bending at the MoO3 interface and yielding a defect response comparable to a pristine cell (Figure S3c, Supporting Information). Similarly, prolonged storage in dry N2 also removes adsorbed H2O (Figure S3d, Supporting Information). In contrast, for an air‐exposed complete solar cell the defect response increases further rather than decreases over time in N2 atmosphere (Figure S3d, Supporting Information) and the increase therefore has a different underlying cause for aging in ambient atmosphere. These results strengthen the proof that the decreased band bending due to H2O‐exposed MoO3 underlies the observed increase in defect response for these devices exposed to H2O‐saturated compressed air as more filled defect states become available for photo‐excitation. Concurrently, since band bending occurs primarily near the PM6:Y6 | MoO3 interface, the defect contribution could be dominated by this interface.The increased, and reversible, defect response upon H2O exposure of finished cells closely resembles the defect response upon ambient air aging (Figure S3b, Supporting Information). Therefore, we attribute both defect responses as coming from the same defect. Additionally, this provides evidence that the defects present in pristine PM6:Y6 cells are the same as those formed upon exposure to ambient air.The reversible increase in defect response of a finished device upon aging in H2O‐saturated compressed air could alternatively be explained by the presence of water‐induced defects, as often discussed in literature.[41,54] The reversibility of this process would then be ascribed to H2O leaving the active layer. However, the same defect EQE is observed when aging a finished cell in ambient atmosphere, but in that case the effect is irreversible. Therefore these defects cannot simply be ascribed to the uptake of water because this mechanism should be reversible (or irreversible) regardless the aging in ambient air or H2O‐saturated compressed air. In fact, the results shown in Figure 1c give no indication for the presence of water‐induced defects. Of course, it is possible that a change in water‐induced defect‐density is not recorded in the EQE when these defects do not generate a photocurrent when excited.Trace Constituents of Ambient AirHaving found that a trace constituent of ambient air likely causes the formation of defects in PM6:Y6 films, the question arises which compound is responsible for this effect. Considering the various trace compounds in ambient air, ozone (O3) is known for its reactivity toward organic substrates,[47] especially in oxidizing unsaturated bonds. To investigate its effect, we exposed semi‐finished inverted PM6:Y6 devices to O3‐enriched air for times between 30 s and 5 min (see Experimental Section for details), followed by completing the device with a MoO3 | Ag top contact. The J–V and EQE characteristics (Figure 2a,b) reveal that such short exposures to O3‐enriched air result in a strong decrease in device performance, similar to exposure to ambient air for prolonged periods (48 h). The VOC, FF, and JSC (verified by EQE) suffer from O3 exposure and after 5 min the photovoltaic behavior is completely lost.2Figurea) J–V measurements recorded under simulated AM1.5G light of ITO | ZnO | PM6:Y6 | MoO3 | Ag solar cells with a pristine active layer (dark green) or with active layers exposed for different times to O3‐enriched air before completing the solar cell with a MoO3 | Ag top contact. b) Corresponding EQE spectra versus wavelength. c) Semilogarithmic plot of the corresponding sensitive EQE spectra versus photon energy. d) The difference in optical density of a PM6:Y6 layer on glass exposed to 15 min ambient atmosphere or up to 15 min of O3‐enriched air compared to the pristine film.The sensitive EQE spectra (Figure 2c) of these samples show a substantial increase in defect signal after only 30 s of exposure to O3‐enriched air. The result is similar to 48 h aging in ambient atmosphere as is shown in Figure S5a (Supporting Information). Exposure to O3 for up to 5 min results in only a small additional increase in defect signal, but does decrease the above‐bandgap EQE significantly, consistent with the changes in the J–V characteristics (Figure 2a). Because the measured defect response is the product of defect density and charge collection efficiency from photo‐excited defects, it is possible that the intensity of the defect signal remains similar for prolonged O3 exposure due to a decrease in collection efficiency and a concurrent increasing defect density. Overall, these experiments indicate that aging in ambient atmosphere and exposure to O3‐enriched air present the same effects to device performance and defect density (Figure S5a, Supporting Information).O3 is highly reactive to unsaturated carbon–carbon bonds forming an initial ozonide adduct followed by cleavage (ozonolysis) resulting in a variety of oxidation products.[55,56] Exposure of polythiophene to O3 enriched air (<1 ppm) results in the formation of carboxyl (−COOH) and sulfone (−SO2−) groups.[57] At higher concentrations, O3 fully bleaches the π‐π* absorption of conjugated polymers as a result of chemical oxidation to carbonyls, carboxylic acids, hydroperoxides, and sulfones (for polythiophenes).[27,58–61]The thermodynamically favorable oxidation by nitrous oxide (N2O) can yield similar reaction products as oxidation by O3. N2O, however, is much less reactive due to unfavorable kinetics.[62,63] To rule out oxidation of PM6:Y6 by N2O, we exposed a semi‐finished inverted PM6:Y6 device to 10% N2O in He for ≈30 min and recorded the J–V, and sensitive EQE (Figure S6, Supporting Information). We find that, even at these high concentrations, N2O had a negligible influence on the defects and device performance and thus we conclude that O3 is the leading cause of defect formation in PM6:Y6 in ambient air. These results imply that stable solar cells need to be encapsulated to protect them from ingress of O3, and additionally from H2O in case MoO3 is used as hole transport layer.To further study the oxidative degradation by O3, we recorded the ultraviolet‐visible‐near infrared (UV–vis–NIR) absorptance spectra of PM6:Y6 layers on glass before and after exposure to O3‐enriched air for up to 15 min (Figure 2d). As a control, we included the difference after 15 min exposure to ambient atmosphere. We find that exposure to O3‐enriched air causes a bleaching of the absorption of both PM6 (centered ≈600 nm) and Y6 (centered ≈800 nm), similar to the O3‐induced bleaching of the p‐p* absorption reported for poly(3‐hexylthiophene) and poly[2‐methoxy‐5‐(3″,7″‐dimethyloctyloxy)‐1,4‐phenylenevinylene].[27] We note that after 30 s the decrease in absorption was small while the sensitive EQE spectrum showed a large increase in defect response, which is consistent with an interface effect on the sensitive EQE. After 15 min of exposure to O3‐enriched air, the loss in absorbance (or optical density (OD)) is more substantial for PM6 (ΔOD = −0.21) than for Y6 (ΔOD = −0.056), while their original optical densities are similar (0.82 vs 076 as shown in Figure S7, Supporting Information). The circa 3.5 times higher susceptibility of PM6 than of Y6 to O3 is possibly related to the slightly higher (by 0.09 eV) HOMO energy of PM6 in combination with its lower crystallinity that enables easier diffusion of ozone.[48] The strong effect of crystallinity on the susceptibility to photostability was recently demonstrated for isomeric, amorphous and crystalline, non‐fullerene acceptors.[64] The higher susceptibility of PM6 to O3 raises the question whether the defects observed in sensitive EQE are also primarily related to PM6.Resolving the Defect ResponseTo differentiate between defects states originating from either PM6 or Y6, we compare devices with PM6:Y6, PM6, and Y6 active layers. In general, comparing sensitive EQE spectra of devices with different active layers and different layer thickness is hampered by interference effects that distort measured spectra.[44,45,65] By using semitransparent devices, the effects of interference can be reduced substantially. Hence, we fabricated semitransparent devices with an active layer of PM6:Y6, PM6, and Y6 in which we replaced the Ag top electrode by a sputtered ITO electrode and recorded their sensitive EQE spectra (Figure 3a). Compared to the opaque PM6:Y6 cell with an Ag top electrode (Figure 1c), the semitransparent cell with a sputtered ITO electrode yields a substantial increase in the defect response (Figure 3a), which we ascribe to the presence of reactive oxygen species during ITO sputtering. Apparently, these reactive oxygen species oxidize the PM6:Y6 blend similar to O3.3Figurea) Semilogarithmic plot of sensitive EQE spectra versus photon energy of semitransparent ITO | ZnO | active layer | MoO3 | ITO devices with an active layer of PM6:Y6 (purple), PM6 (dark red), or Y6 (green). b,c) Semilogarithmic plot of the DOS versus energy obtained from ER‐EIS for films of b) PM6 and c) Y6. Films are either pristine or exposed to O3 for 30 s or 2 min prior to the measurements. d) Semilogarithmic plot of sensitive EQE spectra versus photon energy of ITO | ZnO | PM6:Y6 | MoO3 | Ag cells (dark green), of ITO | PEDOT:PSS | PM6:Y6 | PFN‐Br | Ag cells (yellow), and of semitransparent ITO | PEDOT:PSS | PM6:Y6 | AZO | ITO cells (red).By comparing the EQE spectra of the semitransparent devices, the lowest‐energy defect in PM6:Y6 (≈0.7 eV onset) can be assigned to a defect originating from PM6. The Y6 device shows a defect onset near 1.0 eV and the same signal can also be discerned in the EQE spectrum of PM6:Y6 as a slight deflection. Comparing the defect response of the semitransparent devices to those of opaque devices shown in Figure 1 and Figure 2, we find that the defect response for the opaque devices extends to at least 0.8 eV and thus beyond the defect state present in Y6. For the opaque devices, the Urbach tail extends to almost 1.0 eV and overshadows possible Y6‐defect contributions. We conclude that the defect EQE observed for opaque devices is dominated by the contribution of PM6 but that an additional (small) defect contribution originating from Y6 cannot be fully excluded.We use ER‐EIS to investigate the changes in the DOS of PM6 and Y6 films after O3 exposure. ER‐EIS is able to provide information on (defect) states in the bandgap and relate these to the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energies. Details on the measurements can be found in the Experimental Section. The DOS of PM6 and Y6 determined via ER‐EIS are shown in Figure 3b,c, respectively. For pristine PM6, there are distinct onsets of the HOMO and LUMO close to −6.2 and −3.8 eV versus vacuum, respectively. The HOMO‐LUMO difference in ER‐EIS corresponds to the electrochemical bandgap, which is known to often be higher than the optical bandgap.[66,67] Further, a minor defect contribution (2 × 1017 eV−1 cm−3) appears near the HOMO of PM6 at ≈−5.6 eV. The plateau in the bandgap is likely related to leakage currents. Upon exposure to O3 the defect density near the HOMO is significantly increased, and after 2 min the defect density is enlarged by two orders of magnitude. The LUMO level also shifts after O3 exposure, but the electrochemical bandgap remains larger than the optical bandgap. For pristine Y6, the DOS is markedly different compared to PM6. The HOMO (−6.5 eV) and LUMO (−4.7 eV) of Y6 are peaked much sharper and no obvious defect signal can be discerned in the bandgap. Exposure to O3 for up to 2 min did not alter the DOS significantly, in contrast to the observations for PM6. The DOS of PM6 and Y6 determined via ER‐EIS for other exposure times is included in the Figure S8 (Supporting Information).For reference, ER‐EIS spectra were also recorded after exposure to ambient air (Figures S5b and S9, Supporting Information). We find that the spectra are extremely similar to those recorded after O3 exposure (Figure S5b, Supporting Information). Again the Y6 spectrum remains constant while the DOS of PM6 shows an increased defect density near the HOMO. This is further confirmation that O3 exposure and ambient aging amount to the same defect formation in PM6.Thus, we conclude from ER‐EIS that the density of defects present in pristine PM6 increases upon O3 exposure or aging in ambient air, while Y6 is much less affected. This corroborates the results from the sensitive EQE experiments where the defects in pristine layers also increase in ambient aged and O3 exposed layers and originate from PM6. Additionally, ER‐EIS indicates that the relevant defect is located near the HOMO of PM6.With defects predominantly stemming from PM6 and energetically located near the HOMO of PM6, one might expect the defect response to be altered when the polarity of the device is switched. Nevertheless, after comparing inverted (n–i–p) and regular (p–i–n) devices, we find that the device polarity does not influence the defect response (Figure 3d). Further, a regular device in opaque and semitransparent configuration shows a practically identical defect response, ruling out the effect of interference on the defect response (Figure 3d). The oxygen scavenging nature of Al‐doped ZnO (AZO) and the larger thickness of AZO (≈60 nm) compared to MoO3 (15 nm) might underly its protective capability against ITO sputter damage.Location of the DefectsInformation on the spatial location of defects observed in sensitive EQE spectra can be obtained by changing the optical interference in a semitransparent device by evaporating an optical spacer−mirror stack (MgF2 | Ag) on top of the ITO top contact, as recently shown for perovskite solar cells.[65] Because the device itself is not changed in this experiment, any change in the EQE upon including the optical spacer−mirror stack must stem from the altered optical interference. The optical interference can be accurately modeled using the transfer matrix method,[68] and by comparing the modeling results with measured EQE spectra the defect location can be assessed.Here we apply this method to semitransparent PM6:Y6 devices to uncover the location of the observed defects within the active layer. Sensitive EQE spectra were recorded for three inverted semitransparent PM6:Y6 devices before and after evaporation of 100, 126, or 185 nm MgF2, and 100 nm Ag. Figure 4a reveals significant changes in the defect region as a consequence of changing the interference via different optical spacers. In Figure 4b–d we plot the normalized ratios of the normalized EQE spectra versus wavelength of semitransparent cells with and without the optical spacer−mirror stack as solid lines that are offset vertically for clarity. As further detailed below, the dashed lines correspond to the results of the optical simulations and represent the ratio of the photon densities with and without the optical spacer−mirror stack (expressed as the ratio of the normalized modulus squared of the optical electric field, |E|2), experienced by the absorbing species, being the PM6:Y6 active layer or the defect.[44,65]4Figurea) Semilogarithmic plot of the normalized sensitive EQE spectra versus wavelength of inverted PM6:Y6 devices in a semitransparent configuration or with an optical spacer−mirror stack using either 100, 126, or 185 nm MgF2. b) The ratio of the sensitive EQEs of semitransparent cells with and without an optical spacer−mirror stack versus wavelength for the three MgF2 thicknesses (continuous lines). The dashed lines represent the optically‐simulated ratio of the |E|2 spectra for semitransparent cells with and without the optical spacer−mirror stack assuming absorption occurs throughout the active layer. The measured and simulated ratios are offset per MgF2 thickness and normalized for clarity. The defect fraction versus wavelength is included to visualize the transition between above‐bandgap (PM6:Y6) dominated EQE response and below‐bandgap (defect) dominated EQE response (thick black line). c) Same as b) assuming absorption near the back (PM6:Y6 | MoO3) interface. d) Same as b) assuming absorption near the front (ZnO | PM6:Y6) interface.For the optical modeling it is imperative to first define the defect region in the EQE spectra. To this end, we fit the absorption edge with an Urbach tail (Figure S10a, Supporting Information) and define the onset of the defect region as the wavelength where the experimental EQE starts to deviate from the Urbach tail. The defect contribution can then be expressed as a fraction of the total EQE response. For the inverted semitransparent PM6:Y6 devices, we find that the defect contribution dominates the EQE spectrum from ≈1200 nm onwards (Figure S10b, Supporting Information).To rationalize the observed changes in the EQE spectra we employ transfer‐matrix modelling, using the refractive index − extinction coefficient (n‐k) spectra and thickness of all layers involved in the device stack as input (Figure S11, Supporting Information). Then we average the calculated |E|2 either over the entire active layer (throughout), or for the first 10 nm near the back interface (back, PM6:Y6 |MoO3), or near the front interface (front, ZnO | PM6:Y6).Finally, we calculate the ratio of the |E|2 spectra for semitransparent devices without and with an optical spacer−mirror stack. In Figure 4b,c,d we compare the ratios of the normalized EQEs with the calculated |E|2 ratios. The defect fraction is included as a thick black line for reference. Comparing the measured EQE ratio with the calculated |E|2 ratios assuming charge generation throughout, we find very good agreement for wavelengths below 1000 nm corresponding to above‐bandgap absorption. More specifically, the minima and maxima are correctly predicted at their corresponding wavelengths (Figure 4b). This is to be expected because the above‐bandgap absorption and photocurrent generation are bulk phenomena.Beyond 1000 nm, the defect response starts to dominate the EQE spectrum and the agreement between measurement and simulation is lost when assuming absorption throughout the active layer. If, however, the absorption is assumed to be located near the back interface (Figure 4c) there is excellent agreement between measurement and simulation in the defect region (>1200 nm). Assuming absorption near the front interface (Figure 4d) does not give simulated ratios in agreement with the experimental ratios in any wavelength region. From these results, we therefore conclude that the defect response recorded in Figure 4a stems from the back interface.We repeated the same procedure for three regular architecture semitransparent devices employing AZO and a MgF2 thickness of 100, 230, or 380 nm (Figure S12, Supporting Information). Again, we find that the defect response must be originating from the back interface, being in this case the PM6:Y6 | AZO interface. The defects in PM6:Y6 are thus located near the back interface regardless the device polarity. This could point towards defective molecules or (reactive) contaminants accumulating near the solvent‐air interface during spin coating. Alternatively, defect formation could occur during the thermal evaporation or sputtering of the back contact. During these high energy processes (temperature/plasma), adsorbed species could react with the PM6:Y6 blend.ConclusionIn this work, we used sensitive EQE measurements to analyze the defects present in a layer of PM6:Y6 in an OPV architecture. By exposing the active layer to ambient atmosphere and H2O‐saturated compressed air, we find that a trace constituent in air increases the defect density and reduces device performance. O3 was identified as responsible for the defect formation because it yielded similar results in exposure experiments as ambient atmosphere. From UV–vis–NIR absorption spectroscopy, we find that defect formation involves reaction of O3 with the PM6:Y6 blend. To explain the observed (reversible) increase in defect response upon aging complete inverted‐configuration cells in H2O‐saturated compressed air, we postulate the change in work function of MoO3 to influence the band bending at the PM6:Y6 | MoO3 interface, resulting in an increased density of occupied defect states and consequently higher defect response. Semitransparent devices were fabricated to circumvent spectral distortions by interference effects and the sensitive EQE spectra of these devices showed the low energy defect to originate from PM6. The results from sensitive EQE are supported by ER‐EIS experiments on the DOS of PM6 and Y6 which indicate that O3 causes an increased defect signal near the HOMO of PM6 but did not change the DOS of Y6. Finally, by comparing the EQE ratios of semitransparent devices with and without an optical spacer−mirror stack added, we could attribute the defect response to an absorptive site located near the back interface regardless the n–i–p or p–i–n polarity of the cell architecture.Given the reactivity of O3 and the common design motives in organic semiconductors, it can be assumed that defect formation by O3 is common to many organic semiconductors. In general, during synthesis, purification, or processing π‐conjugated polymers and molecules are exposed to ambient air for shorter or longer time. Oxidation by trace O3 concentrations could be responsible for defects observed in organic semiconductor based devices.Experimental SectionMaterials2‐Methoxyethanol (99+%, Acros Organics), ethanolamine (>99.5%, Aldrich), zinc acetate dihydrate (98+%, Acros Organics), MoO3 (99.97%, Aldrich), 1‐chloronaphthalene (>85% TCI), chloroform, acetonitrile (>99.9%, extra dry over mol. sieves, Thermo Scientific), and TBAPF6 (>99.0%, electrochemical grade, Merck) were employed without further purification. PM6, Y6, and poly(9,9‐bis(3′‐(N,N‐dimethyl)‐N‐ethylammoinium‐propyl‐2,7‐fluorene)‐alt‐2,7‐(9,9‐dioctylfluorene))dibromide (PFN‐Br) were acquired from Solarmer Materials. AZO (N‐21X) and poly(3,4‐ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS) (Clevios P, VP Al 4083) were obtained from Avantama and Heraeus, respectively. ITO substrates were obtained from Naranjo Substrates and the ITO sputtering target was obtained from Angstrom Engineering.Fabrication and AgingPre‐patterned and fully covered ITO‐covered glass substrates were rinsed by sonication in acetone, scrubbing with a soap solution in water, flushing with deionized water and sonication in isopropanol. Prior to spin coating the PEDOT:PSS, ZnO, or PM6:Y6 solution, the substrate was subjected to 30 min UV/O3 cleaning (UVP PR‐100). A ZnO colloidal solution was obtained by dissolving zinc acetate dihydrate (109.7 mg) in 2‐methoxyethanol (1 mL) and adding ethanolamine (30.2 µL). The ZnO solution was stirred for >1 h before use. For the inverted device architecture, the patterned ITO was spin coated with the ZnO solution at 4000 rpm for 60 s and subsequently annealed at 150 °C for 5 min. The PM6:Y6 (1:1.2) active layer was cast from chloroform containing 0.5 vol.‐% 1‐chloronaphthalene at a concentration of 16 mg mL−1. Spin coating proceeded at 3000 rpm for 30 s in nitrogen atmosphere followed by annealing at 110 °C for 10 min. Inverted devices were finalized by thermal evaporation of 15 nm MoO3 and 100 nm Ag.For regular devices, PEDOT:PSS was spin coated at 4000 rpm for 60 s on top of pre‐patterned ITO‐covered glass substrates followed by annealing at 150 °C for 5 min. After depositing the active layer following the procedure described above, PFN‐Br (0.5 mg mL−1 in methanol) was deposited on a rotating substrate at 3000 rpm in N2. Alternatively, AZO was spin coated at 2000 rpm for 60 s, followed by annealing at 75 °C for 1 min, another spin coating at 2000 rpm for 60 s and a final annealing at 75 °C for 5 min. The cells were completed by depositing Ag (100 nm) via thermal evaporation. For semitransparent devices, an ITO (180 nm) electrode was deposited by radiofrequency magnetron sputtering (Angstrom Engineering).PM6 (≈70 nm) and Y6 (≈120 nm) films for ER‐EIS were cast on fully ITO covered glass substrates from chloroform containing 0.5 vol.‐% 1‐chloronaphthalene at a concentration of 8 mg mL−1 and 16 mg mL−1 respectively. Spin coating of these films was done at 1200 rpm for 60 s in nitrogen atmosphere, followed by annealing at 110 °C for 10 min. The films were exposed to high vacuum (<5 × 10−7 mbar) overnight before exposure to different atmospheres and measurement with ER‐EIS.Samples for absorption measurements were produced by casting the PM6:Y6 active layer on a cleaned glass substrate following the procedure for the inverted/regular devices.Exposure to different atmospheres was performed in an airtight flow box with inlet and outlet. For other atmospheres than ambient, the flow box was kept at a slight overpressure relative to ambient pressure. Samples were kept in the dark during aging. Intentional O3 exposure was conducted by guiding the O3 enriched internal atmosphere of an UV/O3 photoreactor (UVP PR‐100) in operation over the semi‐finished devices or PM6:Y6 layers on (ITO covered) glass. During intentional O3 exposure, samples were kept in the dark.Compressed air was taken from a central facility consisting of a large tank, kept at constant pressure. The tank was refilled by ambient air that was filtered, dried, and passed over active coal to remove reactive trace gasses. The air was further dried by a H2O‐absorber upon exiting the tank and before entering the central distribution system. Following initial intake, the air was never exposed to light and could remain inside the tank and central facility for extended periods of time. Owing to the short half‐life of O3 (indicated as several tens of minutes to several hours)[69–71] and the filtering with active coal, the compressed air was virtually free of O3 at the point of use.CharacterizationThe current density–voltage (J–V) characterization was performed using a Keithley 2400 sourcemeter unit while the device was exposed to simulated AM1.5G illumination. The incident spectrum was achieved by using a Schott GG385 UV filter and a Hoya LB120 daylight filter in combination with a tungsten halogen lamp light source.EQE spectra were obtained by illuminating the device with chopped (Stanford Research Systems SR540) light of a tungsten‐halogen lamp that passed a monochromator (Oriel Cornerstone 130) and aperture. The generated current was measured by a low‐noise current preamplifier (Stanford Research Systems SR570) and a lock‐in amplifier (Stanford Research Systems SR830). The devices were simultaneously exposed to bias light (730 nm light emitting diode, Thorlabs M730L4) to better simulate operating conditions under AM1.5G illumination. The device response was quantified by comparing with a calibrated Si solar cell.Sensitive EQE in the sub‐bandgap region was measured using an Oriel 3502 light chopper, a Cornerstone 260 monochromator (CS260‐USB‐3‐MC‐A) with appropriate sorting filters, a Stanford Research SR 570 low‐noise current preamplifier, a Stanford Research SR830 lock‐in amplifier, and a 250 W tungsten‐halogen lamp. A detailed description of the set‐up and measurement can be found in Ref. [65].A PerkinElmer Lambda 1050 UV–vis‐NIR spectrophotometer was used to measure the absorption spectra. n–k spectra were obtained by variable angle spectroscopic ellipsometry using a WVASE32 ellipsometer (J.A. Woollam Co.). Thicknesses were also obtained from VASE and validated by profilometry (Veeco Dektak 150). Simulations were performed using an in‐house adapted version of the open‐source script provided by McGehee et al.[72]The energy resolved‐electrochemical impedance spectroscopy (ER‐EIS) method was adapted from the work of Nádaždy et al.[73] In a three‐electrode electrochemical cell, a glass | ITO working electrode was covered with either PM6 or Y6 of which 28.3 mm2 was exposed to the electrolyte (200 µL, 200 mm TBAPF6 in dry acetonitrile). The counter electrode was a platinum wire (0.5 mm, >99.99%) cleaned in a roaring blue flame. The pseudo‐reference electrode was a silver wire (0.5 mm, >99.9%) cleaned in a roaring blue flame and dipped in aqua regia before being rinsed with water and acetonitrile. The electrodes were connected to a Biologic VSP potentiostat with impedance capabilities and a “p” low‐current option. Before each measurement a voltaic hold of 20 s was applied with the potential set to the starting potential of the ER‐EIS measurement. The potential was increased/decreased with steps of 40 mV to measure the HOMO/LUMO of the organic films, starting at a point in the bandgap. The frequency was fixed at 0.5 Hz with an amplitude of 60 mV. At each potential step an average of six impedance measurements was taken and later converted to the DOS using the equations described by Schauer et al.[74] Measuring HOMO and LUMO was done in separate measurements on different parts of the same film.AcknowledgementsThe authors would like to thank Guus J. W. Aalbers for measuring the sensitive EQE spectrum of cells employing aged ZnO layers. The authors acknowledge funding from the Ministry of Education, Culture, and Science (Gravity program 024.001.035). The work is further part of the Advanced Research Center for Chemical Building Blocks, ARC CBBC, which is co‐founded and co‐financed by Netherlands Organisation for Scientific Research (NWO) and the Netherlands Ministry of Economic Affairs (project 2016.03.Tue). The authors acknowledge the financial support by NWO via a Spinoza grant.Conflict of InterestThe authors declare no conflict of interest.Data Availability StatementThe data that support the findings of this study are available from the corresponding author upon reasonable request.J. M. Shaw, P. F. Seidler, IBM J. Res. Dev. 2001, 45, 3.R. M. Owens, G. G. Malliaras, MRS Bull. 2010, 35, 449.Y. van de Burgt, A. Melianas, S. T. Keene, G. Malliaras, A. Salleo, Nat. Electron. 2018, 1, 386.J. Rivnay, S. Inal, A. Salleo, R. M. Owens, M. Berggren, G. G. Malliaras, Nat. Rev. Mater. 2018, 3, 17086.J. T. Mabeck, G. G. Malliaras, Anal. Bioanal. Chem. 2006, 384, 343.Y. van de Burgt, E. Lubberman, E. J. Fuller, S. T. Keene, G. C. Faria, S. Agarwal, M. J. Marinella, A. Alec Talin, A. Salleo, Nat. Mater. 2017, 16, 414.P. Cheng, G. Li, X. Zhan, Y. Yang, Nat. Photonics 2018, 12, 131.Z. Hu, J. Wang, X. Ma, J. Gao, C. Xu, K. Yang, Z. Wang, J. Zhang, F. Zhang, Nano Energy 2020, 78, 105376.G. Horowitz, Adv. Mater. 1998, 10, 365.G. Simone, M. J. Dyson, S. C. J. Meskers, R. A. J. Janssen, G. H. Gelinck, Adv. Funct. Mater. 2020, 30, 1904205.J. Bauri, R. B. Choudhary, G. Mandal, J. Mater. Sci. 2021, 56, 18837.A. Salehi, X. Fu, D. Shin, F. So, Adv. Funct. Mater. 2019, 29, 1808803.H. F. Haneef, A. M. Zeidell, O. D. Jurchescu, J. Mater. Chem. C 2020, 8, 759.S. Zeiske, O. J. Sandberg, N. Zarrabi, W. Li, P. Meredith, A. Armin, Nat. Commun. 2021, 12, 3603.P. H. Nguyen, S. Scheinert, S. Berleb, W. Brüiting, G. Paasch, Org. Electron. 2001, 2, 105.C. Krellner, S. Haas, C. Goldmann, K. P. Pernstich, D. J. Gundlach, B. Batlogg, Phys. Rev. B 2007, 75, 245115.M. Grünewald, P. Thomas, D. Würtz, Phys. Status Solidi 1980, 100, K139.W. L. Kalb, F. Meier, K. Mattenberger, B. Batlogg, Phys. Rev. B 2007, 76, 184112.M. Meier, S. Karg, K. Zuleeg, W. Brütting, M. Schwoerer, J. Appl. Phys. 1998, 84, 87.D. V. Lang, J. Appl. Phys. 1974, 45, 3023.J. J. Dittmer, E. A. Marseglia, R. H. Friend, Adv. Mater. 2000, 12, 1270.D. Hu, J. Yu, G. Padmanaban, S. Ramakrishnan, P. F. Barbara, Nano Lett. 2002, 2, 1121.M. P. Nikiforov, B. Lai, W. Chen, S. Chen, R. D. Schaller, J. Strzalka, J. Maser, S. B. Darling, Energy Environ. Sci. 2013, 6, 1513.K. Müllen, Nat. Rev. Mater. 2016, 1, 15013.K. H. Hendriks, W. Li, G. H. L. Heintges, G. W. P. van Pruissen, M. M. Wienk, R. A. J. Janssen, J. Am. Chem. Soc. 2014, 136, 11128.M. Streiter, D. Beer, F. Meier, C. Göhler, C. Lienert, F. Lombeck, M. Sommer, C. Deibel, Adv. Funct. Mater. 2019, 29, 1903936.A. Früh, H. J. Egelhaaf, H. Hintz, D. Quinones, C. J. Brabec, H. Peisert, T. Chassé, J. Mater. Res. 2018, 33, 1891.E. J. Meijer, C. Detcheverry, P. J. Baesjou, E. van Veenendaal, D. M. De Leeuw, T. M. Klapwijk, J. Appl. Phys. 2003, 93, 4831.M. L. Chabinyc, R. A. Street, J. E. Northrup, Appl. Phys. Lett. 2007, 90, 123508.H. F. Iqbal, Q. Ai, K. J. Thorley, H. Chen, I. McCulloch, C. Risko, J. E. Anthony, O. D. Jurchescu, Nat. Commun. 2021, 12, 2352.N. B. Kotadiya, A. Mondal, P. W. M. Blom, D. Andrienko, G. J. A. H. Wetzelaer, Nat. Mater. 2019, 18, 1182.J.‐M. Zhuo, L.‐H. Zhao, R.‐Q. Png, L.‐Y. Wong, P.‐J. Chia, J.‐C. Tang, S. Sivaramakrishnan, M. Zhou, E. C. W. Ou, S.‐J. Chua, W.‐S. Sim, L.‐L. Chua, P. K. H. Ho, Adv. Mater. 2009, 21, 4747.P. K. Nayak, R. Rosenberg, L. Barnea‐Nehoshtan, D. Cahen, Org. Electron. 2013, 14, 966.S. N. Yaliraki, R. J. Silbey, J. Chem. Phys. 1996, 104, 1245.T. Huser, M. Yan, L. J. Rothberg, Proc Natl Acad Sci U S A 2000, 97, 11187.Y.‐Y. Noh, D.‐Y. Kim, Y. Yoshida, K. Yase, B.‐J. Jung, E. Lim, H.‐K. Shim, R. Azumi, J. Appl. Phys. 2005, 97, 104504.A. Seemann, T. Sauermann, C. Lungenschmied, O. Armbruster, S. Bauer, H. J. Egelhaaf, J. Hauch, Sol. Energy 2011, 85, 1238.J. Schafferhans, A. Baumann, A. Wagenpfahl, C. Deibel, V. Dyakonov, Org. Electron. 2010, 11, 1693.H. T. Nicolai, M. Kuik, G. A. H. Wetzelaer, B. de Boer, C. Campbell, C. Risko, J. L. Brédas, P. W. M. Blom, Nat. Mater. 2012, 11, 882.M. Nikolka, I. Nasrallah, B. Rose, M. K. Ravva, K. Broch, A. Sadhanala, D. Harkin, J. Charmet, M. Hurhangee, A. Brown, S. Illig, P. Too, J. Jongman, I. McCulloch, J. L. Bredas, H. Sirringhaus, Nat. Mater. 2017, 16, 356.G. Zuo, M. Linares, T. Upreti, M. Kemerink, Nat. Mater. 2019, 18, 588.H. F. Iqbal, M. Waldrip, H. Chen, I. McCulloch, O. D. Jurchescu, Adv. Electron. Mater. 2021, 7, 2100393.S. Zeiske, C. Kaiser, P. Meredith, A. Armin, ACS Photonics 2019, 7, 256.C. Kaiser, S. Zeiske, P. Meredith, A. Armin, Adv. Opt. Mater. 2020, 8, 1901542.J. Melskens, M. Schouten, R. Santbergen, M. Fischer, R. Vasudevan, D. J. Van Der Vlies, R. J. V. Quax, S. G. M. Heirman, K. Jäger, V. Demontis, M. Zeman, A. H. M. Smets, Sol. Energy Mater. Sol. Cells 2014, 129, 70.G. Li, C.‐W. Chu, V. Shrotriya, J. Huang, Y. Yang, Appl. Phys. Lett. 2006, 88, 253503.J. M. Wallace, P. V. Hobbs, in Atmos. Sci., Elsevier, New York, 2006, pp. 153‐207.J. Yuan, Y. Zhang, L. Zhou, G. Zhang, H. L. Yip, T. K. Lau, X. Lu, C. Zhu, H. Peng, P. A. Johnson, M. Leclerc, Y. Cao, J. Ulanski, Y. Li, Y. Zou, Joule 2019, 3, 1140.J. Yuan, T. Huang, P. Cheng, Y. Zou, H. Zhang, J. L. Yang, S.‐Y. Chang, Z. Zhang, W. Huang, R. Wang, D. Meng, F. Gao, Y. Yang, Nat. Commun. 2019, 10, 570.F. Urbach, Phys. Rev. 1953, 92, 1324.D. Y. Kim, J. Subbiah, G. Sarasqueta, F. So, H. Ding, Y. G. Irfan, Appl. Phys. Lett. 2009, 95, 093304.Irfan, H. D., Y. Gao, C. Small, D. Y. Kim, J. Subbiah, F. So, Appl. Phys. Lett. 2010, 96, 243307.M. C. Gwinner, R. Di Pietro, Y. Vaynzof, K. J. Greenberg, P. K. H. Ho, R. H. Friend, H. Sirringhaus, Adv. Funct. Mater. 2011, 21, 1432.M. Shi, T. Wang, R. Sun, Q. Wu, D. Pei, H. Wang, W. Yang, W. Wang, Y. Wu, G. Xie, T. Wang, L. Ye, J. Min, Sci. China Mater. 2021, 64, 2629.P. S. Bailey, Chem. Rev. 1958, 58, 925.R. W. Murray, Acc. Chem. Res. 1968, 1, 313.J. Heeg, C. Kramer, M. Wolter, S. Michaelis, W. Plieth, W.‐J. Fischer, Appl. Surf. Sci. 2001, 180, 36.P. S. Bailey, H. H. Hwang, J. Org. Chem. 1985, 50, 1778.J. Nowaczyk, W. Czerwiński, E. Olewnik, Polym. Degrad. Stab. 2006, 91, 2022.J. Nowaczyk, P. Olszowy, P. Cysewski, A. Nowaczyk, W. Czerwiński, Polym. Degrad. Stab. 2008, 93, 1275.H. Hintz, H.‐J. Egelhaaf, H. Peisert, T. Chassé, Polym. Degrad. Stab. 2010, 95, 818.F. S. Bridson‐Jones, G. D. Buckley, L. H. Cross, A. P. Driver, J Chem Soc 1951, 2999.A. V. Leont'ev, O. A. Fomicheva, M. V. Proskurnina, N. S. Zefirov, Russ. Chem. Rev. 2001, 70, 91.J. Guo, Y. Wu, R. Sun, W. Wang, J. Li, E. Zhou, J. Guo, T. Wang, Q. Wu, Z. Luo, W. Gao, Y. Pan, C. Yang, J. Min, Sol. RRL 2021, 5, 2000704.B. T. van Gorkom, T. P. A. van der Pol, K. Datta, M. M. Wienk, R. A. J. Janssen, Nat. Commun. 2022, 13, 349.R. E. M. Willems, C. H. L. Weijtens, X. de Vries, R. Coehoorn, R. A. J. Janssen, Adv. Energy Mater. 2019, 9, 1803677.S. Admassie, O. Inganäs, W. Mammo, E. Perzon, M. R. Andersson, Synth. Met. 2006, 156, 614.L. A. A. Pettersson, L. S. Roman, O. Inganäs, J. Appl. Phys. 1999, 86, 487.C. J. Weschler, Indoor Air 2000, 10, 269.J. G. Waller, G. McTurk, J. Appl. Chem. 1965, 15, 363.R. Pandiselvam, S. Sunoj, M. R. Manikantan, A. Kothakota, K. B. Hebbar, Ozone: Sci. Eng. 2017, 39, 115.G. F. Burkhard, E. T. Hoke, M. D. McGehee, Adv. Mater. 2010, 22, 3293.V. Nádaždy, F. Schauer, K. Gmucová, Appl. Phys. Lett. 2014, 105, 142109.F. Schauer, J. Appl. Phys. 2020, 128, 150902.

Journal

Advanced Energy MaterialsWiley

Published: Mar 1, 2023

Keywords: energy resolved‐electrochemical impedance spectroscopy; external quantum efficiency; organic solar cells; ozone; sub‐bandgap defects

There are no references for this article.