Stability improvement mechanism due to less charge accumulation in ternary polymer solar cells

Abstract

Ternary polymer solar cells based on a thiazolothiazole-based polymer donor (PTzBT) and a fullerene acceptor (PC₆₁BM) have attracted attention because they show high efficiency and stability by addition of a non-fullerene acceptor (ITIC). However, the performance improvement mechanism is not completely elucidated. Here, we show the stability improvement mechanism due to less charge accumulation in the PTzBT cells with ITIC using operando electron spin resonance from a microscopic viewpoint. We observed two correlations between device performance and number of spins (N_spin) under solar irradiation. One correlation is the decrease in short-circuit current and the N_spin increase in electrons on PC₆₁BM and holes in PTzBT, where the ITIC addition causes the less these N_spin. The other correlation is the increase in open-circuit voltage and the N_spin decrease in holes in ZnO. These findings explain the stability improvement mechanism, showing the correlation between less charge accumulation and higher stability, which is valuable for the development of further efficient and stable polymer solar cells.

Introduction

Polymer solar cells (PSCs) have been actively researched and developed as the next generation of solar cells due to their features such as flexibility, low production cost, and semitransparency, which are difficult to be achieved with silicon solar cells^1,2,3. While binary blend systems in which a p-type material and an n-type material are mixed together to form a photoactive layer have been central to the PSCs, ternary blend systems have also been attracting much attention because they can effectively generate electricity by adding a p-type or an n-type third component material to the binary blend system, improving the power conversion efficiency (PCE)^4,5,6,7,8. However, the performance improvement mechanism is still not completely elucidated from a microscopic viewpoint. Recently, it has been reported that the PCE of a binary blend PSC with a p-type semiconducting polymer based on poly((2,5-bis(3-(2-butyloctyl)thiophen-2-yl)thiazolo[5,4-d]thiazole)-alt-(2,5-bis(3-(2-hexyldecyl)thiophen-2-yl)thiazolo[5,4-d]thiazole)) (PTzBT) and an n-type semiconductor [6,6]-phenyl C₆₁ (or C₇₁)-butyric acid methyl ester (PC₆₁BM or PC₇₁BM) has markedly improved from 7.4% to 10.3% by adding a small amount of a narrow-band-gap non-fullerene acceptor 3,9-bis(2-methylene-(3-(1,1-dicyanomethylene)-indanone))-5,5,11,11-tetrakis(4-hexylphenyl)-dithieno[2,3-d:2’,3’-d’]-s-indaceno[1,2-b:5,6-b’]dithiophene (ITIC) as a third component (Fig. 1a–c)⁸. Furthermore, the stability of the PTzBT cell has been enhanced by the ITIC addition, where the PCE decrease is reduced from 30% to 10% after storing the cells in a nitrogen-filled glove box for 1000 h at 85 °C under dark conditions⁸. As such the PTzBT ternary blend PSC is a promising cell with a high PCE and stability, ITIC is expected as a quite useful additive for improving the photovoltaic performance. Thus, it is interesting to elucidate the ITIC addition effect on the performance improvement mechanism of the PTzBT cells.

Fig. 1: Chemical structures, device structure, and energy-level diagram of PTzBT polymer solar cells.

a–c Chemical structures of PTzBT, ITIC, and PC₆₁BM, respectively. d Device structure of a PTzBT ternary solar cell with ITIC and wiring details in a N₂-filled sample tube. e Energy diagram of the PTzBT cell (ITO/ZnO/PTzBT:ITIC:PC₆₁BM/MoO_x/Ag).

In previous electron spin resonance (ESR) studies on binary blend PSCs, clear correlations between device-performance deterioration and accumulated charges have been studied for typical binary blend PSCs with poly(3-hexylthiophene) (P3HT)⁹, high PCE cells with poly[(4,8-bis-(2-ethylhexyloxy)-benzo(1,2-b:4,5-b)dithiophene)-2,6-diyl-alt-(4-(2-ethylhexyl)-3-fluorothieno[3,4-b]thiophene-)-2-carboxylate-2,6-diyl)] (PTB7)¹⁰ and poly[4,8-bis(5-(2-ethylhexyl)thiophen-2-yl)benzo[1,2-b;4,5-b]dithiophene-2,6-diyl-alt-(4-(2-ethylhexyl)-3-fluorothieno[3,4-b]thiophene-)-2-carboxylate-2,6-diyl)] (PTB7-Th)¹¹, and high stable cells with a copolymer based on poly[(2,5-bis(3-(2-butyloctyl)thiophen-2-yl)thiazolo[5,4-d]thiazole)-alt-(5,10-bis(4-(2-butyloctyl)thiophen-2-yl)naphtho[1,2-c:5,6-c’]bis([1,2,5]thiadiazole) (PTzNTz)¹². These reports have indicated that the accumulated charges contribute to the reduction of device performance. Although encapsulating the cells can eliminate the extrinsic degradation factors such as oxygen and moisture^13,14,15, the cell encapsulation cannot prevent charge accumulation^9,10,11,12. The accumulated charge states in cells or materials can be directly observed and investigated by ESR spectroscopy at a molecular level^{9,10,11,12,16,17,18,19,20,21,22,23,24}. It is very interesting to study the difference of accumulated charge states between binary and ternary blend PSCs and to clarify the effect of non-fullerene acceptor addition and the performance improvement mechanism of ternary blend PSCs.

Here, we performed operando light-induced ESR spectroscopy on the PTzBT:PC₆₁BM binary blend PSCs with and without ITIC to investigate the accumulated charge states at molecular and atomic levels during cell operation. With the time increase of solar irradiation, multiple ESR signals were observed and the ESR intensities of all the PTzBT cells increased in low magnetic fields at around 321.5 mT regardless of with or without ITIC. Interestingly, the ESR intensity increased first then decreased in high magnetic fields at around 327.5 mT. The ESR measurements with layered films of PTzBT binary and ternary blend PSCs and their fitting analyses in the low magnetic fields at around 321.5 mT show several ESR signals which are assignable to holes in PTzBT, electrons on PC₆₁BM, and holes in ZnO, respectively. The accumulation of electrons on PC₆₁BM and holes in ZnO mainly contributes to the deterioration of device performance. Furthermore, the ITIC addition to the PTzBT:PC₆₁BM binary blend PSCs reduces the charge accumulation of electrons on PC₆₁BM and holes in PTzBT, which leads to the less decrease in short-circuit current density of the cells. In addition, the correlation between the increase in open-circuit voltage and the decrease in accumulated holes in ZnO have been observed regardless of with or without ITIC addition, which is due to less vacuum level shift at the device interface. This deep clarification of the ternary blend PSCs will be essential for optimizing the cell structures and enhancing the PCE and stability of the cells.

Results and discussion

Operando ESR measurements of polymer solar cells

The ternary blend PSCs (ITO/ZnO/PTzBT:ITIC:PC₆₁BM/MoO_x/Ag) were fabricated with PTzBT, ITIC, and PC₆₁BM (Fig. 1d). The ESR signals and performance of the fabricated cells were simultaneously measured for the conditions of dark or solar irradiation by AM1.5 G, 100 mW cm⁻² solar simulator (OTENTOSUN-150LX). The energy-level diagram of the PTzBT cell is show in Fig. 1e, whose energy values are taken from the literatures for ZnO and MoO_x (ref. ²⁵), PTzBT and ITIC⁸, PC₆₁BM²⁶, holes of singly charged zinc interstitial (${{{\mathrm{Zn}}}}_{{{\mathrm{i}}}}^ +$) and oxygen vacancy (${{{\mathrm{V}}}}_{{{\mathrm{O}}}}^ +$) in ZnO (ref. ²⁷), respectively.

ESR experiments were conducted using an X-band spectrometer (JEOL RESONANCE, JES-FA200) with a 1.0 mW microwave power at the center of the resonator. The cells and thin films were measured at the center of the ESR cavity. A continuous-wave ESR method with a lock-in detection for an external magnetic field H with a 100 kHz modulation frequency was utilized. Consequently, we cannot observe the light-induced charges with a short lifetime (<10 μs) which contribute to standard cell operation, and can only observe the charges accompanied with spins with a long lifetime (≥10 μs) in cells^9,10,11,12. The solar cells or thin films and a standard Mn²⁺ marker were simultaneously measured in the same ESR cavity to directly compare these obtained ESR parameters. The g factor was calibrated by an ESR software considering high second-order correction of the effective resonance field. We also used another standard sample 2,2-diphenyl-1-picrylhydrazyl (DPPH) to confirm the correctness of the g factor. Since the fluctuation of microwave power etc. can be calibrated by the marker’s signals, the present ESR system remarkably decreases experimental errors. In our study, the H was parallel to the substrate plane unless otherwise stated. All the ESR signals were averaged over 1 h. When the cell’s electrodes are in the cavity resonator, the occurrence of dielectric loss due to the electrodes causes a decrease of about 25% of the Q values of the cavity resonator. However, the impact is calibrated using the ESR intensity of the Mn²⁺ marker. During the ESR measurements, the microwave frequency and the ESR signal of the Mn²⁺ marker were measured simultaneously, and it was confirmed that all ESR signals were adjusted to the same microwave frequency.

Operando ESR signals of PTzBT cells with and without ITIC under device operation

Figure 2a, b show the operando ESR spectra of the PTzBT cells with/without ITIC under short-circuit conditions and Fig. 2c, d show those under open-circuit conditions, respectively. In the dark conditions, all the cells show one ESR signal where the ESR parameters are g factor g = 2.0024(±0.0002) and peak-to-peak ESR linewidth ΔH_pp = 0.16(±0.02) mT. The g factors are calculated by the resonant magnetic field. We calibrated the ESR parameters by the Mn²⁺ marker. The axes of ordinates are plotted with a unit of a peak-to-peak ESR intensity of the Mn²⁺ marker (I_Mn). As the duration of solar irradiation increased, multiple ESR signals were observed and the ESR intensities of all the cells regardless of with or without ITIC increased in the low magnetic fields at around 321.5 mT, respectively. Interestingly, the ESR intensity in the high magnetic fields at around 327.5 mT increased when the solar light was irradiated for 1 h, and then decreased gradually with time, which indicated that the N_spin first increased and then decreased under solar irradiation (Fig. 2c, d). Investigation of the origins for these charge-accumulation states are essential for clarifying the mechanisms of performance improvement and internal deterioration. Thin films of PTzBT:ITIC:PC₆₁BM, PTzBT:PC₆₁BM, ZnO, ITIC, and MoO_x, and PC₆₁BM and PC₇₁BM films on the ZnO layer (ZnO/PC₆₁BM and ZnO/PC₇₁BM) were fabricated and investigated to clarify the origins of these ESR spectra of the cells.

Fig. 2: Operando ESR signals of PTzBT cells with and without ITIC.

a–d Time variation of operando ESR spectra of PTzBT cells (a, c) with and (b, d) without ITIC under solar irradiation under (a, b) short- and (c, d) open-circuit conditions at room temperature, respectively. The substrate planes are parallel to the direction of external magnetic field H.

Identification of cation signal of PTzBT (Signal 1)

Figure 3a shows the ESR spectra of the PTzBT:ITIC:PC₆₁BM and PTzBT:PC₆₁BM thin films under solar irradiation. Under dark conditions, no signal has been observed for these thin films. Under solar irradiation, the above two thin films exhibited same ESR signals where the ESR parameters are g₁ = 2.0023 ± 0.0002 and ΔH_pp1 = 0.16 ± 0.02 mT. An ESR signal of ITIC was not observed in both the PTzBT:ITIC:PC₆₁BM and PTzBT:PC₆₁BM thin films because the same ESR spectra were obtained. Furthermore, almost no signal was observed for ITIC thin film under the dark and irradiation conditions (Supplementary Fig. 3a). Under simulated solar irradiation, excitons are generated by solar irradiation and separated at the interface between PTzBT and PC₆₁BM. Thus, the cation signal of PTzBT and the anion signal of PC₆₁BM can be observed. Since the spectrum of PC₆₁BM anion cannot be observed at room temperature due to the short spin-lattice relaxation time^9,10,11,12, the obtained ESR spectrum is confirmed as a cation (or hole) signal of PTzBT. The cation signal of PTzBT in PTzBT:ITIC:PC₆₁BM film is larger than that in PTzBT:PC₆₁BM film, which suggests that ITIC can promote the charge separation and enhance the cation (or hole) formation in PTzBT.

Fig. 3: ESR signals of thin films of PTzBT:ITIC:PC₆₁BM and PTzBT:PC₆₁BM, and the anisotropy measurements.

a ESR spectra of thin films of PTzBT:ITIC:PC₆₁BM (red line) and PTzBT:PC₆₁BM (blue line) (defined as Signal 1) after 20 h solar irradiation. b Anisotropies of the ESR signals of the PTzBT:ITIC:PC₆₁BM and PTzBT:PC₆₁BM thin films measured by changing the angle θ between the substrate plane and H direction. The error bars are ±0.00005.

The ESR parameters of the Signal 1 accord with those of the cell’s signals under dark conditions (Fig. 2). This indicates that the cations (or holes) in PTzBT formed in the cells are due to the hole transfer from MoO_x under dark conditions because the energy level of the valence band of MoO_x has been reported approximately −5.6 eV^25,28, which is deeper than that of the highest occupied molecular orbital (HOMO) of PTzBT (−5.12 eV) (Fig. 1e)⁸.

Since polymer orientation plays a crucial role in device performance and stability, we studied the orientation in the PTzBT:ITIC:PC₆₁BM and PTzBT:PC₆₁BM thin films by observing the anisotropy of the ESR signal of PTzBT (Signal 1) with changing the angle θ between the substrate plane and H direction (Fig. 3b). The anisotropy of the g factor is observed as g = 2.00233 ± 0.00005 at θ = 0° and g = 2.00258 ± 0.00005 at θ = 90° for the PTzBT:ITIC:PC₆₁BM thin film, which is larger than that for the PTzBT: PC₆₁BM thin film. The results directly show the ITIC addition can enhance the molecular orientation of PTzBT with hole accumulation in the active layer.

To get more accurate anisotropy of the PTzBT molecular orientation with accumulated holes (Signal 1), we conducted density functional theory (DFT) calculation using Gaussian 16W^10,11,18,29. The g-factor shift can be used to determine whether charges are accumulated at polymer-chain ends. It has been reported that residual bromine exists on polymer-chain ends in conducting polymers, such as P3HT^18,29, PTB7¹⁰, and PTB7-Th¹¹, even if the chain-end treatment is conducted. Thus, the model molecules of PTzBT with residual bromine-end (monomer, dimer, and trimer defined as TzBT-Br, 2TzBT-Br, and 3TzBT-Br, respectively) and of PTzBT without residual bromine-end (monomer, dimer, and trimer defined as TzBT-H, 2TzBT-H, and 3TzBT-H, respectively) were calculated using 6-31 G(d,p) basis set and UB3LYP functional. The optimized chemical structures of the PTzBT model trimers and the spin density distribution in cationic states are shown in Fig. 4, where the coordinate axes of x, y, and z-axis are defined as the long, short, and molecular-plane-perpendicular axis of the molecule, respectively. As shown in Fig. 4e–h, the side views of the optimized chemical structures and spin density distribution obtained from the DFT calculation show good coplanarity in the cationic states, respectively. Supplementary Figs. 1 and 2 show the chemical structures of the model dimers and monomers with their spin density distribution in cationic states, respectively. Table 1 summarizes the DFT results of all model molecules. The averaged g factors (g_μν) can be calculated with a root mean square of principal values of g tensors g_μ and g_ν (μ, ν = x, y, z) as follows:

$$g_{\mu \nu } = \sqrt {\frac{{g_\mu ^2 + g_\nu ^2}}{2}} ,\left( {\mu ,\nu = x,y,z} \right).$$

(1)

Fig. 4: Model molecules for DFT calculation and spin density distribution in cationic states.

a, c Model molecules of PTzBT trimers for the DFT calculation: a 3TzBT-H, c 3TzBT-Br. b, d Spin density distribution in cationic states: b 3TzBT-H, d 3TzBT-Br. e, g Side views of model molecules of PTzBT trimers for the DFT calculation: e 3TzBT-H, g 3TzBT-Br. f, h Side views of spin density distribution in cationic states: f 3TzBT-H, h 3TzBT-Br.

Table 1 Calculated g-tensors’ principal values and their averaged values of model molecules of PTzBT with and without bromine end.

Since the g factors of the monomer model-molecule TzBT-Br are so large that they cannot explain the observed g factors, we focus on other g factors. According to the DFT calculation (Table 1), g_y of all the model molecules are larger than other calculated g factors. Thus, the observed large g factor at θ = 90° (g = 2.00258) can be described by the g_y. Thus, the orientation of molecules with charge accumulation is found to be “edge-on”, that is, the molecular plane is perpendicular to the substrate plane, which is inconsistent with the previous study of two-dimensional grazing incidence X-ray diffraction (2D-GIXD) (see the inset of Fig. 3b)^8,30. The reason is thought to be that the accumulated charges are not in the main face-on orientation area but in the edge-on orientation area of PTzBT. Although the g_y of 2TzBT-H (g_y = 2.00258) and 3TzBT-H (g_y = 2.00260) accord with the observed g factor at θ = 90° (g = 2.00258), the g_xz of 2TzBT-H (g_xz = 2.00169) and 3TzBT-H (g_xz = 2.00176) are much smaller than the observed g factor at θ = 0° (g = 2.00233) (Table 1). Consequently, the cations (or holes) may accumulate in the regions with molecular structures similar to TzBT-H, 2TzBT-Br, and 3TzBT-Br under solar irradiation. Since the observed g factor at θ = 90° (g = 2.002583) are smaller than the calculated g_y of TzBT-H (g_y = 2.00277), 2TzBT-Br (g_y = 2.00315), and 3TzBT-Br (g_y = 2.00292), and that at θ = 0° (g = 2.00233) are larger than the calculated g_xz of TzBT-H (g_xz = 2.00165), 2TzBT-Br (g_xz = 2.00199), and 3TzBT-Br (g_xz = 2.00194) (see Table 1), the molecular orientation is not a perfect edge-on orientation and the y-axis direction tilt from the normal direction of the substrate plane. As a result, PTzBT have a slightly disordered molecular orientation and the hole accumulation is thought to exist near the chain ends of PTzBT.

Identification of anion signal of PC₆₁BM (Signal 2) and hole signal of singly charged oxygen vacancy in ZnO (Signal 3)

To identify other origins of observed ESR signals except for the Signal 1 related with PTzBT, thin films of ZnO/PC₆₁BM, ZnO/PC₇₁BM, and ZnO were also measured under solar irradiation. The ESR spectra of the ZnO/PC₆₁BM film under dark condition, where no signal was observed, and 1 h and 20 h under solar irradiation are shown in Fig. 5a. Multiple ESR signals were observed and the ESR intensity in the low magnetic fields at around 321.5 mT increased with irradiation time. On the other hand, the ESR intensity in the high magnetic fields at around 327.5 mT increased at 1 h, then decreased at 20 h. Such signal changes are similar to those observed in the cells as shown in Fig. 2. We performed a least-squares fitting analysis using EasySpin (6.0.0-dev.17) and Igor Pro (6.36 J) to find out the components of the ESR signals for the ZnO/PC₆₁BM film in the low magnetic fields at around 321.5 mT. Figure 5b shows the fitting analysis of the ESR spectrum for the ZnO/PC₆₁BM film obtained after 20 h solar irradiation. The results are evaluated as follows: The component of the ESR signal, whose lineshape is Lorentzian type and ESR parameters are g₂ = 2.0004 ± 0.0008 and ΔH_pp2 = 0.34 ± 0.02 mT, is defined as “Signal 2”. The component of the ESR signal, whose lineshape is Lorentzian type and ESR parameters are g₃ = 2.0027 ± 0.0002 and ΔH_pp3 = 0.21 ± 0.02 mT, is defined as “Signal 3”.

Fig. 5: ESR signals of ZnO/PC₆₁BM layered film and ZnO thin film and fitting analysis.

a Time variation of ESR spectra of ZnO/PC₆₁BM layered film. b Fitting analysis for the ZnO/PC₆₁BM signal in low magnetic fields at around 321.5 mT (Signal 2 and Signal 3). c Time variation of ESR spectra of ZnO thin film in high magnetic fields at around 327.5 mT (Signal 4). These samples were measured with the same conditions used for the cells.

The g factor of the Signal 2 (g₂) is similar to the previously calculated g factor for an anion of PC₆₁BM using a DFT method³¹. We infer that the Signal 2 is an anion signal of PC₆₁BM. To get more accurate identification of the Signal 2, the ZnO/PC₇₁BM film was measured under the same condition used for the ZnO/PC₆₁BM film. Supplementary Fig. 3 shows the ESR signals of the ZnO/PC₇₁BM film. The Signal 2 is found to be the anion signal of PC₆₁BM because the Signal 2 cannot be observed when PC₇₁BM is layered on ZnO (Supplementary Fig. 3b). Thus, we conclude that the Signal 2 (anion signal of PC₆₁BM) can be observed under solar irradiation when PC₆₁BM is layered on ZnO. Since PC₆₁BM anions in the active layer cannot be observed at room temperature due to the short spin-lattice relaxation time as mentioned above^9,10,11,12, the ESR observation of PC₆₁BM anions at room temperature indicates the direct interaction of PC₆₁BM with ZnO, suggesting a mixing of wave function between PC₆₁BM and ZnO. This interaction may enable the ESR observation of PC₆₁BM anions by breaking the symmetry of electron wave function on PC₆₁BM, resulting in the long spin-relaxation time. Thus, the electron accumulation on PC₆₁BM is thought to occur at the PC₆₁BM/ZnO interface. In previous studies, the Signal 3 has been reported as the hole signal of singly charged oxygen vacancies in ZnO (${{{\mathrm{V}}}}_{{{\mathrm{O}}}}^ +$ signal)³². We have observed the Signal 3 (g = 2.0027 ± 0.0002) in the ESR measurements of both ZnO/PC₆₁BM and ZnO/PC₇₁BM films (Supplementary Fig. 3). Therefore, the Signal 2 and Signal 3 are identified as the PC₆₁BM anion signal and the hole signal of ZnO (${{{\mathrm{V}}}}_{{{\mathrm{O}}}}^ +$ signal), respectively.

Identification of hole signal of singly charged zinc interstitial in ZnO (Signal 4)

To identify the ESR signal of the cells in the high magnetic fields at around 327.5 mT (Fig. 2), we studied the ZnO thin film using ESR spectroscopy (Fig. 5c). We define the obtained ESR signal, whose ESR parameters are g₄ = 1.9697 (±0.005) and ΔH_pp4 = 0.54(±0.02) mT, as Signal 4. The ESR parameters and the ESR intensity change accord with those of the cells in the high magnetic fields at around 327.5 mT (Fig. 2). In previous studies, the signal with g $\approx$ 1.96 has been identified as the hole signal of singly charged zinc interstitials in ZnO (${{{\mathrm{Zn}}}}_{{{\mathrm{i}}}}^ +$ signal) (ref. ³²). This g factor is consistent with that of the Signal 4. Thus, the origin of the Signal 4 is assigned to the hole signal of singly charged zinc interstitials in ZnO (${{{\mathrm{Zn}}}}_{{{\mathrm{i}}}}^ +$ signal) (ref. ³²). Immediately after solar irradiation, a large number of holes were suddenly trapped in zinc interstitials, where zinc interstitials can be formed by solar irradiation with short wavelengths (Frenkel reaction²⁷), and the Signal 4 (${{{\mathrm{Zn}}}}_{{{\mathrm{i}}}}^ +$ signal) of the ZnO thin film first greatly increased. After that, during solar irradiation, the Signal 4 (${{{\mathrm{Zn}}}}_{{{\mathrm{i}}}}^ +$ signal) of the ZnO thin film decreased with the time increase of solar irradiation, which indicates that the accumulated charges are detrapped by solar irradiation. The reason for this detrapping is thought to be the recombination between ${{{\mathrm{Zn}}}}_{{{\mathrm{i}}}}^ +$ and electrons generated by solar irradiation with short wavelengths. We can assume the same mechanism occurs for the cells’ signal in Fig. 2 because of the same behavior of the Signal 4. Thus, the N_spin of the cells (Fig. 2) first increases then decreases at 327.5 mT.

We also measured the MoO_x thin film to confirm the ESR signal of oxygen vacancy in MoO_x and hyperfine structures from Mo and its isotopes. As a result, almost no signal was observed under dark and irradiation conditions (Supplementary Fig. 3c). Thus, MoO_x has no effect on the identifications of the ESR signals of the cells.

Fitting analyses of the operando ESR signals for PTzBT cells

Least-squares fitting analyses were further performed to identify the components of the operando ESR signals for the cells obtained after 20 h solar irradiation in the low magnetic fields at around 321.5 mT (Fig. 2). After having tried various functions based on the ESR parameters of the Signal 1, 2, and 3, the sum of one Gaussian and two Lorentzian functions shows the best results for each operando ESR spectrum as shown in Fig. 6. Similar component signals have been obtained using the fitting analyses under short- (Fig. 6a, b) and open-circuit conditions (Fig. 6c, d) regardless of with or without ITIC, respectively. Table 2 summarizes the fitting analysis results. From the results, the same components have been identified under short- and open-circuit conditions, respectively.

Fig. 6: Fitting analyses of the operando ESR signals of the PTzBT cells after 20 h solar irradiation.

a–d Fitting analyses of the ESR signals of the PTzBT cells (a, c) with and (b, d) without ITIC under (a, b) short- and (c, d) open-circuit conditions in low magnetic fields at around 321.5 mT, respectively. Red lines represent the sum of green dotted line (Signal 1, cation of PTzBT), orange dotted line (Signal 2, anion of PC₆₁BM), and purple dotted line (Signal 3, holes in ZnO (${{{\mathrm{V}}}}_{{{\mathrm{O}}}}^ +$)).

Table 2 Results of fitting analyses of g factors and ΔH_pp of Signal 1, Signal 2, and Signal 3 of PTzBT solar cells with and without ITIC under short- or open-circuit conditions.

Correlations between operando ESR signals and device performance

It is essential to clarify the correlation between the device performance and the number of spins (N_spin) that corresponds to the number of accumulated charges in the cells. For this purpose, the device performance (Supplementary Fig. 4) and ESR signals were simultaneously measured in solar irradiation, and their correlations were analyzed. The standard Mn²⁺ marker sample was used to calculate the N_spin. Figure 7a, b show the transient responses of the relative short-circuit current density (J_SC(t)/J_SC(0)) and the calculated N_spin in the low magnetic fields at around 321.5 mT from the signals of the cells with and without ITIC, respectively, where J_SC(t) is normalized to the initial value of J_SC(0) at irradiation time t = 0. A clear correlation is observed between the J_SC(t)/J_SC(0) decrease and the N_spin increase in the low magnetic fields at around 321.5 mT under solar irradiation, as reported in the ESR researches of other polymer solar cells^9,10,11,12.

Fig. 7: Time variation of the N_spin of PTzBT cells and device performance.

a–d Time variation of the N_spin (red circles) and (a, b) J_SC (t)/J_SC (0) (blue line) or (c, d) V_OC (t)/V_OC (0) (blue line) for the cells (a, c) with and (b, d) without ITIC under solar irradiation at room temperature, respectively. Here, J_SC(t) and V_OC(t) are normalized to their initial values of J_SC(0) and V_OC(0) at irradiation time t = 0, respectively. The N_spin are plotted at each averaged time over 1 h. The error bars in a, b, d and in c are ±0.3 × 10¹² and ±0.5 × 10¹², respectively.

The transient responses of the N_spin of the Signal 1, Signal 2, and Signal 3 upon solar irradiation are shown by Green rhombus, orange triangles, and purple squares in Fig. 7a, b, respectively. The total N_spin corresponds to the sum of these three components. The N_spin for the Signal 1, Signal 2, and Signal 3 were calculated by the deconvolution of the ESR signals by the fitting analyzes shown in Fig. 6. The N_spin of the Signal 2 dominates the total N_spin. The N_spin of the Signal 3 shows the similar transient response regardless of with or without ITIC (Fig. 7a, b). However, the N_spin for the Signal 1 and Signal 2 for the cell with ITIC are obviously smaller than those of the cell without ITIC (Fig. 7a, b). The transient responses of the relative open-circuit voltage (V_OC(t)/V_OC(0)) and the N_spin in the high magnetic fields at around 327.5 mT obtained from the cells with and without ITIC are shown in Fig. 7c, d, respectively, where V_OC(t) is normalized to the initial value of V_OC(0) at t = 0. For this case, we observe another correlation between the increase in V_OC(t)/V_OC(0) and the decrease in the N_spin of the Signal 4 (hole signal of ZnO (${{{\mathrm{Zn}}}}_{{{\mathrm{i}}}}^ +$)) under solar irradiation from 1 h to 20 h regardless of with or without ITIC as shown in Fig. 7c, d, respectively. The above two correlations for the J_SC(t)/J_SC(0) and V_OC(t)/V_OC(0) cases are further discussed below.

Performance improvement and deterioration mechanisms of PTzBT solar cells

On the basis of the correlations between device performance and N_spin mentioned above, we discuss the mechanisms of the performance improvement and deterioration of the cells. In the case of cells with charge accumulation, the J_SC can be described as follows^9,10,11,12:

$$J_{{{{\mathrm{SC}}}}}\left( {N_{{{{\mathrm{spin}}}}}} \right) = n{{{\mathrm{e}}}}\mu E = n{{{\mathrm{e}}}}\frac{{\mu _{{{{\mathrm{SC}}}}}\mu _{{{{\mathrm{CA}}}}}}}{{\mu _{{{{\mathrm{CA}}}}} + (\mu _{{{{\mathrm{SC}}}}}/c)N_{{{{\mathrm{spin}}}}}}}E.$$

(2)

Here, n is charge density, e is elementary charge, $\mu$ is the mobility of Matthiessen’s rule, E is internal electric field, $\mu _{{{{\mathrm{SC}}}}}$ and $\mu _{{{{\mathrm{CA}}}}}$ are the mobilities without and with charge accumulation, and c is a proportional constant^9,10,11,12. From Eq. (2), the J_SC(0) can be calculated, which derives J_SC(N_spin)/J_SC(0) as follows^9,10,11,12:

$$\frac{{J_{{{{\mathrm{SC}}}}}\left( {N_{{{{\mathrm{spin}}}}}} \right)}}{{J_{{{{\mathrm{SC}}}}}\left( 0 \right)}} = \frac{{\mu _{{{{\mathrm{CA}}}}}}}{{\mu _{{{{\mathrm{CA}}}}} + \left( {\mu _{{{{\mathrm{SC}}}}}/c} \right)N_{{{{\mathrm{spin}}}}}}} = \frac{1}{{1 + \left( {\frac{{\mu _{{{{\mathrm{SC}}}}}}}{{c \cdot \mu _{{{{\mathrm{CA}}}}}}}} \right)N_{{{{\mathrm{spin}}}}}}}.$$

(3)

Equation (3) shows that the J_SC decreases while the N_spin increases, as observed in Fig. 7a, b. The collision time of charge carriers, and thus their mobility, in the PTzBT cells with and without ITIC decreases because of the scattering of charge carriers by accumulated charges under solar irradiation. This deterioration mechanism is the same as that reported for other polymer solar cells^9,10,11,12.

Compared to the PTzBT cell without ITIC, the PTzBT cell with ITIC was found to have less hole accumulation in PTzBT and less electron accumulation on PC₆₁BM. In the case of holes in PTzBT, the probability of the hole carriers going to the chain ends of PTzBT is thought to be reduced because the ITIC addition enhances the molecular orientation and thus enhances the hole transport between stacked PTzBT molecules with π-π stacking interaction in the active layer (Fig. 3b), which prevents the hole traps at chain ends and thus leads to the less hole accumulation in PTzBT. The enhancement of the molecular orientation of PTzBT have been confirmed by GIWAXS (or GIXD) in the previous study (Fig. S11 of ref. ⁸). The intensities of the PTzBT:PC₆₁BM blend film with ITIC (more than 0 wt%) at q_xy = 0.35 are larger than that without ITIC (0 wt%) (Fig. S11b of ref. ⁸). However, the intensities at q_z = 0.35 are almost same regardless of with or without ITIC addition (Fig. S11a of ref. ⁸). The larger ratio of q_xy to q_z means more molecular orientation of PTzBT. Thus, the ITIC addition can enhance the molecular orientation of PTzBT, which corresponds to the observed PTzBT anisotropy in the ESR measurements (Fig. 3b).

The ITIC molecules can co-exist with PC₆₁BM molecules because the surface energies of ITIC (31.2 mJ m⁻²) and PC₆₁BM (33.3 mJ m⁻²) are similar to each other⁸. Moreover, the ESR signal intensity of PC₆₁BM anions in ZnO/ITIC:PC₆₁BM, which corresponds to the Signal 2, has been reported to be smaller than that of ZnO/PC₆₁BM, as shown in previous study³³, which indicates that the electron accumulation on PC₆₁BM at the ZnO interface is reduced by the ITIC addition. As discussed above, the electron accumulation on PC₆₁BM is thought to occur at the ZnO/PC₆₁BM interface. Consequently, a part of the ITIC aggregations maybe directly contacts with the ZnO layer and thus some electrons can directly transfer from ITIC to the ZnO layer, which reduces the electron accumulation on PC₆₁BM at the ZnO interface. Thus, the ITIC addition in the active layer causes the less N_spin of holes in PTzBT and electrons on PC₆₁BM, which leads to the less decrease in the J_SC(t)/J_SC(0) of the cells. Although the ITIC addition can reduce the electron accumulation on PC₆₁BM, we still observe a lot of electrons accumulated on PC₆₁BM. From our results, PC₆₁BM is not a very ideal n-type material for PSCs because a large number of electrons are accumulated on PC₆₁BM during device operation, which deteriorates the device stability. To improve the stability of the cells, it is essential to reduce the charge accumulation in the active layer during device operation.

Next, the reason of the correlation between the increase in V_OC(t)/V_OC(0) and the decrease in the N_spin of the Signal 4 (hole signal of ZnO (${{{\mathrm{Zn}}}}_{{{\mathrm{i}}}}^ +$)) regardless of with or without ITIC is discussed. As shown in Fig. 7c, d, the V_OC(t)/V_OC(0) increased as the duration of solar irradiation increased. According to the theoretical study, the V_OC of organic solar cells has been reported to be evaluated as follows³⁴:

$$V_{{{{\mathrm{OC}}}}} = \left| {{{{\mathrm{HOMO}}}}_{{{\mathrm{D}}}}} \right| - \left| {{{{\mathrm{LUMO}}}}_{{{\mathrm{A}}}}} \right| + BB_{{{\mathrm{D}}}} + BB_{{{\mathrm{A}}}} - \Delta \Phi _{{{\mathrm{D}}}} - \Delta \Phi _{{{\mathrm{A}}}}.$$

(4)

Here, $\left| {{{{\mathrm{HOMO}}}}_{{{\mathrm{D}}}}} \right| - \left| {{{{\mathrm{LUMO}}}}_{{{\mathrm{A}}}}} \right|$ is the energy difference between the HOMO of the donor and the lowest unoccupied molecular orbital (LUMO) of the acceptor, BB_D and BB_A are the net band bendings in the electrostatic potential, and $\Delta \Phi _{{{\mathrm{D}}}}$ and $\Delta \Phi _{{{\mathrm{A}}}}$ are the energy offsets at the contact³⁴. In previous ESR studies, electrons and holes have been reported to be accumulated, which form an electric dipole layer at the interface between a hole transport layer (HTL) poly(3,4-ethylenedioxythiophene):pol(4-styrenesulfonate) (PEDOT:PSS) and an active layer such as P3HT, PTB7, and PTB7-Th under solar irradiation^9,10,11,12. These charge-accumulation states can lead to the occurrence of energy-level shift and the variance of the $\Delta \Phi _{{{\mathrm{D}}}}$. With the time increase of solar irradiation, the N_spin of holes in P3HT, PTB7, or PTB7-Th (active layer) and electrons in PEDOT:PSS (HTL) increases, which contributes to the increase in the $\Delta \Phi _{{{\mathrm{D}}}}$ and leads to the decrease in the V_OC of those cells^9,10,11,12. In the present case of the PTzBT cells, electrons and holes mainly accumulate on PC₆₁BM (active layer) and in ZnO (electron transport layer (ETL)) under solar irradiation, respectively. These charge-accumulation states lead to the variation of the $\Delta \Phi _{{{\mathrm{A}}}}$, which is different from the above cases of the P3HT, PTB7, or PTB7-Th cells.

From Fig. 7a–d, we find that electrons on PC₆₁BM (Signal 2) and holes in ZnO (${{{\mathrm{Zn}}}}_{{{\mathrm{i}}}}^ +$, Signal 4) dominant the N_spin of the cells, respectively. The hole and electron accumulation states at the interfaces between PC₆₁BM and ZnO under open-circuit conditions are schematically shown in Fig. 8. Before solar irradiation (Fig. 8a), electrons on PC₆₁BM and holes in ZnO are not trapped so we observe no anion signal of PC₆₁BM (Signal 2) and no hole signal of ZnO (${{{\mathrm{Zn}}}}_{{{\mathrm{i}}}}^ +$ signal, Signal 4). Immediately after solar irradiation (Fig. 8b), a large number of electrons on PC₆₁BM and holes in ZnO are suddenly trapped at the interface between PC₆₁BM in the active layer and the ZnO layer, respectively, then the anion signal of PC₆₁BM (Signal 2) and the hole signal of ZnO (${{{\mathrm{Zn}}}}_{{{\mathrm{i}}}}^ +$ signal, Signal 4) begin to be observed because of the Frenkel reaction²⁷ (Figs. 2 and 5a, b). The hole accumulation also can be confirmed by the cells’ transient responses under open-circuit conditions (Fig. 7c, d). Along with the increases in electrons on PC₆₁BM and holes in ZnO, the energy level of PC₆₁BM near the interface become shallower and that of ZnO become deeper (Fig. 8b). This energy-level shift causes the $\Delta \Phi _{{{\mathrm{A}}}}$ increase, which means that the performance deterioration due to the charge accumulation has already occurred at the time of the immediately after solar irradiation regardless of with or without ITIC. This sudden performance deterioration cannot be observed by usual device characterization measurements because they need to use solar irradiation to obtain the photovoltaic properties. Very interestingly, it should be noted that if the cells are irradiated for a long time (Fig. 8c), the N_spin of holes in ZnO (Signal 4) has massively decreased (Fig. 7c, d), which leads to a large decrease in the energy-level shift and causes the decrease in $\Delta \Phi _{{{\mathrm{A}}}}$. In contrast to the slight increase of the N_spin of electrons on PC₆₁BM, the large decrease of the N_spin of ZnO partly recovers the energy-level shift between of the PC₆₁ BM and ZnO layers, which causes the $\Delta \Phi _{{{\mathrm{A}}}}$ decrease and leads to the increase in the V_OC. This V_OC increase during device performance occurs the cells with and without ITIC, which indicates that the optimization of the trapping states in ZnO is very important for further improving the device performance, which can be demonstrated by the present operando ESR study of the cells.

Fig. 8: Schematic variation of the interface energy level between PC₆₁BM and ZnO under solar irradiation.

a Before solar irradiation, electrons and holes are not trapped on PC₆₁BM and in ZnO so the anion signal of PC₆₁BM (Signal 2) and the hole signal of ZnO (${{{\mathrm{Zn}}}}_{{{\mathrm{i}}}}^ +$ signal, Signal 4) are not observed. b Immediately after solar irradiation, a large number of electrons and holes are suddenly trapped at the interface between PC₆₁BM and ZnO layer. c After long-time solar irradiation, the N_spin of holes in ZnO (Signal 4) has massively decreased.

In conclusion, the PTzBT PSCs with and without ITIC have been studied by operando ESR spectroscopy under solar irradiation. We have clarified the origins of four ESR components for the cells from the measurements of the thin films and the fitting analyses for the ESR signals, where the signal 1, 2, 3, and 4 are identified as the cation signal of PTzBT (Signal 1), the anion signal of PC₆₁BM (Signal 2), the hole signal of singly charged oxygen vacancies in ZnO (${{{\mathrm{V}}}}_{{{\mathrm{O}}}}^ +$, Signal 3), and the hole signal of singly charged zinc interstitials in ZnO (${{{\mathrm{Zn}}}}_{{{\mathrm{i}}}}^ +$, Signal 4), respectively. With the time increase of solar irradiation, the intensities of the Signal 1, 2, and 3 increased monotonically, the intensity of the Signal 4 firstly increased then decreased regardless of with or without ITIC. From the ESR anisotropy measurements, we have found the ITIC addition can enhance the molecular orientation of PTzBT with hole accumulation in the active layer. We have observed the clear correlation of the J_SC(t)/J_SC(0) decrease with the N_spin increase regardless of with or without ITIC, which has demonstrated that the charge scattering by accumulated charges lead to the decrease in charge-carrier mobility. The degree of the N_spin increase of the cells with ITIC is lower than that of those without ITIC because of the enhancement of the molecular orientation by ITIC and the direct contact between ZnO and ITIC, resulting in higher performance and stability for the cells with ITIC compared to those without ITIC. This result explains the stability improvement mechanism by the ITIC addition from a microscopic viewpoint. In addition, regardless of with or without ITIC addition, the correlation between the V_OC(t)/V_OC(0) increase and the N_spin decrease has been also observed, which demonstrates that the decrease of hole accumulation in ZnO causes the $\Delta \Phi _{{{\mathrm{A}}}}$ decrease and lead to the increase in the V_OC. From our findings, further decrease in the N_spin under device operation will be essential for the development of PSCs with higher efficiency and stability.

Methods

Device fabrication

An ITO-patterned (15 mm × 2 mm) quartz substrate (20 mm × 3 mm) was ultrasonically cleaned with 2-propanol and acetone. A ZnO (electron transport layer, 40 nm) layer was prepared by spin-coating (1500 rpm, 15 s) a diluted solution of ZnO nanoparticles. PTzBT, ITIC, and PC₆₁BM (purchased form Solenne BV) (1:0.2:2 w/w/w) dissolved in chlorobenzene solvent (approximately 5 g L⁻¹ concentration based on the PTzBT) was mixed at 100 °C for 30 min with a vibrational method. Then, the active layer was spin-coated (600 rpm, 20 s) on the ITO substrate in the nitrogen-filled glove box (O₂ < 0.2 ppm, H₂O < 0.5 ppm). The MoO_x (hole transport layer, 7.5 nm) and Ag (hole collecting electrode, 100 nm) layers were fabricated by a vacuum-deposition method below 4.0 × 10⁻⁵ Pa. The active area is 10 mm × 2.6 mm^8,10,11,12. Thin films of PTzBT:ITIC:PC₆₁BM (1:0.2:2 w/w/w) and PTzBT:PC₆₁BM (1:2 w/w) were spin-coated (600 rpm, 20 s) on a quartz substrate (20 mm × 3 mm) with the same condition used for cells. Thin films and layered films of ZnO, ITIC, ZnO/PC₆₁BM, and ZnO/PC₇₁BM were spin-coated on the quartz substrate with the same condition used for cells using spin-coating conditions for ZnO (1500 rpm, 15 s) and for ITIC, PC₆₁BM, and PC₇₁BM (600 rpm, 20 s)^8,10,11,12. MoO_x thin films (7.5 nm) were fabricated on the quartz substrate by the vacuum-deposition method with the same condition used for cells. The wiring details of a cell are schematized in Fig. 1d and Supplementary Fig. 6. All the samples were encapsulated in an ESR sample tube in the glove box. Other details are described in the references^10,11,12.

Device performance of PTzBT ternary solar cells

The device performance of the PTzBT cells is shown in Supplementary Fig. 4. The reason for the high ITIC response that extends to over 800 nm is an interference effect caused by the reflection of light at the metal electrode, which have been clarified by the optical measurements and simulations in the previous study⁸. Other details are described in the reference⁸.