Introduction

Organic light-emitting diodes (OLEDs) are a type of organic light-emitting device that exhibits several advantages, such as high efficiency, spontaneous light emission, and flexibility, compared to the liquid crystals currently used in displays; as a result, OLEDs have attracted attention in recent years and have been utilized in practical situations1,2,3,4. Since organic materials are easily soluble in solvents and solution processes can be used for organic thin films, OLEDs should be more easily fabricated in the future3,5,6. In addition, since organic materials can be synthesized, there is no fear that resources will be exhausted or the price of raw materials will increase1,6,7.

As a new next-generation light-emitting device, light-emitting electrochemical cells (LECs) have been developed with the same organic light-emitting materials used in OLEDs. LECs are organic light-emitting devices that utilize electrochemiluminescence, and the first LEC with polymers was reported by Pei et al.8. Since then, LECs with phosphorescence due to metal complexes9,10,11,12 or with thermally activated delayed fluorescence (TADF)13,14,15,16,17,18 have been reported. While certain aspects of LECs (speed, lifetime, etc.) cannot compete with OLEDs, there is a unique niche in which LECs can compete. Compared to OLEDs, LECs are less expensive, have a simpler device structure, and do not require electrodes with low work functions19,20. Moreover, low-cost production by a printing method and/or a roll-to-roll method is possible21,22,23,24.

Compared to OLEDs, LECs exhibit several disadvantages, such as a slow response speed and short driving lifetime25, as mentioned above. It is necessary to improve these disadvantages for practical application; however, the performance improvements needed for practical applications have not been obtained, even though much research has been conducted to address these issues. Furthermore, the details of the operation mechanisms have not been clearly clarified8,9,10,11,12,13,14,15,16,17,18,20,26. For example, whether the number of electrochemically doped charges changes under luminescence has never been quantitatively measured, and the orientation of the molecules in which the electrochemically doped charges reside in the LEC structure has not been determined8,9,10,11,12,13,14,15,16,17,18,20,26. Although the electrochemical doping model was reportedly applied to LECs, quantitative doping numbers for p-type and n-type doping have not been revealed during device operation27.

To elucidate the detailed operation mechanisms, it is necessary to directly observe the change in the charge states due to electrochemical doping that occurs inside LECs. Electron spin resonance (ESR) spectroscopy is an effective method for directly observing the charge states inside devices. The ESR technique can nondestructively perform characteristic evaluation of materials at the molecular level by revealing the spin states of charges. We have studied the charge states, molecular orientation with charges, and deterioration mechanisms in OLEDs28, solar cells29,30,31,32,33,34,35, and transistors36,37,38,39,40 with ESR under device operation. By this operando ESR method, we can directly observe the spin states in LECs being driven, which reveals the electrochemically doped charge states in the LECs under device operation and the operation mechanism from a microscopic viewpoint. Electrochemically doped charge states in an LEC with poly(9,9-dioctylfluorene-alt-bithiophene) (F8T2) under light emission at an applied voltage have been observed using the ESR method41. However, the ESR spectrum has been obtained at only one applied voltage41; thus, detailed investigations, such as the change in charge states due to voltage application or light emission, have not yet been reported.

Here, we show the microscopic investigation of LECs with operando ESR spectroscopy from a microscopic viewpoint at a molecular level. We observe the spin states of electrochemically doped charges, which shows the detailed change in electrochemically doped charge states accompanied by voltage application and light emission for the first time. We fabricated an LEC using the typical yellow light-emitting polymer Super Yellow (SY, Fig. 1a) as a light-emitting material and observed the electrochemically doped charge states and their changes under device operation by the operando ESR method. We directly observed the increase in electrochemically doped holes (or cations) and electrons (or anions) with a correlation to the increase in luminance. The results suggest that the electrochemically doped charges are distributed over the light-emitting active layer in the LECs under device operation for the operation mechanism. Moreover, we determined that the SY molecules with electrochemically doped charges are oriented in the LEC structure from the anisotropy of the observed g-factor from the ESR spectra. Thus, fundamental knowledge was obtained by directly observing the spin states of LECs during device operation and will be useful for research on other LECs to elucidate the operation mechanism in detail and further improve device performance.

Fig. 1: Organic materials and device structures for light-emitting electrochemical cells.
figure 1

a, b The chemical structures of Super Yellow (SY) (a), an organic semiconductor yellow light-emitting material, and trihexyltetradecylphosphonium bis(trifluoromethylsulfonyl)amide ([P66614][TFSI]) (b), an ionic liquid. c Cross section of the LEC used in this study. d Schematic structure of the LEC used in this study.

Full size image

Results and discussion

Device fabrication and characterization

To attain a high signal-to-noise (S/N) ratio of the ESR signal by increasing the active area of the device, we utilized a rectangular device structure (3 mm × 20 mm) in an ESR sample tube with an inner diameter of 3.5 mm and a total length of 270 mm29,42. The light-emitting material SY43 (Sigma Aldrich, Fig. 1a) and ionic liquid trihexyltetradecylphosphonium bis(trifluoromethylsulfonyl)amide ([P66614][TFSI]) (Sigma Aldrich, Fig. 1b) were used for device fabrication. The device structure is indium tin oxide (ITO) (150 nm)/SY:[P66614][TFSI] (300 nm)/Ag (100 nm), and the schematics of the device structure are shown in Fig. 1c, d. The light-emitting layer was formed on an ITO substrate by a spin-coating method. The Ag cathode was formed by a vacuum evaporation method. The active area was 0.2 cm2. The fabricated device was sealed in the ESR sample tube together with a desiccant in a N2-filled glove box (O2 < 0.2 ppm, H2O < 0.5 ppm) to suppress the influence of oxygen and moisture44.

The fabricated LECs were evaluated with an X-band ESR spectrometer (JEOL, JES-FA200), a source meter (Keithley, 2612 A), and a luminance meter (TOPCON BM-9M). A standard Mn2+ marker sample was used to calibrate the g-factor, the ESR linewidth, and the number of spins. All measurements were carried out at room temperature. The external magnetic field H was parallel to the substrate unless otherwise stated. The anisotropy of the ESR signal was measured by changing the H direction from parallel to perpendicular with respect to the substrate plane with a step of 15°.

The operand ESR measurements of the LECs were carried out under device operation, in which the voltage applied to the LEC, Vbias, was increased by a step of 0.2 V for 0 V ≤ Vbias ≤ 2 V or by a step of 0.1 V for 2 V ≤ Vbias ≤ 3 V. The change in ESR signal with increasing Vbias was measured along with the current and luminance of the LEC. Figure 2 shows the dependence of the current density (J) and luminance characteristics on Vbias, which were simultaneously measured with the ESR spectra of the LEC. The J and luminance are shown with semilogarithmic and linear plots, respectively. The averaging time to obtain the data was 60 min for 0 V ≤ Vbias < 2 V, 20 min for 2 V ≤ Vbias < 2.8 V, and 10 min for 2.8 V ≤ Vbias ≤ 3 V. Thus, the nominal voltage scan rate was ~0.056 mV s−1 for 0 V ≤ Vbias < 2 V, 0.083 mV s−1 for 2 V ≤ Vbias < 2.8 V, and 0.17 mV s−1 for 2 V ≤ Vbias ≤ 3 V. The J value increases by approximately 4 orders of magnitude for Vbias ≤ 1 V, which may be due to the formation of electric double layers at the interfaces between the organic active layer and electrodes41. The LEC started light emission at Vbias = 2.4 V, and the luminance monotonically increased as Vbias increased, reaching 245 cd m−2 at Vbias = 3 V. Here, we define the threshold luminance level as 1 cd m−2. On the basis of the J and luminance values in Fig. 2, the maximum current efficiency was approximately 1.7 cd A−1 at Vbias = 2.6 V. Although we did not measure the lighting duration, simultaneous measurement of ESR and luminance confirmed that the LECs emitted light for at least 3 h. Here, the lighting duration is the time needed for the luminance to decrease from the initial value to 70%. The above LEC parameters are lower than those in the previous study21, possibly because the device structure used in this study is specialized to achieve a high S/N ratio for the ESR signal (see Fig. 1d) and is not optimal for obtaining high device performance. In other words, the ESR signal intensity is proportional to the active area of the LEC, and the LEC must be inserted into an ESR sample tube with an inner diameter of 3.5 mm. In this case, to increase the active area, an LEC with a special elongated structure, as shown in Fig. 1d, was needed, and then the film quality of the light-emitting layer of the LEC became poor, making it difficult to obtain high device performance. As demonstrated later, J increases in the region of 1 V ≤ Vbias ≤ 2.4 V has been found to be derived from the progression of electrochemical doping into the active layer, which has been suggested by a previous ESR study41.

Fig. 2: Current density and luminance characteristics of the LEC.
figure 2

Dependence of current density (J) and luminance (L) characteristics of the LEC on applied voltage (Vbias). The error bars for J and L are ±2 × 10−6 mA cm−2 and ±1 cd m−2, respectively.

Full size image

Operando ESR of the LECs

Direct evidence for radical (or spin) formation in LECs was obtained by ESR measurements. In these experiments, a continuous-wave method with a modulation frequency of 100 kHz for the external magnetic field H was used29,30,31,32,33,34,35. Thus, spin-accompanied charge carriers with a lifetime of <10 μs, which contribute to the normal current flow in the LECs, cannot be observed in this ESR method. Hence, the observed ESR signals are attributed to stably formed radicals or accumulated (or deeply trapped) charges with spins with a lifetime of >10 μs; charges such as polarons in typical organic semiconductors are accompanied by spins.

Figure 3a shows the dependence of the ESR spectra on Vbias. Almost no ESR signal was observed at 0 V. Therefore, chemical doping by oxygen, etc., to SY does not occur. The intensity of the ESR signal began to change when Vbias was increased from 1 V. This change is attributed to electrochemical doping of SY, as mentioned above. The intensity of the ESR signal further increased with further increasing Vbias above 2.4 V at which the LEC started emitting light. Figure 3b shows the ESR spectrum at Vbias = 3 V application. For the ESR parameters, the g-factor and the peak-to-peak ESR linewidth ΔHpp were determined to be g = 2.00278 ± 0.00003 and ΔHpp = 0.32 ± 0.01 mT, respectively. Here, the g-factor was evaluated from the resonance magnetic field at which the value of the ESR spectrum in first derivative form is zero, and the ΔHpp value was evaluated as the difference between the two magnetic fields at the peak and valley of the ESR spectrum.

Fig. 3: Operando ESR spectra of the LEC.
figure 3

a Vbias dependence of the operando ESR spectra of the LEC. b Operando ESR spectrum of the LEC at Vbias = 3 V. c Difference in ESR spectra at Vbias = 2.1 V and 2.0 V for the LEC.

Full size image

The g-factor reflects the spin-orbit interaction of organic molecules28,29,30,31,32,33,34,35,36,37,38,39,40. This is the magnetic interaction between the spin angular momentum and orbital angular momentum of the molecule. Organic molecules are generally planar in structure, and the orbital angular momentum of their molecular orbitals varies from molecule to molecule. The contribution of orbital angular momentum to the g-factor depends on the direction of the external magnetic field applied relative to the molecular plane. Therefore, by measuring the g-factor, we can identify the molecular species in which the spin is present and evaluate the orientation of the molecule. Additionally, for organic molecules, the ESR linewidth, such as ΔHpp, reflects the magnetic interaction, that is, the hyperfine interaction between the spins of the molecule and the nuclear spins in the molecule28,29,30,31,32,33,34,35,36,37,38,39,40. The interaction depends on the nuclear spins of the organic molecule and varies from molecule to molecule. It also reflects the spin density distribution of positive and negative charges on the molecule. Therefore, the linewidth information is also useful for identifying molecules with spins and their charge state (positive or negative charge).

DFT calculation of charge states

To identify the origin of the ESR signal, density functional theory (DFT) calculations were performed on model molecules of SY using the C.01 revision of the Gaussian 09 program package45. SY is a copolymer, as shown in Fig. 1a, and the x, y, and z parts are randomly copolymerized in the ratio x:y:z = 1:12:12; the ratio has been reported previously43. For this reason, the calculation in this study was carried out using respective model dimer molecules of the x, y, and z parts (see Fig. 4). The values of g-tensors for SY were calculated by weighting the calculated values of the model molecules for the x, y, and z parts of SY by the ratio of the x, y, and z parts of SY, 1:12:12. The molecular coordinate axes are defined as the principal axes of g-tensors of the molecules shown in Fig. 4. Here, the X and Y axes are the directions of the short and long axes of the polymer, respectively, and the Z axis is the direction perpendicular to the molecular plane. The DFT calculations were performed by optimizing the structures using the B3LYP functional and 6-31+G(d,p) basis set for the cations and anions of the x, y, and z parts of SY, respectively. Table 1 summarizes the principal values of g-tensors obtained from the DFT calculations for the cationic and anionic states of the x, y, and z parts of each dimer. Table 2 shows the average of the above principal values of the g-tensors for the cationic and anionic states of SY, which are calculated by weighting and averaging the dimer results of the x, y, and z parts in the ratio 1:12:12. The g-tensors show anisotropy, which means that we can discuss the molecular orientation of SY in the LEC based on the calculated anisotropic g-factors.

Fig. 4: Model molecules and obtained spin density distributions for DFT calculations of Super Yellow.
figure 4

a–c Model dimer molecules for the x (a), y (b), and z (c) parts of Super Yellow. d–i Spin density distribution obtained from DFT calculations for the cationic (d–f) and anionic (g–i) states of the model molecules. The blue and green areas in the middle and lower panels represent positive and negative spin density distributions, respectively.

Full size image
Table 1 DFT calculations for Super Yellow model molecules.
Full size table
Table 2 Principal values of g-tensors (gXX, gYY, gZZ) and their averaged values for Super Yellow obtained from DFT calculations.
Full size table

Cation states of SY with molecular orientation

First, we will discuss the cation states of SY (hole doping). To clarify the cation states, we need to discuss the molecular orientation of SY from the anisotropy of the ESR signal because the observed g-factors can depend on the molecular orientation, as mentioned above (Table 2). Figure 5a shows the angular dependence of the g-factor of the ESR spectra obtained from the measurements. Here, θ is defined as the angle between the H direction and the substrate plane, where θ = 0° and θ = 90° indicate the H direction parallel and perpendicular to the substrate plane, respectively. If the molecular orientation of SY is assumed to be perfectly amorphous, the g-factor should not vary when θ is changed. However, as θ increased, the g-factor monotonically decreased from a maximum value (g = 2.00280 ± 0.00003) to a minimum value (g = 2.00250 ± 0.00003) at θ = 90°. This result demonstrates that charge-doped SY has a preferential molecular orientation.

Fig. 5: Anisotropy of ESR signals reflecting the molecular orientation of Super Yellow in the LEC.
figure 5

a Angular dependence of the g-factor of the ESR spectra of the LEC at Vbias = 3 V. The error bars are ±0.00003. b, c Definition of face-on (b) and edge-on (c) orientations of the model molecule of Super Yellow when the molecular planes of the phenyl groups of the molecular backbone of Super Yellow are parallel and perpendicular, respectively, to the substrate plane. The coordinate axes in b, c are defined as the principal axes of the molecular g-tensors shown in Fig. 4. Here, the X and Y axes are the directions of the short and long axes of the polymer, respectively, and the Z axis is the direction perpendicular to the molecular plane.

Full size image

If the H direction is parallel to the substrate plane (θ = 0°) and perfect face-on orientation with random molecular chain direction (Y direction) occurs on the substrate plane, the averaged g-factors from the X and Y components at θ = 0° are predicted to be gave-XY = \(\sqrt{\left({({g}_{{XX}})}^{2}+{({g}_{{YY}})}^{2}\right)/2}\) = 2.002794 for cations and gave-XY = 2.002885 for anions, as shown in Table 2. If perfect edge-on orientation occurs with a random Y direction on the substrate plane, the averaged g-factors from the Y and Z components at θ = 0° are predicted to be gave-YZ = \(\sqrt{\left({({g}_{{YY}})}^{2}+{({g}_{{ZZ}})}^{2}\right)/2}\) = 2.002408 for cations and gave-YZ = 2.002600 for anions (see Table 2). Here, the definitions of the face-on and edge-on orientations are shown in Fig. 5b, c, in which the molecular plane of the phenyl group of the SY molecular backbone is parallel and perpendicular to the substrate plane, respectively. The observed g-factor at θ = 0° (g = 2.00280) is very close to that of gave-XY = 2.002794 for cations with the face-on orientation compared to other the anions and edge-on orientation mentioned above. Moreover, if the face-on orientation of SY for cations is dominant, the calculated g-factor from the Z component, gZZ = 2.002484 (Table 2), should be observed for the H direction perpendicular to the substrate plane (θ = 90°) because the Z direction of the model molecules is perpendicular to the substrate plane for the face-on orientation (see Fig. 4 and Fig. 5b). As shown in Fig. 5a, the observed g-factor at θ = 90° (g = 2.00250) is in good agreement with the calculated gZZ = 2.002484, which strongly demonstrates the face-on orientation of SY for cations. Thus, the main origin of the ESR signal observed in the measurements is the cations in SY with the face-on molecular orientation. That is, the cations of SY (hole doping or p-type doping) are electrochemically generated, and the corresponding ESR signal is observed because of the hole doping to SY by the Vbias application.

The above p-type doping is supported by previous work in which poly(p-phenylene vinylene)-based polymers, such as SY, were reported to exhibit predominant hole transport46. This predominant hole transport may be related to the spin density distributions of cations and anions calculated for the model dimers, as shown in Fig. 4. That is, in contrast to the anions of the y and z parts, in the cations of the y and z parts, a large distribution of the spin density spreads over the phenyl side chains, which may facilitate hole transport in SY, as discussed in the previous work46.

Anion states of SY with molecular orientation

Next, we will discuss the anion states of SY (electron doping or n-type doping). The knowledge that luminescence occurs as described above suggests that not only SY cations but also SY anions are present due to electron doping. A previous ESR study of LECs reported that the signal intensity of cations is much larger than that of anions in F8T241. Thus, it is inferred that the anion signal is masked by the cation signal in this measurement. Therefore, an analysis was performed using the difference in the ESR signal obtained by subtracting the ESR signal before n-type doping from that after n-type doping. Here, the difference signal between the signals at Vbias = 2.1 V (after n-type doping) and 2.0 V (before n-type doping) was evaluated while the signal intensity at Vbias = 2.0 V was adjusted. In other words, the line shape of signals below Vbias = 2.0 V does not change, suggesting that these signals are derived from p-type doping. As Vbias increases, both p-type and n-type doping-derived signals increase. Therefore, by varying only the signal intensity at Vbias = 2.0 V, the p-type doping-derived signal at each voltage can be reproduced, and the n-type doping-derived signal can be obtained by subtracting only the p-type doping-derived signal contribution from the signal at each voltage. The scaling factor for the intensity at Vbias = 2 V was calculated so that the n-type doping-derived signal exhibited the correct ESR line shape, as shown in Fig. 3c. If the coefficient was calculated incorrectly, the line shape of the n-type doping-derived signal became corrupted, making it impossible to evaluate the correct number of spins by the n-type doping-derived signal, as shown later. Figure 3c shows the difference in the ESR spectrum between the ESR spectra at Vbias = 2.1 V and 2.0 V at θ = 0°. For the ESR parameters, the g-factor and ΔHpp were determined as g = 2.0025 ± 0.0001 and ΔHpp = 0.37 ± 0.05 mT from the difference signal, respectively. These values are clearly different from those of the SY cations determined above. Based on the DFT calculations, the g-factor of this difference signal is close to gave-YZ = 2.002600, which is the average value of the Y and Z components of SY anions (electron doping), compared to gave-XY = 2.002885 for anions in the face-on orientation of SY; thus, the edge-on orientation of SY was revealed (see Table 2). This different molecular orientation, that is, the edge-on orientation for anion formation (electron doping) and the face-on orientation for cation formation (hole doping), can be explained by the dependence of energy levels on molecular orientation for a conduction-band minimum, CBmin (or lowest unoccupied molecular orbital, LUMO), and a valence-band maximum, VBmax (or highest occupied molecular orbital, HOMO), of molecules in aggregate states on substrate planes47,48. Therefore, although the signal intensity of anions is considerably lower than that of cations, we demonstrate that the signal derived from SY anions by electron doping is included in the observed ESR signal after light emission.

We further confirm the origins of the observed ESR signal by measuring the reversible behavior of the ESR signal with a Vbias scan. After Vbias was increased to 3 V and the ESR measurement was obtained, Vbias was returned to 0 V again, and the ESR measurement was carried out. As a result, although there was some hysteresis, the ESR signal finally matched completely with the ESR signal before the Vbias application. This result proves that the ESR signal obtained by this measurement is not due to irreversible changes in molecular structures or oxygen doping but rather to charges derived from pure electrochemical doping.

The molecular structure of SY may be a relevant driving force for the molecular orientation of SY (including face-on or edge-on) with respect to cations or anions. As shown in Fig. 1a, the molecular structure of SY consists mainly of a PV unit in the molecular backbone and phenyl groups with alkyl groups in the side chains. As seen from the optimized structures based on the DFT calculations in Fig. 4, the PV unit of the molecular backbone is planar, and intermolecular interactions are thought to form molecular stacks and contribute to the formation of molecular orientation. On the other hand, the phenyl group of the side chain is twisted and not in plane with the PV unit. This twisting may act as steric hindrance between molecules, resulting in a distribution of face-on and edge-on orientations in molecular orientation.

To further confirm the molecular orientation, X-ray diffraction (XRD) measurements were performed on thin films of light-emitting layers containing SY and ionic liquids; the thin films were fabricated by the spin-coating method as performed for device fabrication. As a result, no clear molecular orientation was observed, as shown in Fig. 6. This result means that no large, molecularly oriented crystal grains are present in the light-emitting layer, which can be attributed to the steric hindrance mentioned above. Thus, while XRD structural analysis is inadequate for detecting small amounts of oriented molecules, this study shows that the ESR method is sensitive enough to detect even the presence of small amounts of oriented molecules by determining the anisotropic g-factor.

Fig. 6: XRD data of a thin film of the light-emitting layer used for the LEC.
figure 6

No clear molecular orientation is observed by XRD measurements, meaning that no large, molecularly oriented crystal grains are present in the light-emitting layer.

Full size image

The molecular orientation can be determined through other methods, such as obtaining optical anisotropy measurements of the refractive index and Fourier transform infrared spectroscopy measurements. However, the sensitivity of these methods is generally less than that of the ESR method, and small amounts of oriented molecules may not be observed accurately, as in the case of XRD measurements. Thus, the edge-on and face-on orientations may not be separated by the other measurement methods mentioned above.

To summarize the above molecular orientation of SY, we conclude that for p-type doping, the molecular plane of the phenyl group of the SY molecular backbone is parallel to the substrate plane (face-on orientation), while for n-type doping, the molecular plane is perpendicular to the substrate plane (edge-on orientation). The orientation is determined by comparing the experimental g-factor anisotropy with the g-factor anisotropy calculated by DFT. The anisotropy of the g-factor indicates the presence of molecular orientation in SY, indicating that it is not a completely amorphous state. Not all SY molecules are face-on oriented; some are edge-on oriented. In this study, we find that p-type doping occurs preferentially in the face-on orientation and n-type doping in the edge-on orientation, which can be explained by the molecular orientation dependence of energy levels for CBmin (or LUMO) and VBmax (or HOMO) of molecules in aggregate states on substrate planes47,48. The different orientation between p- and n-type doping does not mean that the SY molecules rearrange when voltage is applied; it indicates that p- and n-type doping occur in molecules with different orientations. In both the face-on and edge-on orientations obtained in this study, the molecular backbone, or molecular chain, is parallel to the substrate plane. Therefore, our result is consistent with the conclusion of previous work, in which the dipole orientation of SY and the molecular backbone are horizontal to the plane49.

Operation mechanism of the LECs from the correlation between ESR and luminescence

It is interesting to compare the increase in ESR signal with the increase in luminance. Here, the number of spins (Nspin) is used as an index of the ESR signal intensity. The Nspin was evaluated by integrating the ESR spectrum twice and by comparing it with the standard Mn2+ marker sample. The ESR spectrum is obtained by a highly sensitive lock-in detection method and is the first derivative form of the magnetic resonance absorption. Therefore, it is necessary to integrate the ESR spectrum twice to obtain the absolute value of its absorption. Figure 7a shows the Vbias dependence of the Nspin and luminance of the LEC at θ = 0°. Here, the values of Nspin (Total) are derived from the observed ESR spectra including cations and anions of SY, and those of Nspin (Cation) are derived from only cations of SY in the LEC. Figure 7b shows the Vbias dependence of Nspin (Anion) derived from only anions of SY in addition to the luminance. The Nspin (Anion) at each Vbias is evaluated from the difference in the ESR spectrum between the ESR spectra at each Vbias and 2.0 V at θ = 0°, as discussed for Fig. 3c. Nspin (Cation) at each Vbias is calculated by subtracting Nspin (Anion) from Nspin (Total) at each Vbias. Since Nspin (Anion) is two orders of magnitude smaller than Nspin (Total, Cation), as shown in Fig. 7b, the Nspin (Cation) data almost overlap with the Nspin (Total) data (see Fig. 7a). The Nspin (Total, Cation) shown in Fig. 7a gradually increased as Vbias increased above Vbias = 1.4 V and further increased significantly when light emission occurred at Vbias = 2.4 V. Since the spins observed in the measurements have a long lifetime of ≥10 μs and the Nspin increase is proportional to the progress of electrochemical doping in the active layer, the luminance increase is proportional to the progress of doping from this graph. The Nspin (Anion) shown in Fig. 7b gradually increased above Vbias = 2.1 V before light emission, decreased once with light emission at Vbias = 2.4 V and then increased with increasing Vbias. Overall, Nspin (Anion) tended to increase with increasing Vbias.

Fig. 7: Correlation between electrochemical doping progression and luminescence in LECs.
figure 7

a Vbias dependence of the number of spins (Nspin) derived from cations in Super Yellow and luminance (L) of the LEC. As shown in b, in Super Yellow, cation Nspin (Nspin (Cation)) is two orders of magnitude larger than anion Nspin (Nspin (Anion)), so in a, the total Nspin (Nspin (Total)) and Nspin (Cation) are almost the same value. b Vbias dependence of anion Nspin (Nspin (Anion)) in Super Yellow and L of the LEC. The L and Nspin error bars in a and b are ±1 cd m−2 and ±4 × 1013 in a and ±2 × 1011 in b, respectively.

Full size image

As shown in Fig. 7, the two orders of magnitude difference in the number of Nspin between cations and anions indicates that the p-type and n-type doping in the LEC are not balanced. This is probably due to the difference in energy levels between the luminescent material and the ITO anode or Ag cathode. Unpaired cations form an electrical double layer with charges in the electrode at the electrode interface, and there is no electrochemical doping to the light-emitting material. Charges in the electrode are not clearly observed because they are energy degenerate and are in a Pauli-paramagnetic state. Therefore, the ESR signal intensity is lower than that of the isolated spin state following the Curie law and is below the detection limit of the ESR measurement. Additionally, the number of electrochemically doped charges (both n and p) is much less than the actual number of ions in the film. This means that other ions do not contribute to electrochemical doping and that the cation and anion occur in a paired state without polarization.

The correlation between the Nspin increase and luminance increase obtained in Fig. 7 provides insight into the doping process and operation mechanism of the LECs. Figure 8 shows the schematic model diagram of the doping process in the LECs proposed in this study. The horizontal axis represents the channel length of the active layer, and the left and right ends represent the ITO (anode) and Ag (cathode) electrodes, respectively. The vertical axis represents the number of ions of the ionic liquid plus the Nspin of cations (p-type doping) and anions (n-type doping) in SY. The N+ and N- in Fig. 8 represent the number of cations (P66614+) and anions (TFSI−) of the ionic liquid (Fig. 1b), respectively. First, when a voltage is applied, the ions of the ionic liquid polarize to form an electric double layer at the electrodes, reducing the barrier to charge injection from the electrodes (Fig. 8b). Next, only holes are electrochemically doped from the ITO anode at Vbias ≥ 1.4 V, and p-type electrochemical doping of SY occurs first (Fig. 7a and Fig. 8c). As Vbias increases, the number of cations doped in SY (Nspin (Cation)) increases, expanding the p-type doping region (Fig. 8d). When Vbias ≥ 2.1 V, electrons are electrochemically doped from the Ag cathode, and n-type electrochemical doping to SY occurs (Fig. 8e). When the light-mission start voltage Vbias = 2.4 V is reached, light emission occurs due to recombination of hole and electron carriers injected from the ITO and Ag electrodes, respectively. These injections can be attributed to reduced energy barriers due to the enhanced electric double layer at the electrode interfaces (Fig. 8e). As Vbias increases further, the number of cations and anions (Nspin (Cation) and Nspin (Anion)) doped in SY increases, expanding the p-type and n-type doping regions, and the luminance increases further (Fig. 8f). As seen from the Nspin values in Fig. 7, the number of cations electrochemically doped in SY (Nspin (Cation)) is two orders of magnitude larger than that of anions electrochemically doped in SY (Nspin (Anion)).

Fig. 8: Mechanism of LEC operation by electrochemical doping.
figure 8

The vertical axis represents the number of ions in the ionic liquid and the number of Super Yellow doped charges, and the horizontal axis represents the channel length of the light-emitting layer between the ITO and Ag electrodes. The symbols N+ and N− represent the number of cationic and anionic ions in the ionic liquid, respectively. The orange and blue areas represent the p- and n-type doping regions of Super Yellow, respectively. The progression of p- and n-type doping in Super Yellow results in the injection of hole and electron carriers from the ITO and Ag electrodes, respectively, leading to light emission. a Before Vbias application. b Formation of an electric double layer at the ITO and Ag electrodes. c P-type doping by hole injection. d Progression of p-type doping. e N-type doping by electron injection and light emission by recombination of hole and electron carriers injected from the electrodes. f Enhancement in p- and n-type doping regions and associated light emission enhancement.

Full size image

The above model in which electrochemically doped charges are distributed in the light-emitting active layer within the LEC is supported by a recent preliminary study using transmission scanning X-ray microscopy; in this study, a linear distribution of TFSI anions was observed with respect to p-type doping in a light-emitting layer within an LEC structure50, which is consistent with the model described above.

Conclusion

To conclude, we will summarize our most important findings. To microscopically elucidate the charge state and operation mechanism of LECs, operando ESR measurements were performed during device operation while simultaneously measuring the J and luminance of SY-based LECs. The results show that the ESR signals increase with Vbias and light emission and that Nspin, evaluated from the ESR signals, is correlated with luminance. Comparison of the ESR signals with the results of the DFT calculation shows that the ESR signals originate from holes (cations) and electrons (anions) electrochemically doped in SY. The correlation between ESR and luminescence reveals the doping process and suggests that electrochemically doped charges are distributed in the light-emitting active layer of the LECs during device operation. The g-factor anisotropy in the ESR signals of electrochemically doped SY molecules demonstrates that different charge doping occurs in different molecular orientations. In other words, we show that electrochemical doping occurs preferentially with face-on oriented molecules in the p-type doping state and with edge-on oriented molecules in the n-type doping state. These findings are expected to provide deep insight into the charge states and operation mechanisms of LECs and help further improve the device performance.

Methods

Fabrication of LECs

Devices were fabricated using 3 mm × 20 mm ITO substrates with ITO electrodes formed on the surface of quartz glass. An ITO electrode was used for the anode. The ITO substrates were ultrasonically cleaned with 2-propanol and acetone in an ultrasonic cleaner, followed by UV ozone cleaning in an ozone cleaner (Filgen, UV253E) as needed. The solution for the light-emitting layer consists of a light-emitting polymer Super Yellow (Sigma Aldrich, 900438-1 G), ionic liquid trihexyltetradecylphosphonium bis(trifluoromethylsulfonyl)amide ([P66614][TFSI]) (Sigma Aldrich, 50971-5G-F), leveling agent (AGC Seimi Chemical, Surflon S-651), and chlorobenzene (Nacalai Tesque, 08122-25), and the polymer to ionic liquid weight ratio was 5:1. The concentration of the solution relative to the weight of Super Yellow and ionic liquid was 0.95 wt%, and the leveling agent was added at a weight ratio of 1 ppm to the total solution. The solution was stirred in a N2-filled glove box (O2 < 0.2 ppm, H2O < 0.5 ppm) at 500 rpm for more than half a day at room temperature to dissolve completely. The solution of light-emitting layers was spin-coated onto the ITO substrate in the glove box. The spin-coating speed was increased to 500 rpm over 30 s, held at 500 rpm for 30 s, and then increased to 3000 rpm for 0.5 s. The fabricated layer was then dried naturally in the glove box for 1 to 2 min and annealed on a hot plate at 90 °C for 30 min to fabricate the light-emitting layer. Ag cathodes with a film thickness of 100 nm were fabricated using a vacuum evaporation system (ULVAC VPC-260F) at a deposition rate of 3 Å s−1 under 5 × 10−4 Pa. The fabricated device was fixed on a 3 mm × 250 mm polyethylene terephthalate film and wired with copper wires and silver paste. The wired devices were sealed into ESR sample tubes in the glove box.

Device characterization

ESR measurements were performed using an X-band ESR spectrometer (JEOL, JES-FA200), a source meter (Keithley, 2612 A), and a luminance meter (TOPCON, BM-9M). The ESR signals were measured as a function of Vbias by averaging ESR spectra, typically over 10–60 min, depending on signal intensity. The g-factor and linewidth of the ESR signals were calibrated using Mn2+ marker standards. The peak-to-peak ESR linewidth (ΔHpp) was evaluated as the difference between the two magnetic fields at the peak and valley of the ESR spectrum. The number of spins (Nspin) was evaluated by integrating the ESR spectrum twice and comparing it with that of the Mn2+ marker sample. Absolute values of Nspin in Mn2+ marker samples were calculated using a solution (220 μL) of 4-hydroxy-2,2,6,6-tetramethylpiperidin-1-oxyl (TEMPOL) as a standard. Calibration of the g-factor was performed using the JEOL ESR system software program, which considers high second-order correction for the effective resonance field. Another standard sample, 2,2-diphenyl-1-picrylhydrazyl (DPPH), was used to confirm its correctness. XRD measurements were performed using a PANalytical X’Pert Pro MPD powder diffractometer equipped with a rapid counting system (X’Celerator). X-rays were nickel-filtered Cu Kα rays (λ = 1.5418 Å) from a sealed horizontal line focus tube operating at 45 kV and 40 mA.