Flexible and Transparent Electrodes of Cu2−XSe with Charge Transport via Direct Tunneling Effect

In this paper, it is demonstrated that copper selenide (Cu2−XSe) films onto polyester sheets may serve as transparent electrodes in inorganic–organic hybrid light emission devices (IOHLED), as possible replacement to indium tin oxide or fluorine‐doped tin oxide. The Cu2−XSe film synthesized via bath chemical deposition is electrically stable with a sheet resistance of 148 Ω sq−1 and optical bandgap of 2.3 eV. IOHLED are made with poly(3,4‐ethylenedioxythiophene)‐poly(styrenesulfonate) (PEDOT:PSS) as an organic layer for hole transport and poly[2‐metoxy‐5‐(2‐ethylhexyloxy)‐1,4‐phenylenevinylene] (MEH‐PPV) as electroluminescent semiconductor. The IOHLED emits in the visible range owing to the simultaneous emission from Cu2−XSe and MEH‐PPV layers. The enhanced performance is explained by analyzing the charge transport mechanisms at the inorganic–organic interface, which for Cu2−XSe/PEDOT:PSS changes from Fowler‐Nordheim to direct tunneling regardless of the device temperature (90–370 K). The onset voltage is 75% smaller than in the absence of the PEDOT:PSS layer due to a 27 meV decrease in the potential barrier, and the direct tunneling becomes more relevant to device performance than the sheet resistance of the Cu2−XSe layer. Upon adding transparency, mechanical flexibility, and covering large areas, the ultrathin Cu2−XSe films on polyester substrates permit new designs for electro‐optical devices with inorganic–organic heterojunctions.


Gap Energy and Surface Morphology of Cu 2−X Se Films
The optical and morphological properties of Cu 2−X Se films deposited onto polyester sheets were compared with the conventional FTO electrodes. To analyze the absorption coefficient (α) and optical energy gap (E g ) of Cu 2−X Se films we followed the methodology by Tauc applied to amorphous semiconductor materials, [27][28][29] using Equation (1) for the allowed direct transition [4,10,30] α ν ν ) ( where A o is a proportionality constant and hν is the incident photon energy. [28] Figure 1 shows the transmittance spectrum of the polyester/Cu 2−X Se film at room temperature. The transmittance spectrum and Tauc plot of the glass/FTO film are shown in Figure S1 in the Supporting Information, for comparison. The FTO substrate has higher transmittance in the visible spectrum range above 350 nm, while the transmittance for CuSe substrates is higher than 80% above 550 nm. Both films had significant transmittance in the emission spectral region of conjugated and light-emitting polymers (400-750 nm), which is of interest to produce displays. For the polyester/Cu 2−X Se film, light scattering is observed above 350 nm, probably due to the film roughness to be discussed later on. The insets in Figure 1 and Figure S1 (Supporting Information) are the Tauc plots according to Equation (1). From the extrapolation of the linear region of (αhν) 2 versus hν we estimated an optical gap energy E g for FTO and Cu 2−X Se as 3.4 and 2.3 eV, respectively, consistent with the literature. [4,[9][10][11]24,[31][32][33][34] The bandgap for Cu 2−X Se is suitable for organic optoelectronic devices. Figure 2 shows the AFM images (5.0 µm x 5.0 µm) in the tapping mode for polyester, polyester/Cu, and polyester/Cu 2−X Se films. The image of the glass/FTO substrate is shown in Figure  S2 in the Supporting Information. Both polyester and polyester/Cu are more homogeneous and smoother than polyester/ Cu 2−X Se and glass/FTO surfaces. The Cu 2−X Se film is uniform and covers the entire substrate, similar to what is observed for FTO. The thickness of the copper and FTO films was obtained from the AFM measurements (see Supporting Information). The thickness of the Cu 2−X Se film was estimated assuming a parabolic growth with coefficients taken from reference 39. The thickness and mean square roughness (σ RMS ), skewness (σ SK ) and kurtosis (σ KU ) parameters for each image in Figure 3 are given in Table 1. [35] The values σ RMS , σ SK, and σ KU are similar for polyester and polyester/Cu, which suggests that the copper layer is conformable to the polyester substrate. The parameters for copper selenide were close to those for FTO. In fact, both surfaces are similar qualitatively (see Figure 2 and Figure S2 (Supporting Information)). The copper surface changes due to a reaction with the selessosulphate ion forming copper selenide. The thickness of the initial layer increases with the agglomerates originating from nucleation of multiple clusters. The σ KU parameter suggests that the peaks on the film surface are sharper than for an FTO film. This can be an advantage to reduce kurtosis (sharpness of the peaks) on the electrodes, an effect that decreases the performance of electronic devices in the electrode/semiconductor region. The parameter σ SK indicates that the Cu 2−X Se surface has more peaks than valleys.
The height distribution function p(h), where h is the height along the surface, has a Gaussian profile for FTO and Cu 2−X Se  www.advelectronicmat.de films, with the profile for Cu 2−X Se film being 35% narrower. The height profile of copper resembles that of polyester. Overall, we may consider all samples as having a random distribution of heights. The increase in mean quadratic roughness for Cu 2−X Se indicates a larger height dispersion on the film surface. The histograms of area distribution for the cross section of polyester/Cu 2−X Se and Glass/FTO are given in Figure S3 (Supporting Information), with circular clusters for Cu 2−X Se and rectangular clusters for FTO. The average areas of the grain section are 38713 and 20715 nm 2 for Cu 2−X Se (average radius of 160 nm) and FTO (length of 312 nm and width of 123 nm), respectively. These data are consistent with the lowest σ RMS for Cu 2−X Se and reflect the large difference in morphology between FTO and Cu 2−X Se surfaces. Table SI1 in the Supporting Information presents the parameters h c (center) and Δh (width) of the Gaussian fitting of p(h) for the images in Figure 2 and Figure S3 (Supporting Information). The selfcorrelation image between heights in Figure S4 (Supporting Information) was obtained using the second order statistical model, [35] where light regions indicate patterns of distances with most correlated heights. The correlation for polyester has a groove pattern probably due to the lamination used in the industrial process, and this pattern is followed by deposited copper. This pattern disappears after the chemical reaction in the Cu 2−X Se synthesis process with the film growing in a more isotropic manner, similarly to FTO (see Figure S4 in the Supporting Information). We may conclude that Cu 2−X Se grows by homogeneous random nucleation and non-correlated throughout the Cu film. This morphology is similar to those of FTO or ITO in the market, with the advantage of grain shapes that affect the conductivity. Furthermore, it is suitable for adhesion of the organic semiconducting layer.

Electrical Analysis of the IOHLEDs
We used the four-point methodology to determine the sheet resistance (R s ) and electrical resistivity (ρ) of FTO, Cu and Cu 2−X Se. [36,37] The details of layout and parameters are given in the Supporting Information. A constant electric current of 1 mA was used. The results for FTO agree with the data provided by the manufacturer with R s of 14.6 Ω sq −1 , [32] while the average of R s for copper selenide was 148.8 Ω sq −1 , compatible with applications in electronic devices. [38,39] In subsidiary experiments with 100 repeated bending cycles (0-170°), we obtained an average value of 148 ± 5 Ω sq −1 for R s of Cu 2−X Se films, according to the results in Figure S8 in the Supporting Information. Figure 3 shows the characteristic curves of I versus F (electric field) of IOHLEDs at room temperature. The threshold voltages differ in the potential range studied with significant carrier injection in the device with PEDOT:PSS (5 MV m −1 ). This should lower the potential barrier at the junction compared to the IOHLED without the PEDOT:PSS layer (21.7 MV m −1 ). The magnitude of the electric current in these devices is as expected from the literature. [35,36] Note that for conventional OLEDs, for example FTO/PSS:MEH-PPV/Al, the threshold voltage is ≈80 MV m −1 . [40] Hence, despite the increased sheet resistance of the Cu 2−X Se layer in comparison with conventional electrodes, the decrease in the energy barrier at the inorganic-organic semiconductor junction can be significantly more effective in the charge transport along the IOHLED layers. The electrical device stability was verified with 120 consecutive scans polarizing directly the IOHLEDs devices ( Figure S9 in the Supporting Information). No change was observed in the I versus V curves after the 60th scan.
The results in Figure 3 represent a demonstration that efficient IOHLEDs may be obtained with Cu 2−X Se in the transparent electrode and a PEDOT:PSS layer for hole injection. Furthermore, one infers that it is also possible to fabricate a hybrid LED without a PEDOT:PSS layer, but then the electroninjecting layer should be improved for both devices. In order to propose new designs, we investigated the charge transport and interfacial mechanisms in inorganic/organic layers, as discussed next.
Three main charge transport mechanisms are involved on the interfaces of these devices. The first includes the Schottky barrier (BS) formed at the interface between a metal and a semiconductor owing to the difference in work functions. The second mechanism is the Fowler-Nordheim (FN) process, which manifests itself at low temperatures and high electric fields. The third is direct tunneling (DT), significant at small electric fields because carrier injection depends exponentially on the difference in carriers energy and potential energy at the semiconductors interface. [41] In the devices studied here, the Schottky barrier (BS) is formed on the electrode/polymer interface (MEH-PPV/Al) due to the difference between the energy levels, increasing contact resistance and being rectified with the application of a polarization voltage. The injection of charge carriers into organic films occurs through tunneling and where I is the electric current, F is the electric field, φ is the height of the barrier, m eff is the effective mass of the holes injected into the polymer layer, e is the electron charge, h is the Planck's constant, and the thickness of the thin barrier is 1 nm. Note that DT occurs when the applied bias voltage is less than the average barrier height and FN occurs if the bias voltage exceeds that barrier. FN tunneling is a consequence of reducing the width of the barrier. Electrical measurements were performed at various temperatures to examine the competition between thermionic emission and charge transport via tunneling, and the I versus V curves are shown in Figure S9 in the Supporting Information. The device without PEDOT:PSS had a distinctive characteristic curve at higher temperatures, with an increased electric current. However, the characteristic curve had its shape altered slightly with the temperature for the PEDOT:PSS device. The change in electrical current with increasing temperature is due to the thermal activation of charge carriers, with the results for the device without PEDOT:PSS being in accordance with the literature. [44] Figure 4 shows the tunneling plot for both devices from 90 to 370 K. Different charge injection behaviors through the inorganic-organic semiconductor interfacial barrier are identified. FN and DT tuning occur in the device without PEDOT:PSS and DT occurs only in the device with PEDOT:PSS. A detailed analysis of the Cu 2−X Se/MEH-PPV interface shows a non-linear increase in the region of lower polarization, which reveals an FN to DT transition. In contrast, the Cu 2−X Se/PEDOT:PSS/MEH-PPV device shows increasing curves over the polarization interval, which indicates a dominating DT mechanism. When a lower polarization is applied to the Cu 2−X Se/MEH-PPV contact in Figure 4a, carriers have to overcome a wide barrier (rectangular barrier) and charge transport takes place via DT. This result indicates that the DT effect is more important than the sheet resistance of the CuSe layer, i.e., the relevant physical parameter is the decrease in the potential energy barrier at the inorganic/organic interface. For such a decrease allows for considerable carrier injection in the device when the electrode is a p-type inorganic semiconductor. At a higher polarization, the energy barrier reduces, becoming thin and approximately triangular, increasing the probability of FN tunneling. Therefore, there is a change in transport mechanism from DT at the smallest polarization (or larger 1/F) to FN at the highest polarization (smallest 1/F). The intersection occurs at 0.026 m MV −1 at an ambient temperature of 290 K and at 0.07 m MV −1 for the higher temperatures of 330 and 370 K. The crossing point is known as the transition voltage (V trans ). Figure S10 in the Supporting Information shows details of the transition of the two regimes for the polyester/Cu 2−X Se/MEH-PPV/Al device.
Two factors compete to determine how the tunneling plot changes with temperature in Figure 4, which can be detected at lower polarizations. The first relates to the vibrational levels of the polymer, which become more accessible with increasing temperatures (charge carriers have more energy to jump over barriers on the interfaces). The second is the aforementioned vacancy defects (traps) in Cu 2−X Se, where the charge carriers are susceptible to being trapped. In the device without PEDOT:PSS in Figure 4a the access to the vibrational levels of the polymer increases with temperature, while trapping due to Cu 2−X Se defects decreases. In the device with PEDOT:PSS in Figure 4b, the two processes occur equally, one compensating the other. Therefore, the values of electric current for this sample vary little with temperature as shown in Figure  S9d in the Supporting Information. At low temperatures, we observed only DT in the device without PEDOT:PSS. With less thermal energy, the charge carriers are less likely to overcome the energy barrier on the interface. Therefore, the low temperature mechanism is tunneling. The transition from DT to FN occurred only at temperatures 290, 330, and 370 K. We do not rule out the mechanism via hopping because the crossing point between the two tunneling processes varies. This transition occurs at lower voltages at higher temperatures (330 and 370 K). We estimate the barrier height, φ, at 27 meV based on the FN tunneling equation. We assume the linear dependence in the higher field interval (≈30 V m −1 ) and the effective mass of the On the interface of a p-type semiconductor (Cu 2−X Se) with an intrinsic semiconductor (MEH-PPV), there is a discontinuity in the energy bands due to the reorganization of the charge carriers. Discontinuity represents a barrier that hinders carrier diffusion from Cu 2−X Se to MEH-PPV. An energy barrier also exists in the Cu 2−X Se/PEDOT:PSS/MEH-PPV device. However, PEDOT:PSS favors hole injection in the active layer. In both devices, the energy barrier is a consequence of intrinsic defects such as vacancies of non-stoichiometric compounds like Cu 2−X Se, causing a charge decompensation. The traps make it difficult to inject holes in MEH-PPV, contributing to the balance of charge carriers. The PEDOT:PSS layer causes passivation of the copper selenide surface, minimizing the interaction of Cu 2−X Se defects in the charge injection process. This explains the difference in behavior of the devices analyzed. The energy band diagram models proposed in Figure 5 are approximate models, since intrinsic defects in Cu 2−X Se are neglected. The values of LUMO and HOMO energies for Cu 2−X Se films are -4.4 and -6.6eV, respectively. [4,45] Based on our results, we can assume that for the IOHLED Cu 2−X Se/MEH-PPV/Al there is an interface region creating a small barrier, originated in the interstitial defects of Cu 2−X Se. In the Cu 2−X Se/PEDOT:PSS junction, the residual charges of the organic layer should passivate the surface of the inorganic semiconductor.

Electroluminescence of the IOHLEDs
Electroluminescence is observed only with the device without PEDOT:PSS after the stability test (after 120 scans), as shown in Figure S9 in the Supporting Information. Probably, the hole injection through the PEDOT:PSS layer is higher than the electron injection provided by aluminum, resulting in the imbalance of electron-hole pairs required in the active layer for light emission. Figure 6 shows the normalized photoluminescence spectra of MEH-PPV, polyester/Cu 2−X Se and polyester/Cu 2−x Se/ MEH-PPV films excited at 405 nm at room temperature. It is worth noting that the presence of PEDOT:PSS does not change the emission spectrum (not shown) of polyester/PEDOT:PSS/ Cu 2-x Se/MEH-PPV film. The optical absorption spectra of these films are shown in Figure S11 in the Supporting Information. Two characteristic peaks of the MEH-PPV polymer represent the emission of zero phonon at 588 nm (transition 0-0) and the first phonon replica at 632 nm (transition 0-1), and there is an emission band of the Cu 2−X Se at 504 nm. The emission spectrum of the polyester/Cu 2−X Se/MEH-PPV sample shows the overlap of both emission bands of Cu 2−X Se and of MEH-PPV semiconductors. The electroluminescence spectrum (15 V at room temperature) is also shown in Figure 6 for the device without PEDOT:PSS. We identified two peaks from the emission of Cu 2−X Se at 504 nm and zero phonon emission from MEH-PPV at 578 nm. The third peak at 637 nm represents the first phonon replica of MEH-PPV. The line intensity at 637 nm increased substantially for the electroluminescence spectrum probably due to the self-absorbing effect of the light transmitted through the Cu 2−X Se layer. This can be controlled if we change the thickness of each active layer, thus introducing another parameter to adjust the spectral emission center of the device.
An important feature to be highlighted is the wide emission spectrum for the IOHLED upon combining emitting inorganic and organic semiconductors, with the inorganic one also serving as hole transporting layer. Since this latter layer of Cu 2−X Se is transparent and flexible, it may replace ITO or FTO. Table 2 lists transparent oxides, graphene, carbon nanotubes, composites, and conducting polymers used in p-type hole injecting electrodes. Useful characteristics of the electrodes are: transparency above 60% in the visible region (≈500 nm), sheet resistance ≈100 Ω sq −1 , mechanical flexibility, even when synthesized with different techniques. Also in Table 2 is a comparison of physical properties of Cu 2−X Se and other electrode materials, with chemical bath deposition being the technique that permits large area applications.

Conclusions
Copper selenide is stable and can be used as a hole injector electrode in IOHLEDs devices in the form of thin transparent  www.advelectronicmat.de flexible films. The following characteristics were determined for 140 nm thick films: bandgap energy of 2.3 eV, edge of the absorption band close to 450 nm, maximum photoluminescence spectrum at 504 nm (excitation at 405 nm) and sheet resistance of 148 Ω sq −1 . The roughness and homogeneity are comparable to those of commercially-acquired FTO films, with the advantage that its morphology can favor reduction of the tip effects and the accumulation of charge in small regions of the electrode. In the Cu 2−X Se/MEH-PPV/Al IOHLED there is a transition from tunneling FN to DT, with a barrier height of 27 meV. With the addition of an intermediate layer of PEDOT:PSS hole carrier, the IOHLED featured only the DT charge injection mechanism. This explains the considerable decrease in its operating voltage, demonstrating that even under high electric fields there is no significant reduction in the energy barrier. As a result, the direct tunneling in the Cu 2−X Se/PEDOT:PSS interface allows for a low operating voltage in the device, despite the increased sheet resistance. In both devices the charge transport mechanism via hopping must occur along with the tunneling mechanism in these interfaces. The PEDOT:PSS device did not show detectable electroluminescence in our experimental apparatus, probably due to the significant imbalance of charge injection. In the device without PEDOT:PSS, the EL curve shows a red shift in relation to the PL spectrum. While the wide emission spectrum is not suitable for emission of pure colors, it creates the opportunity to fabricate devices for lighting. Its emission region may be possibly controlled by changing the thickness of the Cu 2−X Se layer or the organic active layer. The optical and electrical results are demonstration of the possible use of Cu 2−X Se films as transparent, flexible electrodes in hybrid lightemitting devices processed at low cost and large areas.

Experimental Section
Cu 2−X Se Films: Cu 2−X Se substrates were obtained by evaporating a thin (< 50 nm) layer of copper onto a transparent polyester sheet (polyester film for inkjet printers, thickness ≈ 100 µm) under a vacuum of ≈8 × 10 −3 mbar. A sodium selenosulfate (Na 2 SeSO 3 ) stock solution was prepared. [46] This reactant cannot be isolated in the pure state, but it can be kept stable for several months in aqueous 0.10 mol L −1 Na 2 SO 3 . A mixture of selenium (5 g), Na 2 SO 3 (12 g) and ultrapure water (150 mL) was boiled for 12 h using a condenser to prevent loss of solvent. The mixture was left to rest for several hours at room temperature, and residues were discarded. The actual concentration in the clear solution was determined by gravimetry. An aliquot (5 mL) of the stock solution was decomposed with concentrated HCl. Selenium precipitates were filtered, washed, dried and weighed (0.039 g). For the chemical bath deposition, the stock solution was diluted to 8.0 mmol L −1 with 0.10 mol L −1 Na 2 SeSO 3, where the copper/polyester substrate was immersed for 1 min to 2 min under magnetic stirring. After removal, it was washed with ultrapure water. X-Ray diffraction (XRD) and energy dispersive X-ray spectroscopy (EDS) were carried out using, respectively, a diffractogram X-Ray Shimadzu XRD6000 and a Tescan Vega 3 LMU. XRD and EDS curves typical of berzelianite phase are shown in Figures  S12 and SI3, respectively, in the Supporting Information. [7,47] Light Emitting Diodes with Inorganic-Organic Multilayers: The substrates containing copper selenide were cleaned with ultrapure water and detergent. Figure 7a depicts the film thus obtained. A nickelplated copper wire was used as electrical contact on the copper selenide electrode (anode). Silver glue and epoxy glue were used to cover, insulate and protect the contact. Spin-coating was employed to prepare the active layer of MEH-PPV from a 3.3 mg mL −1 solution in chloroform (Aldrich). [40]  Transmittance ≈ 90% Flexibility: no (rigid glass substrate) Synthesis: DC pulsed (500 W) or RF (120 W) sputtering process on single-layer graphene on a Cu foil and subsequent wet-transfer process to a target substrate Graphene/Ag nanowire [57,58] Optical bandgap ≈ 4.9 eV Sheet resistance ≈ 26.4 Ω sq −1 Transmittance ≈ 91.5% σ RMS ≈ 6.4 nm Flexibility: no (rigid glass substrate) Synthesis: CVD Carbon Nanotubes (CNT) [59] Optical bandgap ≈ 4.1 eV Sheet resistance ≈ 100 Ω sq −1 Transmittance ≈ 90% Flexibility: no (rigid glass substrate) Synthesis: micro-contact printing CNT films with polydimethylsiloxane (PDMS) stamp Single-Wall Carbon Nanotube (SWNT) [60] Optical bandgap ≈ 4.7 eV Sheet resistance ≈ 60 Ω sq −1 Transmittance ≈ 45% Flexibility: no Synthesis: pulsed laser vaporization technique SWNT/PEDOT [61] Optical bandgap ≈ 1.6 eV Sheet resistance ≈ 160 Ω sq −1 Transmittance ≈ 86% Flexibility: yes Synthesis: in situ polymerization of PEDOT on poly(ethylene naftalina)/SWNT Polyethylene Terephthalate/ polyaniline:camphor sulfonic acid (PET/ PANI:CSA) [52] Optical bandgap ≈ 2.5 eV Transmittance > 70% Sheet resistance ≈ 100 Ω sq −1 Flexibility: yesSynthesis: spin-cast film of dissolved in m-cresol PANI:CSA on PET sheet www.advelectronicmat.de The final film thickness was ≈200 nm. The rotation speed was 1400 rpm for 60 s under a relative humidity of 20%. A second device was made by interposing a hole injection layer of PEDOT:PSS (Sigma-Aldrich), which was deposited by spin coating at 3000 rpm for 60 s from a 1.3% dispersion in chloroform. The sample was then heated under vacuum at 120 °C for 1 h. Finally, aluminum was evaporated under vacuum of ≈8 × 10 −3 mbar and the active area of the devices was 7.1 mm 2 . Figure 7b,c shows the IOHLED architectures which is referred to as Cu 2−X Se/ PEDOT:PSS/MEH-PPV/Al and Cu 2−X Se/MEH-PPV/Al, respectively. Device Characterization: Optical absorption measurements were performed using the UV-VIS 800XI Femto spectrophotometer and surface morphology was studied using an atomic force microscope (AFM) Shimadzu SPM-9600. For continuous current (CC) four-probe electrical measurements, a Keithley 2410-C voltage source was used. Photoluminescence spectra were obtained using the Laser Line iZi as the excitation source at 405 nm, and the light emitted by the sample was guided by a set of biconvex lenses and detected with a portable USB4000 spectrometer from Ocean Optics. A 450 nm cutoff high-pass filter was placed in front of the spectrometer to cut the corresponding excitation wavelength. In the electroluminescence measurements, a Keithley 2410-C source was as CC source and the emitted light was collected and detected as described above. Conductivity measurements in CC condition were performed at various sample temperatures (90-370 K) using the Keithley 2410-C voltage source in the planar IOHLED geometry. The samples were kept in a closed-circuit cryostat of helium gas and in a He atmosphere with pressure of 10 −3 mbar to prevent photodegradation.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.

www.advelectronicmat.de
Silésia F. C, Silva, Institute of Physics, Federal University of Uberlândia, Brazil. Silésia de Fátima Curcino da Silva has a MSc in physics from the Federal University of Uberlândia (UFU), Brazil, working on electrical properties of organic field effect transistors. She obtained a PhD degree in 2016 from the same university with a study of the performance of PLEDs. She completed postdoctoral research in copper selenide electrodes in 2018, and is a lecturer at UFU. She is the author of 8 publications, and has experience in polymers and their applications in organic electronic devices and periodic 2D nanostructures using lithography techniques. Bruno S. Zanatta, Institute of Physics, Federal University of Uberlândia, Brazil. Bruno Souza Zanatta received his BSc from the Federal University of Uberlândia, Brazil, working on using Arduino to automate the release of curcumin. He is currently a MSc student at the same university, working on organic light-emitting devices under the supervision of Prof. Alexandre Marletta. www.advelectronicmat.de