1. Introduction

Recently, solid-state light-emitting electrochemical cells (LECs) [1] have received intense research interest since they are viewed as cost-effective alternatives of organic light-emitting diodes (OLEDs) [2]. In general, LECs show several promising advantages such as simple fabrication procedures, low operation voltage and compatibility with air-stable cathode metals. The emissive layer of an LEC contains ionic species and ionic movement results in electrochemical doping of the emissive material to facilitate carrier injection from electrodes. To explore the physical insights of electrochemical processes in LECs, some related studies have been reported. The potential profiles of planar polymer LECs were measured by scanning kelvin probe microscopy (SKPM) to correlate the local potential and electroluminescence (EL) profiles [3,4]. The p-n junction of width of a planar LEC was estimated by concerted optical-beam-induced-current (OBIC) and scanning photoluminescence (PL) imaging studies [5]. The ion redistribution in LECs was depicted by secondary ion mass spectrometry (SIMS) [6–8]. In addition, Fourier-transform infrared (FTIR) microscopy was used to monitor anion density for time-resolved mapping of electrochemical doping in an LEC [9]. By analyzing incident photon-to-current conversion efficiency, angular emission spectra with optical simulations and capacitance measurements, a clear p−i−n structure along with emission zone position in an LEC can be obtained [10]. Electrochemical doping of the emissive material also affects the refractive index and the refractive index change has been correlated to electrochemical doping concentration [11]. The refractive index distribution of the emissive layer is required in optical simulation of LECs to calculate the light outcoupling efficiency [10,12–16] and to estimate the emission zone position [10,17–20]. However, the optical simulation based on the refractive index distribution estimated from the assumed electrochemical doping profile, e.g., constant, graded [14] and constant gradient doping [15], was not accurate since the actual electrochemical doping profile in an LEC remains unclear. Time-dependent refractive index distribution of the emissive layer is critical in optical simulation of temporal evolution in optical properties of LECs.

In this work, we demonstrate a novel technique to in situ probe time-dependent refractive index change profile in a thin film based on an ionic Ru (II) complex during electrochemical doping by employing position sensitive detector (PSD)-based Nanoscale Confocal Microscope (NCM). In this way, time-dependent refractive index change profile in the emissive layer of an LEC can be directly measured and these data can be used to improve the accuracy of optical simulation.

2. Experimental

2.1 Experimental setup of PSD-NCM

The optical layout of the PSD-NCM is illustrated in Fig. 1(a). A 638-nm CW laser (LP637-SF70, Thorlabs, USA) with a 70 mW maximum output power and a ∼1.2 nm full-width-half-maximum (FWHM) bandwidth was employed as the excitation source [21,22]. The operation wavelength in the PSD-NCM is selected as ∼ 638-nm because the photon energy is mostly outside the absorption window of our organic compound. The output beam from the light source was guided across a non-polarizing beam-splitter cube (CM1-BP145B1, Thorlabs) then entered a 2D galvanometric (x-y) scanner. As shown on the left-hand side of Fig. 1(a), between the 2D scanner and the microscope objective (LMPLN-IR 100X, Olympus), two lenses (L₁ and L₂) were placed in accordance with the 4f-configuration [21,23]. By varying the tilted angles of the 2D x-y scanners, the focused laser beam was sequentially scanned over the surface of our LEC specimen. The optical path length (OPL) variation in the z direction arisen from ion migration inside LEC cell was recorded across the x-y interface. The acquisition time of an image was around 2 seconds. With different driving voltages between two electrodes via a power supply, the OPL variation measured between the coverslip (Surface 1 of Fig. 2(a)) and the substrate (Surface 2 of Fig. 2(a)) revealed (Fig. 1(b)) the refractive-index changes (Δn) in LEC while the gap between two surfaces was fixed. The principle to measure the nanoscale OPL differences is through the combination of Grating and PSD, which will further be illustrated in Sec. 3.1 and Fig. 2.

 

Fig. 1. (a) Schematic illustration showing the setup of the PSD-NCM system including the LEC specimen. GM: galvanometric (x-y) scanner; OBJ: Microscope Objective; PH: Pinhole; PSD: position-sensitive detector. (b) Cross-sectional configuration of the LEC sample with two silver electrodes sandwiched between two transparent slabs. (c) Molecular structure of LEC material, Ru(dtb-bpy)₃(PF₆)₂.

Download Full Size | PDF

2.2 Device configuration and molecular structure of organic compound

The configuration of our LEC sample was depicted in Fig. 1(b). The ionic Ru (II) complex material was sandwiched between two transparent slabs, the coverslip and the glass substrate, with a gap in between. The LEC layer was spin-coated on the substrate from the acetonitrile solution of Ru(dtb-bpy)₃(PF₆)₂ (120 mg mL⁻¹) at 3000 rpm. The sample was then baked at 70 °C for 10 hours in a vacuum oven to remove the residual solvent. Two metallic electrodes (Ag) were then thermally evaporated at the two ends of the glass substrate before the encapsulation of the top slab. The separation between the two electrodes is around 40 µm along y-direction in this cell design. The x-y-z- coordinate was also displayed in Fig. 1(a) for comparison. While performing the OPL scanning by the PSD-NCM system, the lower surface of the LEC layer was set to be the scanned surface as illustrated by the red dotted line in Fig. 1(b). Since the gap between the coverslip and the substrate was fixed to L (around 0.3 µm), the change in refractive-index solely contributed to the measured OPL change of LEC. Thus, the OPL was calculated as “refractive-index in the gap” times L. The molecular structure of LEC material for this study, Ru(dtb-bpy)₃(PF₆)₂, was presented in Fig. 1(c). Under an applied bias across two electrodes, the molecule will be ionized and anions/cations will be formed inside the cell gap followed by ions migration.

 

Fig. 2. Working principle underlying the proposed PSD-NCM. (a) Positive/negative ions distribute uniformly in bulk material with zero applied voltage. The OPL of the incident light is indicated by the green dotted line under surface 2. (b) Ions distribution after applying a positive voltage potential pointing from left toward right with corresponding OPL shown on the bottom. (c) and (d): The read-out optical system of a nanoscale optical ruler whose depth position depends on different ion-distributions within the LEC cell. The effective index of LEC cell, n_eff, is equal to n₀ in 2(c) and equal to n₀-Δn in 2(d). The black-dashed lines show the refraction at Surface 1 due to the effective index of LEC cell. BS: Beam Splitter; PSD: Position Sensitive Detector; OPL: Optical Path Length.

Download Full Size | PDF

3. Results and discussion

3.1 Principle of probing OPL difference

Figure 2 illustrates the principle for the non-invasive probing of refractive-index change profile. Assume a bulk material with refractive index n, which is sandwiched between Surface 1 and Surface 2 (scanned surface of Fig. 1(b)) as shown in Fig. 2(a). The distance between these two surfaces is L and anions/ cations are uniformly distributed in between. The OPL of the incident light can be simply calculated by the fundamental formula, OPL= $n \cdot L$. Since positive/negative ions are homogeneously located inside the material, the OPL across the bulk from left to right is a constant as indicated by the green dotted line beneath Surface 2. After a potential difference was applied at the two ends of the specimen, positive and negative ions migrated to the ends with lower and higher voltage potential, respectively. As shown in Fig. 2(b), the symmetry thus broke after the ions migration and redistribution. It is worthwhile to note that the ions usually do not distribute evenly in bulk after the drift process because the mobilities of anions and cations are typically different, e.g., resulting in higher ion concentration on the left in this case. The concentration change thus leads to the OPL change. As depicted in Fig. 2(b), the OPL exhibits a much larger value on the left-hand side as compared to the right due to larger ion density and refractive index on the left (tilted green dotted line). The above describes the principle of detection of refractive-index change and a home-made optical system was designed and built up to realize this concept on investigating the spatial ion profile in LEC specimens.

To read-out the nanoscale OPL variation from ion-migrations, the Grating-PSD combination was utilized. In Fig. 2(c), the LEC cell with uniform anions/ cations distribution has an effective index, n₀. On the other hand, in Fig. 2(d), when only the positive ions exist within the LEC cell, the effective index is reduced to n₀-Δn (Δn is positive). The situation of Fig. 2(d) in the carrier distribution is similar to that on the right side described in Fig. 2(b).

A nanoscale optical ruler which originates from the chromatic aberration of the focusing objective [21,22] was focused near the Surface 2. In Fig. 2(c), the central position of the chromatic nanoscale optical ruler was above the Surface 2 and the corresponding reflection spectrum from Surface 2 is shown in the inset of Fig. 2(c). Then the beam passed through an unmoving pinhole and was diffracted by the grating. The dispersed beam then propagated and was detected by the PSD whose output signal was dependent on the central position of the illumination beam instead of its intensity.

In Fig. 2(d), the change of refractive index, -Δn, in the LEC cell from ion distribution moved the vertical position of the nanoscale optical ruler. The reflected beam with blue-shifted reflection spectrum, as shown in the inset, passed through the same pinhole. The grating transformed the variances of spectral shift into the wave-vector (the direction of beam propagation) changes. The wave-vector-variations were then converted into position differences after propagation. The position differences were then detected by the PSD. By comparing Figs. 2(c) and 2(d), the working principle in this work shows that the combination of Grating and PSD in the PSD-NCM can be used to read the tiny (nanoscale) OPL differences from the migration of ions in the LEC cell.

3.2 Investigation of time-dependent OPL difference

As shown in Fig. 3, while applying a positive electric field pointing to the right (Fig. 3(b)), PF₆^- anions will be attracted by the higher potential side (left) and gradually migrate to the positive electrode over time. On the other hand, the remaining part, Ru(dtb-bpy)₃²⁺, will also move to the right but with a much slower speed due to larger molecular weight. This is a critical process leading to the following electrochemical doping in LEC. The voltage application leads to the ion migration and electrochemical doping, and is typically referred to as “charging” to the LEC devices.

 

Fig. 3. Schematics of cations/ anions distributions before (a) and after (b) applying a bias in the LEC sample based on Ru(dtb-bpy)₃(PF₆)₂. (c) to (h) are the time-dependent OPL difference (ion profile) as a function of fractional distance in y-direction from 0 min to 60 mins. 3D contour of OPL for 10 min and 60 min are shown in the inset of (c) and (h), respectively.

Download Full Size | PDF

The time-dependent OPL difference was successfully captured by the PSD-NCM as shown in Figs. 3(c) to Fig. 3(h). Figure 3(c) shows the ion profiles of the charged LEC with fractional distance (y-direction) presented as the abscissa. The OPL difference was clearly monitored spatially from left to right inside the organic layer between two electrodes. The high-resolution design allows us to record the OPL difference up to ∼5 nm accuracy, which is capable to provide numerous insights of the organic compound during operation [24]. The ion profile before applying an electric field is presented by the red curve, which shows a nearly flat feature. While the charging time started to ramp up, the OPL difference on the left started to rise and the count at the right-end descended as indicated by the red arrows. This result suggests that only ten minutes was already enough to produce a clear ion drift flow in our organic Ru(dtb-bpy)₃(PF₆)₂ thin film. The 3-dimentional (3D) image based on the collection of 2D cross-sectional profiles after the first 10-minute voltage application was also recorded by our system and displayed in the inset of Fig. 3(c). The 3D image shows even more obvious troughs in terms of OPL difference at the right-end of LEC and ridges on the left. This result provides a strong implication that the ionized molecules started to accumulate at the left-end of LEC, which caused the increase of refractive-index followed by OPL rise. In a sharp contrast, an ion-deplete zone at the right-end of LEC formed after bias application, resulting in lower localized refractive-index along with smaller OPL.

It is particularly important to mention that because direct profiling of ion locations in organic layers is very difficult, early studies have only relied on indirect evidence such as the profiles of the electric field.⁶ Those methods provided very limited information about ion concentration which is crucial in studying LECs operation and improving devices performance. More recent reports started to utilize time-of-flight SIMS technique together with sputter bombardment to investigate depth profiling inside the organic films [7,8]. However, positive charges induced on the surface of an ion-included film by the incident sputtering ions could produce large electric field, resulting in ion-migration within the film and interfering the pristine ions from the organic molecules. Considering these issues, we are the first group to propose this 100% optical strategy to investigate the time-dependent ion profile in LECs without invasive ion-sputtering and obtained competitive profiles to the previous literatures.

Ion distributions of each ten-minute time frame were detailedly recorded from the beginning to 60 minutes as shown in Figs. 3(c) to Fig. 3(h). As time went by, the OPL difference at the two ends of LEC became more pronounced, e.g., more than 600 nm OPL difference after 60 minutes. Note that in some profile curves, the oscillating phenomenon could be observed, especially at the left-end. We suspect that this is due to the ununiformity of the organic layer after mass anions accumulation near the positive electrode while the applied bias was high. From the 3D image shown in Fig. 3(h), it also suggests that the OPL difference at the left-end was mostly much higher than at the center or the right-end of the LEC. Individual ridges or aggregations (along x-direction) could be found due to large number of anions accumulating at the left-end, causing exceeded density and uneven ion distributions. Therefore, 3D images were also provided to prevent the misleading localized profile from 2D cross-sectional curves. The details about the ion profiles can be more clearly seen and analyzed in Fig. 4.

 

Fig. 4. (a) OPL difference profile as a function of spatial fraction distance in LEC after applying a 100 V bias for 0, 10 mins, 20 mins, 34 mins, 40mins, 50 mins, and 60 mins. The inset presents the 2D image of the monitored region in the LEC gap. (b) OPL difference on the left, middle, and right of LEC gap as a function of time (under 100 V bias application). The data of these three regions are acquired from the corresponding areas specified in (a).

Download Full Size | PDF

Figure 4(a) concludes the OPL difference of LEC in terms of fractional distance every ten-minute under a 100 V bias. By directly comparing the ion profiles by the 10-min time frame, more insights can be retrieved than the individual profiles in Fig. 3. Ion profile initiated from a flat line at 0-min, gradually increased near the two ends over time, ended at maximum values on both two electrodes at 60-min under bias application. A rapid surge was found at around 34 mins so the curve was used replacing the data of 30 mins. The inset in Fig. 4(a) presents the 2D image of the monitored region in the LEC gap. It is interesting to note that the width of the high-OPL-difference-region on the left is around 15 µm, which is much longer than the 7 µm width on the right. The observation comes from the cation Ru(dtb-bpy)₃²⁺ of our organic material is apparently much heavier than the anion PF₆^-, so the anions were expected to have a much higher mobility followed by more vigorous migration after bias application than the cations. Thus, the accumulation of anions on the left formed a wider zone of higher OPL difference than that of heavier cations. The large amount of anions aggregated at the left-end also raises higher possibility to create ununiform ion distribution, consisting well with the oscillating feature found in Fig. 3. This conjecture also explains why the OPL difference on the right does not show clear oscillating profile. The lower OPL difference on the right is merely due to the lower localized ion concentration after left migrating of anions under applied bias as displayed in Fig. 3(b). In addition, the OPL difference is depicted as a function of time (under bias) in Fig. 4(b). After a bias was applied for few minutes, the OPL difference at both left and right ends showed obvious rise (in absolute value) while the middle part stayed nearly the same. This result is a clear demonstration that the fluctuation in refractive-index profile (or OPL difference) near two electrodes is the consequence of anions migration over time after bias application. From Fig. 4(b) (also Fig. 4(a)), an obvious turning point at 34 mins after bias applied was observed, suggesting the gradual movement of anions to the left firstly arrived the left-end electrode at 34 mins. After the turning point, the ions in our Ru(dtb-bpy)₃(PF₆)₂ then gradually approached a saturation state at the two ends in the following 28 mins.

3.3 Mechanisms of the ion-distribution probing

The key mechanism of our investigation connecting the recorded refractive-index differences and ions migration was resolved. The correlation between the refractive-index change and the ions distribution can be derived from the classical Lorentz oscillation model for dielectric materials. The electric polarization (P) as a function of frequency is given by

 

Fig. 5. Schematic illustration showing the correlation between ion density and refractive-index of materials under bias application. A material with high ion density is shown on the left with high refractive-index while the one with low ion density (low refractive-index) is shown on the right.

Download Full Size | PDF

The above model well connects the observed OPL difference and the ions distribution, which is the key mechanism behind our technique. This model therefore explains the enhancement of OPL on the left-hand side of our sample in both Fig. 3 and Fig. 4 under an applied bias. On the other hand, the OPL became smaller on the right due to the lower cation density and electric susceptibility.

3.4 Discussion and outlook

In this work, the dynamic ion migrations in LEC are taken as an example. In many other optical elements, different optical functions, including the absorption, amplification, wavelength filtering and polarization distinction, are executed by doping different ions within transparent optical substrates. Thus, the presented method can further be used for monitoring the static or dynamic ion distributions in many above-mentioned optical elements, such as laser crystals or polyvinyl alcohol polymer films, for both academic and industrial applications [25]. The bulk refractive index of the emissive layer is valid for the LEC under zero bias. However, after a bias is applied on the LEC, the ion migration results in time-dependent refractive index profile in the active layer, which alters the optical interference effect on the output EL spectrum. As such, the time-dependent refractive index profile is critical for simulating the temporal evolution of output EL spectrum from the LEC. Thus, the proposed technique in this work offers a feasible way to obtain the required refractive index data for optical simulation of LECs.

It is worth noting that the 60-min-experiment was actually an ingenious design. While performing the experiment, the OPL difference data were recorded from the beginning of bias application to 60 minutes, when the memory of our computer was full of vast amount of data. The migrations of cations/anions in the end reached a steady-state at around 55 minutes from the beginning based on Fig. 3(h). The ion-profile only exhibited a very slight change after 55 minutes, indicating the achievement of a steady-state. All the anions eventually migrated to the left electrode as shown in Fig. 3(h), causing the lower refractive-index zone on the right (due to depleted ion density) as well as higher refractive-index part near the left electrode. The OPL response therefore became a fixed profile after the steady-state formation. Also, the LEC gap was always a constant during the experiment. This is the exact reason that we can attribute the OPL difference solely to refractive-index change as explained in section 3.1.

In addition, the phenomenon of anions accumulation at the left-end was carefully investigated. We started from analyzing the 3D and cross-sectional images of 0-minute, which represents the original film roughness profile. Because the organic film was supposed to be uniform at the beginning, the OPL difference profile directly reflected the distance between the sample surface and our detector, revealing the information of film roughness profile. The results are shown in Fig. 6. According to the cross-sectional profile and 3D image of 0-minute shown in Fig. 6(a) and 6(b), certain non-uniformity and roughness of the organic film were observed, which might be due to the uneven glass/organic interface. Particularly, a clear trough between 5–10 fractional distance with a 11.5 nm depth was captured. The uneven surface consequently forms ion traps, reducing the migration speed of anions and causing uneven p-type doping front, which is a commonly seen and acceptable phenomenon in LECs [26–29]. In our case, we believe this trough should be the main reason causing the massive anions accumulation after 40 minutes (Fig. 3(g)) as we mentioned in section 3.2. Over time, the anions will eventually reach the positive electrode under a bias (50 to 60 minutes in our case), forming highest doping concentrations near the anode as well as continuous doping profiles as reported by many LEC literatures [26–29]. It should be emphasized again that the scenario at the right-end is very different from the left. Because the cations are much heavier than anions (PF₆)⁻ and hard to migrate, the lower OPL difference at the right-end is merely due to the lower localized ion concentration after left migrating of anions under applied bias. In addition, no obvious trough was captured at the right-end of our organic film at the beginning. Therefore, the ion accumulation behavior on the right was not monitored by our system.

 

Fig. 6. (a) Cross-sectional profile at 0-minute, illustrating the original film roughness profile. (b) 3D image at 0-minute, revealing certain non-uniformity and roughness of the organic film before 100 V bias application.

Download Full Size | PDF

4. Conclusion

In conclusion, a novel strategy by means of a 100% optical design to observe the spatial ion profile in LEC was firstly proposed in this study. The dynamic time-dependent ion profiles of the organic layer, (Ru(dtb-bpy)₃(PF₆)₂), was completely recorded in 3D images after applying a 100 V bias from 0 min to 60 mins. Clear ion profiles were acquired based on the OPL difference followed by the monitored refractive-index variation while the thickness (pathlength) of LEC layer was fixed. Most importantly, in contrast with previous literatures, we are the first group using a pure optical approach to measure ion profiles in organic layer, avoiding the interferences of complex collision cascade from the bombardment of secondary sputter induced ions which affects local chemical composition and ion profiles of the LEC specimen. Apparent ion migrations were recorded near the corresponding electrodes after 34 mins of bias application. This technique opens up a great avenue to investigate ion distributions inside organic/inorganic materials without damaging or disturbing the samples.