Tailor-made nanostructures bridging chaos and order for highly efficient white organic light-emitting diodes

Organic light-emitting diodes (OLEDs) suffer from notorious light trapping, resulting in only moderate external quantum efficiencies. Here, we report a facile, scalable, lithography-free method to generate controllable nanostructures with directional randomness and dimensional order, significantly boosting the efficiency of white OLEDs. Mechanical deformations form on the surface of poly(dimethylsiloxane) in response to compressive stress release, initialized by reactive ions etching with periodicity and depth distribution ranging from dozens of nanometers to micrometers. We demonstrate the possibility of independently tuning the average depth and the dominant periodicity. Integrating these nanostructures into a two-unit tandem white organic light-emitting diode, a maximum external quantum efficiency of 76.3% and a luminous efficacy of 95.7 lm W−1 are achieved with extracted substrate modes. The enhancement factor of 1.53 ± 0.12 at 10,000 cd m−2 is obtained. An optical model is built by considering the dipole orientation, emitting wavelength, and the dipole position on the sinusoidal nanotexture.

In this work, the authors report some interesting results on highly efficient OLED leveraging a new out-coupling scheme. It is well-known that today OLED can have an internal quantum efficiency approaching 100%. However, the external quantum efficiency is usually 20 to 40% (a recent paper demonstrated an external efficiency of >50%), due to the photon trapping effect. Some of the photons are trapped by total internal reflection, and some are trapped by the waveguiding effect. In this work, the authors demonstrate an external efficiency above ~70% by using a pre-patterned substrate with nanostructures. Overall the results are quite sound. However, I do have a few comments regarding the performance of the OLEDs. 1. The reported efficiency is plotted in Fig. 6b. It seems that the external efficiency varies quite significantly. The lowest is about 45% and the highest is about 70%. What is the cause of this large variation? Is the reported ultrahigh efficiency reproducible and uniform across a large area? 2. The nanostructures are patterned using RIE. Those nanostructures are seemed to be very random. I am wondering how reproducible the nanostructures are. If devices are produced by multiple runs, do they exhibit similar performance? 3. Finally, I suggest the authors to use a table to summarize the external quantum efficiency reported by a few top groups and companies and this work. Then the advantage of this reported outcoupling scheme can be clearly seen.

Reviewer #2 (Remarks to the Author):
This manuscript reports on the fabrication of random nanostructures and their use as an internal light-extraction layer in OLEDs. The authors present a recipe how to modify the average "periodicity" and depth of the nanostructures over a relatively large range. They furthermore present a simulation-based analysis of the light extraction capability of such a layer. And, they apply it to tandem white OLEDs to demonstrate the improved light outcoupling experimentally. Technically, this is a sound paper and there is not so much to criticize about the contents. Just a few issues should be addressed: -The formation of random nanostructures and their use in OLEDs was demonstrated many years before (see e.g. Ref. 19 of the manuscript). And there is even older work demonstrating that thin films can spontaneously form such patterns upon inducing morphological instability by certain means. Thus, the main claim here is to have a new, technically easier recipe to achieve this. -The reported external quantum efficiency of 74% is achieved using a bulky macro-extractor, which is not consistent with the thin appearance of OLEDs. And one has to keep in mind that the device is a tandem OLED, where the EQE is effectively doubled. This means that compared to reported values on single-emission unit OLEDs with light extraction schemes preserving the thinfilm device nature, the EQE of this work is nothing exceptional. -The way the manuscript is written is not always appealing. I suggest shifting some of the more technical stuff to the SI.
Reviewer #3 (Remarks to the Author): How to enhance the outcoupling efficiency in white OLEDs is an important topic in OLED lighting field. As we know, OLEDs suffer from notorious light trapping, resulting in that the efficiency of the resulting OLEDs can be further enhanced, hindering its application. In this paper, the authors developed a facile, scalable and lithography-free method to generate controllable nanostructures with directional randomness and dimensional order by reactive ions etching (RIE) to enhance the outcoupling efficiency of the fabruicated white OLEDs. As we see, about 1.3 times enhancement is obtained. They gave the detailed results about experimental conditions that are used to obtain the nanostructure, and optical simulation. The method to obtain the nanostructures is significant and useful and the results are also encouraging. However, the obtained nanostrucutres still exist some problems that need to be considered by the authors. 1. As given, the nanostrucures is directional randomness. This means that the repeatability can not be guaranteed. If so, the simulation are also unreliable. 2. As we see, the fabricated OLEDs exhibit large leakage current due to the large surface roughness, which will greatly influence device efficiency and stability, causing poor applicability. 3. The 1.3 times enhancement in efficiency is not high as internal outcoupling structure. 4. White OLEDs as lighting sources in practical applications are required to have big emission area. Can the nanostructures given in this papaer keep the uniformity and repeatability in big area?

Response to Reviewers' Comments
We sincerely thank the three reviewers for their thoughtful comments and constructive suggestions for improving the manuscript. Responses to each of the comments are At the end of this response letter, we have added a paragraph in purple, discussing a formal change that is not in connection with the reviewers' comments.

Reviewer #1:
In this work, the authors report some interesting results on highly efficient OLED Thank you for this comment. Please note that the devices giving ~45% external quantum efficiency (EQE) are references without nanostructure (the value is so high, because they are tandem structures with up to potentially 200% internal quantum efficiency). The devices with nanostructures N1-N5 yield higher EQEs compared to the reference, but the absolute EQE is dependent on the specific nanostructure, as shown in Figure 6b in the main text. The highest EQE ~70% can be obtained for devices with the nanostructures N2 and N5. The efficiencies are obtained for tandem white OLEDs with an attached half-sphere lens to extract substrate modes, so actually it is quite reasonable, which is also pointed out by the 2 nd reviewer (see below).
The efficiency is reproducible since we observed several samples with the highest EQE ~ 70%. The summary about the efficiency of these samples are presented in the supplementary information Figure S8 in the revised version.
With respect to the question of uniformity, we added supportive data to demonstrate the very good uniformity of the nanostructures generated on the PDMS surface. The uniformity investigation of the nanostructures is done by AFM measurements on different positions (at least four positions) randomly chosen for each sample, followed with a statistical analysis to check the variation of the obtained values measured at each position. As shown in Figure S1, the periodicity distribution of a specific structure (N1) is almost the same for all these measurements. The dominant periodicity for N1 is 245.5 nm. There is negligible deviation of the average depth at different positions as shown in Figure S1. This originates from the intrinsic difference at different positions and the experimental deviation for each AFM measurement.  Table S1 and Table S2   The uniformity investigation is now also added to the supplementary information as Figure S2, Table S1 and Table S2 and explaining paragraph in the revised version.

Reviewer #1's Comment #2:
The nanostructures are patterned using RIE. Those nanostructures are seemed to be very random. I am wondering how reproducible the nanostructures are. If devices are produced by multiple runs, do they exhibit similar performance?

Authors' response to Reviewer #1's Comment #2:
This is a very good comment. The reproducibility and uniformity have been paid special attention in the early stage of this study.
We have checked on the experimental repeatability of these nanostructures by using a tracking sample made with the same fabrication recipe for each run of RIE treatment.
Further this data was collected to check the setup stability.
As presented in Figure 1 in the revised manuscript and Figure S1 in the supplementary information, the nanostructure is randomly orientated on the surface of PDMS after RIE treatment, but statistically, there is a distribution of the periodicity and depth for each nanostructure. The nanostructure is a system bridging chaos and order as a quasiperiodic pattern. According to AFM measurements, one can obtain the statistical properties including the periodicity, depth distribution and average depth. The experimental repeatability is monitored by measuring the quantitative parameters of periodicity and depth for the nanostructure generated in different batches with the same recipe. As shown in Figure S2 in this response letter, for nanostructures generated in multiple runs, the periodicity is peaking at ~350 nm. Only very small deviation of the average depth can be noted, as shown in Figure S2b. Therefore, the method we presented in this study to generate nanostructures is controllable with good repeatability. Now we add this data set as Figure S3 and an explaining paragraph in the supplementary information.

Reviewer #1's Comment #3:
Finally, I suggest the authors to use a table to summarize the external quantum efficiency reported by a few top groups and companies and this work. Then the advantage of this reported outcoupling scheme can be clearly seen.

Authors' response to Reviewer #1's Comment #3:
This is a very good comment. Thank you very much. Having this included will put our study in better context.  table in the supplementary information as Table S6 (named as Table S3 in this response letter), to compare the external quantum efficiency (EQE) of white OLEDs and the enhancement factor from different studies. As shown in Table S3   nanometer to micrometer scales. The important finding is that process parameters can be used to deterministically fabricate tailored structure features -so the 'control' aspect of the process is the key here. An additional benefit is that this process is indeed, as Reviewer #2 noted, the speed and simplicity of the process. Specifically, it is possible to independently control the periodicity and depth, as illustrated in Figure 2 in the main text and Figure S4 in the supplementary information. The detailed analysis is summarized in the main text of the manuscript in the results part (section: nanostructure generation and characterization). In the end, we investigate the mechanism of nanostructure generation and control, which is presented in the main text (section: mechanism of the nanostructure control). To sum up, we presented a facile, scalable, lithography-free, reproducible and controllable method to fabricate nanostructures with sizes ranging from nanometers to micrometers.

Reviewer #2's Comment #2:
The reported external quantum efficiency of 74% is achieved using a bulky macroextractor, which is not consistent with the thin appearance of OLEDs. And one has to keep in mind that the device is a tandem OLED, where the EQE is effectively doubled.

This means that compared to reported values on single-emission unit OLEDs with light
extraction schemes preserving the thin-film device nature, the EQE of this work is nothing exceptional.

Authors' response to Reviewer #2's Comment #2:
The investigation in this study is based on a highly optimized tandem white OLED. In theory, the maximum internal quantum efficiency for the reference should be 200%.
According to the simulation of the outcoupling efficiency for the planar tandem OLED

Reviewer #2's Comment #3:
The way the manuscript is written is not always appealing. I suggest shifting some of the more technical stuff to the SI.

Authors' response to Reviewer #2's Comment #3:
Thank you very much for this feedback, which we believe it is mostly with respect to the detailed description of the RIE process and its variations on the produced structures.
Here, we would like to not follow the suggestion of Reviewer #2, to shift some of this description to the SI, as we strongly believe that these details are vital in differentiating our work from earlier reports on nanostructure generation. Hence, we decided to keep our manuscript structure as it is.

Authors' response to Reviewer #3's general comment:
Thanks to Reviewer #3 for the positive feedback. Just to clarify the key findings in this manuscript, we presented a facile, scalable, lithography-free and controllable method to generate controllable quasi-periodic nanostructures, bridging a chaotic and ordered system. When embedded in a highly optimized tandem white OLEDs, it is possible to enhance the external outcoupling efficiency with substrate modes from 44.4 ± 3.3% to 69.0 % at 10,000 cd m -2 , giving an enhancement factor of 1.53 ± 0.12.

Reviewer #3's Comment #1:
As given, the nanostructures is directional randomness. This means that the repeatability can not be guaranteed. If so, the simulation are also unreliable.

Authors' response to Reviewer #3's Comment #1:
Thank you for pointing us -similar to Reviewer #1 -to the uniformity and repeatability of the reported nanostructure generation. The experimental repeatability is of vital importance for every study and the reproducibility of nanostructures has been paid special attention in this work in the early stage.
As discussed in the main text in the results part (section: nanostructure generation and characterization), the geometry of the nanostructures is randomly orientated on the surface of PDMS after the RIE-treatment. Statistically, the dimensional parameters including the periodicity and depth is quasi-periodic with a distribution that can be measured by AFM. The experimental repeatability is monitored by measuring the periodicity distribution and average depth for the nanostructure generated in different batches with the same recipe as a tracking sample.
As shown in Figure S2 in this response letter (see above), in multiple batches from different time, the deviation of the dominant periodicity and the average depth of the nanostructure (tracking sample) generated with the same recipe is very small. For tracking samples generated in multiple runs, the periodicity is peaking at about 350 nm.
Only very small deviation of average depth can be noted, as shown in Figure S2b. The method we demonstrated here is facile and controllable, with good experimental repeatability.
We made a detailed statistical analysis about the uniformity and repeatability for the nanostructure at different positions. This has been added to the revised manuscript.
Please refer to the reply to Reviewer #1's Comment #2. To be short, the nanostructure is randomly oriented on the surface. However, regarding the quantitative parameters such as periodicity and depth, it is uniform statistically on the entire surface. Hence, when comparing to a simplified and fixed model description, we can guarantee that each individual structure can be seen fixed and reproducible for a given processing This statement has now been added into the main text in the revised version.

Reviewer #3's Comment #2:
As we see, the fabricated OLEDs exhibit large leakage current due to the large surface roughness, which will greatly influence device efficiency and stability, causing poor applicability.
Authors' response to Reviewer #3's Comment #2: Thank you for this important comments.
The leakage current is about 1-2 orders higher compared to our previous reports with the same device structure on commercial ITO, as stated in the main text in results part  The nanostructures generated on the surface of PDMS is uniform in this study. This is confirmed by measuring randomly chosen positions on one sample and checking the deviation. As shown in Figure S1 in this response letter, the periodicity distribution of a specific structure (N1) is almost the same for all these measurements at different  Table S1 and Table S2 in this letter. According to the statistical analysis, the deviation of the proposed aspect ratio (AR) for these nanostructures at different position is very small, indicating that the nanostructures is uniform at different positions among the entire surface.
The experimental repeatability of these nanostructures has been demonstrated, by using a tracking sample for each run of RIE treatment and is illustrated in Figure S2 in this response letter. As shown in Figure S2, for nanostructures generated in multiple runs, the periodicity is peaking at ~350 nm. Only very small deviation of the average depth can be noted, as shown in Figure S2b. Therefore, the method we presented in this study to generate nanostructures is controllable with good repeatability. Now we add these data into the supplementary information as Figure S2.

Additional revision, not based on the reviewers' comments:
In the original version of the manuscript, the term 'extraction efficiency' was used to quantify the ability of the nanostructures to couple out trapped photons. In an independent work of our group, we have recently established a formalism to characterize the outcoupling efficiency in connection with the respective layer architecture of the OLEDs. This formalism takes possible optical limitations of specific devices into account. We have originally called the corresponding quantity 'extraction efficiency' as it was used here, but we have been convinced during the peer review process of this work, that the use of 'extraction efficiency' would cause ambiguities. To follow this reasoning and avoid ambiguities, we have changed in this current revision from the term 'extraction efficiency' to 'efficiency of light outcoupling structures'. The procedure to calculate it and the reported results remain unchanged.