Low-dose measurement of electric potential distribution in organic light-emitting diode by phase-shifting electron holography with 3D tensor decomposition

Microscopy, 2023, 72(6), 485–493 DOI: https://doi.org/10.1093/jmicro/dfad019 Advance Access Publication Date: 28 February 2023 Article Low-dose measurement of electric potential distribution in organic light-emitting diode by phase-shifting electron holography with 3D tensor decomposition Yusei Sasaki 1 1,* , Kazuo Yamamoto1,2,* , Satoshi Anada 2 and Noriyuki Yoshimoto1 Graduate School of Science and Engineering, Iwate University, 4-3-5 Ueda, Morioka, Iwate 020-8551, Japan Abstract To improve the performance of organic light-emitting diodes (OLEDs), it is essential to understand and control the electric potential in the organic semiconductor layers. Electron holography (EH) is a powerful technique for visualizing the potential distribution with a transmission electron microscope. However, it has a serious issue that high-energy electrons may damage the organic layers, meaning that a low-dose EH is required. Here, we used a machine learning technique, three-dimensional (3D) tensor decomposition, to denoise electron interference patterns (holograms) of bilayer OLEDs composed of N,N’-di-[(1-naphthyl)-N,N’-diphenyl]-(1,1’-biphenyl)-4,4’-diamine (𝛼-NPD) and tris-(8-hydroxyquinoline)aluminum (Alq3 ), acquired under a low-dose rate of 130 e− nm−2 s−1 . The effect of denoising on the phase images reconstructed from the holograms was evaluated in terms of both the phase measurement error and the peak signal-to-noise ratio. We achieved a precision equivalent to that of a conventional measurement that had an exposure time 60 times longer. The electric field within the Alq3 layer decreased as the cumulative dose increased, which indicates that the Alq3 layer was degraded by the electron irradiation. On the basis of the degradation of the electric field, we concluded that the tolerance dose without damaging the OLED sample is about 1.7 × 105 e− nm−2 , which is about 0.6 times that of the conventional EH. The combination of EH and 3D tensor decomposition denoising is capable of making a time series measurement of an OLED sample without any effect from the electron irradiation. Key words: electron holography, tensor decomposition, image denoising, low-dose imaging, organic light-emitting diode, organic semiconductor Introduction Electron holography (EH) is a phase imaging technique using the interference phenomenon of coherent electron waves in a transmission electron microscope (TEM). This method enables a quantitative evaluation of electromagnetic fields by analyzing the phase shift of an object wave passing through a TEM sample by it interfering with a reference wave passing through the vacuum [1–3]. The phase shift is obtained from an electron hologram, namely, an interference pattern of electron waves. The relationship between the phase shift and the electrostatic potential of the sample is described by the following equation [4]: Δ𝜙 (𝑥, 𝑦) = 𝐶𝐸 ∫ 𝑉 (𝑥, 𝑦, 𝑧) 𝑑𝑧, (1) where 𝛥𝜙 is the phase shift in the object wave, CE is an interaction constant (6.53 × 106 rad V−1 m−1 at 300 kV) depending on the accelerating voltage of the incident electron beam, and z represents the coordinate along the electron beam direction perpendicular to the x and y planes. To understand the properties of semiconductor devices, accurate measurement of the electronic band structure is important. So far, EH has been used to elucidate the electrical properties of inorganic semiconductors, for example, through quantitative evaluations of the potential map around p–n junctions of Si, GaAs and GaN [5–11]. Meanwhile, few studies have attempted to apply EH to organic semiconductor devices such as organic light-emitting diodes (OLEDs) [12]. Nevertheless, electronic microscopic measurements such as EH will be essential for gaining a better understanding of organic semiconductor devices. Most of the previous studies on OLEDs were carried out using only macroscopic measurements such as displacement current measurements, impedance spectroscopy and sum frequency generation spectroscopy [13–16]. Recently, we applied static EH to an OLED and clearly visualized its nanometer-scale electric potential distribution, in which three different potential regions formed across the organic bilayers [17]. Although we used EH with a conventional electron dose and found that no structural damage occurred during EH, we concluded that lower-dose measurements are probably necessary and the effect of the electron irradiation on the electrical properties should be evaluated more appropriately in order to make reliable characterizations of OLEDs. However, limiting the dose in the EH measurement decreases the signal-to-noise ratio (SNR) in the holograms, leading to a serious deterioration in the precision of the phase measurement. Image processing techniques can provide solutions to the problem of low-dose EH. In particular, machine learning techniques have demonstrated their effectiveness in noise reduction (denoise) for low-dose EH. For example, sparse coding reduces noise in a hologram by representing a target image Received 31 January 2023; Revised 20 February 2023; Editorial Decision 24 February 2023; Accepted 27 February 2023 © The Author(s) 2023. Published by Oxford University Press on behalf of The Japanese Society of Microscopy. All rights reserved. For permissions, please e-mail: 2 Nanostructures Research Laboratory, Japan Fine Ceramics Center, 2-4-1 Mutsuno, Atsuta-ku, Nagoya, Aichi 456-8587, Japan * To whom correspondence should be addressed. E-mail: ; 486 Materials and methods Sample preparation An OLED sample having two kinds of organic layers was prepared by vacuum-heating deposition. Figure 1a schematically shows the device structure of the bilayer OLED sample used for the EH measurement [24]. First, commercially obtained indium-tin oxide (ITO)/glass substrates were cleaned ultrasonically in acetone and isopropyl alcohol and given an ultraviolet/ozone surface treatment. The ITO layer acted as an anode. The ITO/glass substrates were fixed in our specially designed vacuum chamber that has two separate rooms, one for depositing organic layers and the other for depositing metal layers. Next, organic layers, N,N’-di-[(1-naphthyl)-N,N’diphenyl]-(1,1’-biphenyl)-4,4’-diamine (𝛼-NPD) and tris-(8hydroxyquinoline)aluminum (Alq3 ), were deposited at a rate of 0.1–0.2 nm/s on the ITO side of the glass substrate, as shown in Fig. 1a. The film thickness of each layer was 50 nm. The 𝛼-NPD layer acted as a hole transport layer (HTL), and the Alq3 layer acted as both an electron transport layer (ETL) and an emission layer (EML). Finally, an aluminum (Al) cathode layer was deposited at a rate of 0.3–0.4 nm/s for 90 nm in another room partitioned off in the same vacuum chamber. All depositions were carried out in a vacuum of 10−5 Pa order, and the film thickness and deposition rate were controlled by using Fig. 1. (a) Schematic structure of OLED sample composed of ITO/𝛼-NPD/Alq3 /Al. The chemical structures of 𝛼-NPD and Alq (...truncated)