# Fileset

[Cho_IEEE_EDL2026.pdf](https://mdr.nims.go.jp/filesets/7f828c60-302b-4a89-8ab4-7fe79b81bdc5/download)

## Creator

[Hanjun Cho](https://orcid.org/0009-0009-2834-8846), [Masatake Tsuji](https://orcid.org/0000-0002-3404-6037), [Shigenori Ueda](https://orcid.org/0000-0001-9425-0614), [Hideo Hosono](https://orcid.org/0000-0001-9260-6728)

## Rights

[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[Origin of Mobility Reduction in Thin a-IGZO TFTs and Its Improvement](https://mdr.nims.go.jp/datasets/7e9e72c4-3fff-4e0b-b077-0499ee71d17b)

## Fulltext

Origin of Mobility Reduction in Thin a-IGZO TFTs and Its Improvement1394 IEEE ELECTRON DEVICE LETTERS, VOL. 47, NO. 7, JULY 2026Origin of Mobility Reduction in Thin a-IGZOTFTs and Its ImprovementHanjun Cho , Masatake Tsuji , Shigenori Ueda , and Hideo HosonoAbstract— The degradation of field-effect mobility withdecreasing channel thickness in amorphous In-Ga-Zn-Othin-film transistors (a-IGZO TFTs) remains a critical issuewith a controversial origin. Here, we propose that the mobil-ity degradation in very thin a-IGZO TFTs originates from adefective surface layer. Simulated results reveal that thedegradation is primarily governed by the broadening ofunoccupied tail states below the conduction band mini-mum. These large tail state densities arise from enhancedstructural disorder created by sputter-deposition process.As the channel thickness decreases, this defective layerbecomes dominant in carrier transport. The presence ofsurface defect layer was revealed by hard X-ray photo-electron spectroscopy combined with ex situ wet etching.Removal of the surface defective layer by the etching led tothe restoration of the mobility of the 5 nm channel thicknessfrom 13.7 to 28.5 cm2/V·s at 2 MV/cm, which is comparableto the thick TFTs.Index Terms— Amorphous oxide semiconductor, defect,mobility, thin-film transistor.I. INTRODUCTIONAMORPHOUS oxide semiconductors (AOSs), with high-field-effect mobility (µFE) (> 10 cm2/V·s), have becomea de facto standard for flat-panel display backplanes [1],[2]. Recently, the development of high mobility channelmaterials and the reduction of channel thickness (tch) havebeen actively investigated for the high-density scaling innext-generation display and memory applications, includingcurrent-driven organic light-emitting diodes (OLEDs) andcapacitor-free dynamic random-access memory (DRAM) [3],[4], [5]. In particular, as next-generation DRAM employsvertically stacked three-dimensional architectures to achievehigher density, tch below 10 nm are required. However, a severeReceived 6 May 2026; accepted 17 May 2026. Date of publication22 May 2026; date of current version 1 July 2026. This work wassupported in part by the Ministry of Education, Culture, Sports, Sci-ence and Technology (MEXT), Japan, under Grant JPMXP1122683430;in part by JSPS KAKENHI under Grant Number JP23K19266 and GrantNumber JP24K17753; and in part by Samsung Electronics under GrantIO250530-13006-01. The review of this letter was arranged by Editor S.Zhang. (Corresponding authors: Masatake Tsuji; Hideo Hosono.)Hanjun Cho and Masatake Tsuji are with the MDX ResearchCenter for Element Strategy, International Research Frontiers Initia-tive, Institute of Science Tokyo, Yokohama 226-8501, Japan (e-mail:tsuji.m.ac@m.titech.ac.jp).Shigenori Ueda is with the National Institute for Materials Science,Tsukuba 305-0044, Japan.Hideo Hosono is with the MDX Research Center for ElementStrategy, International Research Frontiers Initiative, Institute of Sci-ence Tokyo, Yokohama 226-8501, Japan, and also with the NationalInstitute for Materials Science, Tsukuba 305-0044, Japan (e-mail:hosono.h.aa@m.titech.ac.jp).Digital Object Identifier 10.1109/LED.2026.3695312degradation in µFE and instability in the ultra-thin tch below10 nm hinder their next-generation applications [6], [7]. Pro-posed mechanisms for this thickness-dependent degradationprimarily involve depletion effects induced by acceptor-liketrap states at the back channel [8], as well as reduced dopingefficiency and the formation of deep electron traps [7], [9].Since the degradation has been attributed to surface-relatedtraps, passivation has been commonly employed; however,recent studies show that this approach is ineffective, implyingthat the degradation originates from the inside film rather thanthe surface [6], [10]. Consistently, µFE degradation persistseven in hydroxyl-free IGZO, indicating an intrinsic originof the degradation. The quantum confinement effect has alsobeen considered; however, it is limited to films thinner than3 nm [11]. Recently, it was reported that intrinsic oxy-gen displacements formed in thinner films, which can beregarded as Frenkel defects, deepens the carrier activationenergy and act as carrier traps, leading to serious TFTinstability [6]. Furthermore, such intrinsic structural disor-der induces local electrostatic potential fluctuations, therebyperturbing the conduction band (CB) [12]. The formation ofthese intrinsic defect species strongly correlates with filmdeposition conditions and the chemical composition. On theother hand, strategies to mitigate intrinsic defect speciesgenerated near the surface during film deposition have notbeen established. Moreover, in sputtered AOS films, post-annealing alone cannot effectively suppress this degradation,indicating a fundamental limitation of thermal treatment. Thissuggests that the dominant defects cannot be eliminated bypost-annealing. Consequently, achieving high mobility in ultra-thin films where these intrinsic defects become dominantremains challenging, highlighting the need for alternativestrategies such as direct removal of the defective near-surfaceregion.In this study, we propose a model for µFE degradationand a simple approach to overcome this thickness-dependentdeterioration. We employed a bilayer model in the technol-ogy computer-aided design (TCAD) simulation, consistingof a defective surface and a clean bulk region, consistentwith the defective surface layer observed by depth-resolvedHAXPES. This confirms that the high density of surfacetail states below the CBM is the primary cause of theµFE degradation. It is demonstrated that the removal of thisdefective surface layer via a wet-etching process effectivelysuppresses these defective states near the Fermi level (EF),restoring the µFE of 5 nm tch TFTs to values comparable tothick-channel devices. These results indicate that the intrinsicbehavior of ultra-thin AOS TFTs can be recovered when trapstates originating from the near-surface region are effectivelyeliminated.© 2026 The Authors. This work is licensed under a Creative Commons Attribution 4.0 License.For more information, see https://creativecommons.org/licenses/by/4.0/https://orcid.org/0009-0009-2834-8846https://orcid.org/0000-0002-3404-6037https://orcid.org/0000-0001-9425-0614https://orcid.org/0000-0001-9260-6728CHO et al.: ORIGIN OF MOBILITY REDUCTION IN THIN a-IGZO TFTs AND ITS IMPROVEMENT 1395Fig. 1. Experimental and simulated transfer characteristics of (a) lowPO2 and (c) high PO2, a-IGZO TFTs with varying tch. Insets show theatomic force microscopy images of 5 and 30-nm thick films. Transfercurves were acquired at a drain–source voltage of 0.1 V. (b, d) InputDOS for the surface and bulk regions used in the simulation for the(b) low PO2 and (d) high PO2 in tch = 5 nm TFTs. The dashed linesindicate the equilibrium EF determined by a doping carrier concentrationof 1 × 1017 cm−3, considering the indium-rich composition.II. EXPERIMENTAL DETAILSBottom-gate top-contact TFTs were fabricated on p++-typeSi substrates with a 150 nm-thick SiO2 layer, using a metalshadow mask for active layer patterning to minimize impuritycontamination. a-IGZO (In:Ga:Zn = 60:30:10 at.%) active lay-ers with tch of 5–30 nm were deposited by RF sputtering usinga 3-inch disk target at 150 W, a total pressure of 0.4 Pa, a basepressure of ∼ 2×10−7 Pa, with PO2 = O2/Ar+O2 ranging from0.25 to 25%. Etching was conducted either by a wet processusing a 1.6 wt% tetramethylammonium hydroxide (TMAH)solution at 44 ◦C or by a dry process using Ar+ ion sputtering.Following the etching process, the samples were annealed at400 ◦C for 1 h in air. Next, Ti (5 nm)/Au (60 nm) electrodeswere sputtered for the source/drain contacts. These electrodeswere patterned by photolithography and a lift-off process,defining a channel width/length of 300/30 µm to avoid µFEoverestimation [13], [14]. Finally, a second thermal annealingstep (400 ◦C for 1 h in air) was performed. The degra-dation mechanism was analyzed by 2D TCAD simulations(Silvaco Atlas) employing a bilayer density of states (DOS)model. The DOS parameters for the a-IGZO TCAD modelwere extracted from literature and further calibrated usingtemperature-dependent AC-Hall effect measurements [7], [15],[16]. Depth-dependent electronic structures were examinedusing hard X-ray photoemission spectroscopy (HAXPES) intotal-reflection (TR) [17], [18] and angle-resolved (AR) modesat BL09XU of SPring-8 [19]. AR-HAXPES measurementsat take-off angles (TOAs) of 15◦ and 85◦ provided surface-and bulk- sensitive information with probing depths (3 timesof inelastic mean-free-path of photoelectrons) of 5.7 and19.8 nm, respectively.III. RESULTS AND DISCUSSIONFig. 1a shows that for low PO2 a-IGZO, µFE significantlydecreases from 22.5 to 13.7 cm2/V·s as the tch decreasesfrom 30 to 5 nm. The surface roughness remained low (root-mean-square roughness = 0.4 nm) regardless of tch, excludingmorphological scattering as a dominant factor. Moreover, thedegradation persists even with a passivation layer, indicatingthat its origin lies not in the outermost surface but in thenear-surface region with a finite thickness [6]. To elucidatethe intrinsic structural changes and defect-related degradationmechanism, TCAD simulations were performed. The channellayer was modeled as a bilayer consisting of a bulk regionand a 2 nm-thick defective surface layer, which was experi-mentally eliminated by wet-etching. This model is based onan assumption that a high concentration of oxygen Frenkeldefects is formed during deposition and they remain as aresult of insufficient structural relaxation near the surface.Notably, the formation of Frenkel pairs is experimentallyrecognized in amorphous solids like a-SiO2 because localchemical ordering is preserved in amorphous structure [20].The simulated transfer characteristics are shown as solid linesin Fig. 1a, and the DOSs used in the simulations are presentedin Fig. 1b. The density of surface tail states (Ntail,A) ofCB was increased to reflect the enhanced potential barriernear the CB edge caused by Frenkel-defect-induced structuraldisorder [6]. In addition, the densities of Gaussian-distributedacceptors and donors are simultaneously increased, and theirenergy levels are deepened. Note that these defects are locatedbelow the observed EF at the doping carrier concentration of1 × 1017 cm−3, estimated by Hall effect measurement, andtherefore their contribution to carrier transport is negligiblysmall. For the commercially used In:Ga:Zn= 1:1:1 composi-tion with a low carrier concentration, the EF may be locatedwithin the defect state, potentially causing a threshold voltage(Vth) shift and a large subthreshold swing (SS). Moreover,TFTs fabricated under high PO2 conditions exhibit higher SSvalues compared to those fabricated under low PO2 conditions(0.54 and 0.37 V/dec, respectively, for tch = 5 nm). Basedon these results, the variations in the DOS near the EF underdifferent PO2 conditions are represented in the TCAD modelby adjusting the acceptor-like trap concentration. The originof these traps is assumed to correspond to the concentrationof CB tail states induced by structural disorder, specificallyFrenkel defects. It has been reported that high PO2 inducesnegative ion resputtering during sputtering [8]. We proposethat this resputtering could enhance surface damage, therebyincreasing the concentration of Frenkel defects and resultingin the formation of a structurally disordered, defective surfacelayer. Figs. 1c and 1d present the characteristics of IGZO TFTsdeposited under high PO2 (O-rich) conditions. The surfacedefect density was increased in the simulation (Fig. 1d), repro-ducing a thickness-dependent mobility degradation similar tothat under oxygen-poor conditions, as well as a characteristicVth shift (Fig. 1a). These results strongly suggest that a highdensity of CB tail states localized near the back-channelsurface predominantly governs the µFE degradation in thethinner film.To examine this idea, we examined the electronic structuresof the near-surface and bulk regions of a-IGZO thin films byusing HAXPES combined with TR [17]. Fig. 2a shows thesurface- and bulk-sensitive valence band (VB) spectra of a100-nm-thick a-IGZO film measured with the TR and non-TR conditions, respectively. The surface-sensitive spectrumexhibits an enhanced density of donor states. This observationcan be attributed to an increased fraction of Frenkel defects(structural disorder). Consequently, in thinner films (Fig. 2c),1396 IEEE ELECTRON DEVICE LETTERS, VOL. 47, NO. 7, JULY 2026Fig. 2. (a) Depth-dependent VB HAXPES spectra. The surface- andbulk- sensitive spectra were obtained using TR- and non-TR-HAXPESmodes, respectively. (b, c) Schematics of the surface defective regionand intrinsic bulk region for (b) thick and (c) thin a-IGZO films.Fig. 3. VB HAXPES spectra obtained at TOAs of (a) 15◦ (surface-sensitive) and (b) 85◦ (bulk- sensitive) for a-IGZO thin films before andafter BCE.the volume fraction of this defective surface layer increasesrelative to the thicker films (Fig. 2b), and thus carrier transportis dominated by the surface defective layer.In our model, the origin of the CB tail states in a-IGZOis incorporated as structural disorder perturbed by the Frenkeldefect. Structural disorder in amorphous oxides induces localelectrostatic potential fluctuations, leading to Anderson local-ization and the broadening of unoccupied tail states belowthe CBM. This phenomenon is particularly prominent inthe surface layer, when Frenkel defects are formed duringdeposition and remain structurally unrelaxed. Since these high-density Frenkel defects or strained M–O–M bonds shouldbe chemically unstable and sensitive to etching in amor-phous oxides, we employed a TMAH-based wet solution toselectively remove the defective surface layer [21], [22]. Thenormalized VB spectra before and after the back-channel-etch (BCE) are shown in Figs. 3a and 3b. Notably, a clearreduction in defect states near the EF was observed, while thebulk remains unchanged. This observation can be explainedby the reduction in the donor states density; in this case, thedecrease in fraction of Frenkel defects leads to a suppressionin disorder-induced tail states. Here, the EF is determined tobe 0.05 eV below ECBM, estimated by optical band gap (Eg)(2.85 eV) −EVBM (2.8 eV), which were obtained from Egmeasurements and HAXPES, respectively. Note that thermaldesorption spectroscopy (TDS) measurements confirmed thatH2 and CO2 were not detected, thereby excluding hydrogen-and carbon-related impurity effects.The 5.2 nm tch TFTs showed a significant improvementin µFE after the BCE (Fig. 4a), increasing from 13.7 to28.5 cm2/V·s due to suppression of surface traps. A similarenhancement was also observed for the 15 and 30 nm TFTs(Fig. 4c). This improvement can be attributed to the removalof the defective surface layer, which enhances front-channeltransport by mitigating defect states, even in thicker TFTs.Fig. 4. Improved TFT performance by BCE. (a) Measured transfercharacteristics and —FE of a-IGZO TFTs (tch = 5.2 nm) with and withoutBCE. The BCE device was wet-etched (∼2.4 nm) from 7.6 nm film.(b) Simulated results for 5 nm TFTs. (c) Measured tch dependence of—FE with and without BCE. (d) Measured —FE variation with etchingdepth for wet and dry etching.To examine the origin of the improvement, TCAD simulationswere performed (Fig. 4b). By removing the surface layerwith a high density of tail states from the bilayer model(upper panel of Fig. 1b), the simulations reproduce the restoredtransfer characteristics. This result indicates that the removalof the defective surface layer is the dominant factor in restor-ing the µFE by reducing the concentration of tail states atthe back channel. It is observed that µFE increases as theetching depth approaches ∼2.5 nm but subsequently exhibitsdeterioration (Fig. 4d). This degradation may be attributed tothe stoichiometric imbalance or morphological effect inducedby prolonged selective etching. In contrast, the dry-etchedTFTs exhibited severe degradation, which is attributed to theformation of plasma-induced damage by Ar+ ion sputtering atthe back-channel [23].IV. CONCLUSIONThe defective surface layer was identified as the majororigin of µFE degradation in ultra-thin a-IGZO TFTs. Exper-imental and TCAD analyses revealed that the deteriorationcorrelates with a high density of tail states below the CBM.This finding indicates perturbations of the CB edge, where ahigh density of defects induces potential fluctuations, broad-ening the tail states. Such structural disorder originates frominsufficient relaxation of the as-deposited thin film. Based onthe proposed model, the deterioration in the ultra-thin regimecan be overcome by removing the defective surface region viaa wet-etching process. Consequently, we achieved the 5 nmtch a-IGZO TFT with high µFE of 28.5 cm2/V·s, comparableto that of thick channel devices.ACKNOWLEDGMENTThe HAXPES measurements were performed with theapproval of JASRI/SPring-8 (Proposal Nos. 2025A1748,2025A1524, 2025B1622, and 2025B1875).CHO et al.: ORIGIN OF MOBILITY REDUCTION IN THIN a-IGZO TFTs AND ITS IMPROVEMENT 1397REFERENCES[1] K. Nomura, H. Ohta, A. Takagi, T. Kamiya, M. Hirano, and H. Hosono,“Room-temperature fabrication of transparent flexible thin-film tran-sistors using amorphous oxide semiconductors,” Nature, vol. 432,no. 7016, pp. 488–492, Nov. 2004, doi: 10.1038/nature03090.[2] Y. Kuo, H. Hosono, M. S. Shur, and J. Jang, Oxide Thin Film Transistors.Hoboken, NJ, USA: Wiley, Nov. 2024.[3] S. Samanta, K. Han, C. Sun, C. Wang, A. Kumar, A. V.-Y. Thean, andX. Gong, “Amorphous InGaZnO thin-film transistors with sub-10-nmchannel thickness and ultrascaled channel length,” IEEE Trans. ElectronDevices, vol. 68, no. 3, pp. 1050–1056, Mar. 2021, doi: 10.1109/TED.2020.3048920.[4] J. Y. Kim, H. Kim, D. Kim, and H. W. Jang, “Advancements ofamorphous IGZO-based transistors: Materials, processing, and devices,”ACS Appl. Electron. Mater., vol. 7, no. 11, pp. 4703–4728, Jun. 2025,doi: 10.1021/acsaelm.5c00605.[5] A. Belmonte, H. Oh, N. Rassoul, G. L. Donadio, J. Mitard, H. Dekkers,R. Delhougne, S. Subhechha, A. Chasin, M. J. van Setten, L. Kljucar,M. Mao, H. Puliyalil, M. Pak, L. Teugels, D. Tsvetanova, K. Banerjee,L. Souriau, Z. Tokei, L. Goux, and G. S. Kar, “Capacitor-less, long-retention (>400s) DRAM cell paving the way towards low-power andhigh-density monolithic 3D DRAM,” in IEDM Tech. Dig., Dec. 2020,pp. 28.2.1–28.2.4, doi: 10.1109/IEDM13553.2020.9371900.[6] H. Cho, M. Tsuji, S. Ueda, J. Kim, and H. Hosono, “Oxygen defects andinstability in very thin a-IGZO TFTs,” Adv. Electron. Mater., vol. 11,no. 15, p. e00349, Sep. 2025, doi: 10.1002/aelm.202500349.[7] L. Shao, K. Nomura, T. Kamiya, and H. Hosono, “Operation charac-teristics of thin-film transistors using very thin amorphous In–Ga–Zn–Ochannels,” Electrochemical Solid-State Lett., vol. 14, no. 5, p. H197,May 2011, doi: 10.1149/1.3555070.[8] T.-H. Chiang, B.-S. Yeh, and J. F. Wager, “Amorphous IGZO thin-film transistors with ultrathin channel layers,” IEEE Trans. ElectronDevices, vol. 62, no. 11, pp. 3692–3696, Nov. 2015, doi: 10.1109/TED.2015.2478700.[9] M. Kimura, T. Nakanishi, K. Nomura, T. Kamiya, and H. Hosono,“Trap densities in amorphous-InGaZnO4 thin-film transistors,” Appl.Phys. Lett., vol. 92, no. 13, Mar. 2008, Art. no. 133512, doi: 10.1063/1.2904704.[10] M. J. Kim, H. J. Park, S. Yoo, M. H. Cho, and J. K. Jeong,“Effect of channel thickness on performance of ultra-thin body IGZOfield-effect transistors,” IEEE Trans. Electron Devices, vol. 69, no. 5,pp. 2409–2416, May 2022, doi: 10.1109/TED.2022.3156961.[11] K. Abe, K. Nomura, T. Kamiya, and H. Hosono, “Optical evi-dence for quantization in transparent amorphous oxide semiconductorsuperlattice,” Phys. Rev. B, Condens. Matter, vol. 86, no. 8, Aug. 2012,Art. no. 081202, doi: 10.1103/physrevb.86.081202.[12] T. Kim, H. Jang, B. Park, S. Shin, M. A. Alam, P. D. Ye, andC. Jeong, “Engineering amorphous IGZO thin-film transistors: The roleof composition and channel thickness in mobility-threshold voltageoptimization,” ACS Omega, vol. 10, no. 50, pp. 62182–62188, Dec.2025, doi: 10.1021/acsomega.5c09705.[13] S. Kim, Y. Lee, S. Shin, H. Kim, H. Shin, H. Hosono,C. Jeong, and J. Kim, “Mobility overestimation in thin-film transis-tors: Effects of device geometry and fringe currents,” ACS Nano,vol. 19, no. 41, pp. 36614–36625, Oct. 2025, doi: 10.1021/acsnano.5c11904.[14] T. Takahashi and Y. Uraoka, “Intrinsic field-effect mobility in thin-filmtransistor with polycrystalline In2O3 channel based on transfer lengthmethod,” Appl. Phys. Exp., vol. 18, no. 1, Jan. 2025, Art. no. 014001,doi: 10.35848/1882-0786/ada19e.[15] M. M. Billah, M. D. H. Chowdhury, M. Mativenga, J. G. Um,R. K. Mruthyunjaya, G. N. Heiler, T. J. Tredwell, and J. Jang, “Analysisof improved performance under negative bias illumination stress ofdual gate driving a-IGZO TFT by TCAD simulation,” IEEE ElectronDevice Lett., vol. 37, no. 6, pp. 735–738, Jun. 2016, doi: 10.1109/LED.2016.2557358.[16] K. Abe, A. Sato, K. Takahashi, H. Kumomi, T. Kamiya, andH. Hosono, “Mobility- and temperature-dependent device modelfor amorphous In–Ga–Zn–O thin-film transistors,” Thin SolidFilms, vol. 559, pp. 40–43, May 2014, doi: 10.1016/j.tsf.2013.11.066.[17] S. Ueda, “Depth-resolved electronic structure measurements by hardX-ray photoemission combined with X-ray total reflection: Directprobing of surface band bending of polar GaN,” Appl. Phys. Exp.,vol. 11, no. 10, Oct. 2018, Art. no. 105701, doi: 10.7567/apex.11.105701.[18] S. Ueda, Y. Fujita, and Y. Sakuraba, “Near-interface electronic andmagnetic states of insulator/Co2MnSi heterostructures probed by hardX-ray photoemission combined with X-ray total reflection,” Phys. Rev.B, Condens. Matter, vol. 109, no. 8, Feb. 2024, Art. no. 085109, doi:10.1103/physrevb.109.085109.[19] A. Yasui, Y. Takagi, T. Osaka, Y. Senba, H. Yamazaki, T. Koyama,H. Yumoto, H. Ohashi, K. Motomura, K. Nakajima, M. Sugahara,N. Kawamura, K. Tamasaku, Y. Tamenori, and M. Yabashi, “BL09XU:An advanced hard X-ray photoelectron spectroscopy beamline of SPring-8,” J. Synchrotron Radiat., vol. 30, no. 5, pp. 1013–1022, Sep. 2023,doi: 10.1107/s160057752300629x.[20] H. Hosono, Y. Ikuta, T. Kinoshita, K. Kajihara, and M. Hirano, “Physicaldisorder and optical properties in the vacuum ultraviolet region ofamorphous SiO2,” Phys. Rev. Lett., vol. 87, no. 17, Oct. 2001, Art. no.175501, doi: 10.1103/physrevlett.87.175501.[21] H. Hosono, “Fourier transform infrared attenuated total reflection spec-tra of ion-implanted silica glasses,” J. Appl. Phys., vol. 69, no. 12,pp. 8079–8082, Jun. 1991, doi: 10.1063/1.348925.[22] Y. Shi, Y.-S. Shiah, K. Sim, M. Sasase, J. Kim, and H. Hosono,“High-performance a-ITZO TFTs with high bias stability enabledby self-aligned passivation using a-GaOx,” Appl. Phys. Lett.,vol. 121, no. 21, Nov. 2022, Art. no. 212101, doi: 10.1063/5.0123253.[23] G. W. Mattson, K. T. Vogt, J. F. Wager, and M. W. Graham, “Illuminatingtrap density trends in amorphous oxide semiconductors with ultrabroad-band photoconduction,” Adv. Funct. Mater., vol. 33, no. 25, Jun. 2023,Art. no. 2300742, doi: 10.1002/adfm.202300742.http://dx.doi.org/10.1038/nature03090http://dx.doi.org/10.1109/TED.2020.3048920http://dx.doi.org/10.1109/TED.2020.3048920http://dx.doi.org/10.1021/acsaelm.5c00605http://dx.doi.org/10.1109/IEDM13553.2020.9371900http://dx.doi.org/10.1002/aelm.202500349http://dx.doi.org/10.1149/1.3555070http://dx.doi.org/10.1109/TED.2015.2478700http://dx.doi.org/10.1109/TED.2015.2478700http://dx.doi.org/10.1063/1.2904704http://dx.doi.org/10.1063/1.2904704http://dx.doi.org/10.1109/TED.2022.3156961http://dx.doi.org/10.1103/physrevb.86.081202http://dx.doi.org/10.1021/acsomega.5c09705http://dx.doi.org/10.1021/acsnano.5c11904http://dx.doi.org/10.1021/acsnano.5c11904http://dx.doi.org/10.35848/1882-0786/ada19ehttp://dx.doi.org/10.1109/LED.2016.2557358http://dx.doi.org/10.1109/LED.2016.2557358http://dx.doi.org/10.1016/j.tsf.2013.11.066http://dx.doi.org/10.1016/j.tsf.2013.11.066http://dx.doi.org/10.7567/apex.11.105701http://dx.doi.org/10.7567/apex.11.105701http://dx.doi.org/10.1103/physrevb.109.085109http://dx.doi.org/10.1107/s160057752300629xhttp://dx.doi.org/10.1103/physrevlett.87.175501http://dx.doi.org/10.1063/1.348925http://dx.doi.org/10.1063/5.0123253http://dx.doi.org/10.1063/5.0123253http://dx.doi.org/10.1002/adfm.202300742