Online ISSN 2092-7592 Print ISSN 1229-7607 Transactions on Electrical and Electronic Materials https://doi.org/10.1007/s42341-020-00275-z REGULAR PAPER Physical Modeling on Effective Traps Density Near the Conduction Band Dependence of Electrical Characteristics of Amorphous Indium Gallium Zinc Oxide Thin‑Film Transistors Abdelhafid Marroun1 · Naima Amar Touhami1 · Taj‑eddin El Hamadi1 Received: 10 August 2020 / Revised: 12 November 2020 / Accepted: 21 December 2020 © The Korean Institute of Electrical and Electronic Material Engineers 2021 Abstract An (a-IGZO TFT) is modeled through experimental-based (a-IGZO TFTs) using (TCAD) simulator. A parametric study is performed on the numerical fit of the designed (a-IGZO TFT) current–voltage (I/V) characteristics, to obtain the near conduction band defects parameters optimal values, and to investigate the effect of the near band defects caused by oxygen vacancies on the (a-IGZO TFTs) output parameters. A new model approach is proposed for simulating (a-IGZO) electrical properties. The proposed model is)known as a density of state models (DOS), and ( ) it is composed of two principal compo( nents, conduction band tail gAct (E) and Gaussian distributed donor-like gD . The study of the presented (DOS) models (E) G is based on both conduction band tail elements that are known as tail acceptor density ( gta ) as well as tail acceptor energy ( Ea ), and Gaussian distributed donor-like elements which are donor gaussian energy ( ED ) and donor gaussian distribution ( ggd ). Results show that the tail acceptor states defects ( gta , Ea ) near the conduction band is the cause of the mobility and gaussian donor distribution degradation near the conduction band, and it has a major impact on changes that occur in the subthreshold region data [threshold voltage ­(Vth), subthreshold swing (SS), and on-state/off-states current ratio ­(Ion/Ioff)]. Keywords Numerical simulation · Thin film transistors TFTs · a-IGZO TFT · Electrical characteristics · Defects 1 Introduction Amorphous Indium-Gallium-Zinc-Oxide (a-IGZO) is considered as one of the most important amorphous oxide semiconductors (AOS) because of its large electron mobility. It became the center of attention of a group of studies that adapt amorphous Indium-Gallium-Zinc-Oxide (a-IGZO) thin film transistors (TFTs) as the next generation of (TFT) technology (high-resolution active-matrix organic lightemitting diode displays (AMOLEDs) [1, 2], liquid crystal displays (AMLCD), RFID Tags, and sensors [3, 4] …) because of its outstanding characteristics of a field-effect mobility µeff ranging from 8 to 20 cm2/Vs, a threshold voltage ­(Vth) around 0 V, an off current (­ Ioff) below 1E-12 A, high on-state/off-states current ratio (­ Ion/Ioff) ≥ 1E6, and a subthreshold swing (SS) below 0.2 V/Dec [5–8]. The * Abdelhafid Marroun [email protected] 1 Faculty of Science, Abdelmalek Essaadi University, M’Hannech II, 93000 Tétouan, Morocco most important characteristic that plays an important role in developing this type of technology is the stability of (a-IGZO TFTs) that is defined by the shift of subthreshold region data (threshold voltage ­(Vth), subthreshold swing (SS) and on-state/off-states current ratio ­(Ion/Ioff)) [9–11]. Hence, many types of research study the factors that affect the changes in the subthreshold region data. Among the possible causes of these changes are oxygen vacancies near the conduction band minimum (CBM) states [12–14]. Refer to some studies her. The numerical simulation is an essential and necessary tool for the understanding of the device physics and the (TFT) operation principles. However, there are barely any publications that discuss the numerical simulation of the (a-IGZO TFT) even with the incredible advances in device fabrication. In this paper, A design of (a-IGZO TFT) is proposed and studied. This study is based on the optimization of the current–voltage (I/V) characteristics of the (a-IGZO TFT) numerical fit. Thereafter, a more accurate modeling approach which is named density-of-state (DOS) models is proposed for simulating (a-IGZO) electrical properties using a two-dimensional (2D) technology computer-aided 13 Vol.:(0123456789) Transactions on Electrical and Electronic Materials design (TCAD) simulator. These models contain two com(ponents ) near the conduction band minimum (CBM) states, gAct (E) and ( gD (E) ) that are known as conduction bandG tail and gaussian-distributed donor-like, respectively. So, to investigate the (DOS) components effect on the subthreshold region data ((Vth), (SS), and (­ Ion/Ioff)). A study of the model tail acceptor density ( gta ) and tail acceptor energy ( Ea ) for the conduction band-tail (gAct (E)) on one hand, and donor Gaussian energy ( ED) and donor Gaussian distribution (ggd ) for the gaussian-distributed donor-like ( gD (E)) on the other G hand, is required since the mathematical (DOS) models are an exponential function. 2 Experimental Methods The schematic structure of the (a-IGZO TFT) is a bottomgate staggered structure. The device is composed of a heavily doped (n + +) silicon wafer that acts as a gate, and a 100 nm of ­(SiO2) layer placed on top of it that is used as a gate insulator. Also, a 20 nm thick (a-IGZO) active layer was deposited on the unheated (­ SiO2) layer. Finally, the (Au/Ti) stacked layer 40∕5 nm thick was deposited as the source and drain electrodes. Noting that the channel width (W) and length (L) are 180 μm, 30 μm, respectively. For the operation of a thin film transistor, the following standard Eqs. (1)–(3) are used [15, 16]. The threshold voltage ­(Vth) and field-effect mobility (µsat) were extracted from the expression of the operation of a field-effect transistor, which is a function of the capacitance and applied voltage: IDS = ( ) W Cox 𝜇sat VGS − Vth , L VDS ≥ VDS − Vth (1) where ­Ids is the drain current, (W) is the channel width, (L) is the channel length, (­ Vgs) is the gate voltage, (­ Vds) is the drain bias voltage, ­(Cox) is the capacitance per unit area of the gate insulator, and ­(Vth) and (µsat) are the threshold voltage and Saturation mobility that were extracted from the linear fit of Eq. 1. Subthreshold swing (SS) is a gate voltage required for the decade increase in drain current at a constant drain voltage. Typically, the (SS) of an (a-IGZO TFT) is in the range of 0.2–0.5 V/decade , and it can be further reduced to be lower than 0.1 V/decade by, for example, reducing the channel bulk trap density, applying high-K dielectrics, or adopting fully depleted states. Subthreshold swing is obtained from the equation given below: SS = 𝜕Vgs || ( ) |Vds = Constant 𝜕 logIds || 2 𝜇sat ⎡ √ ⎤ ⎢ d Ids 1 ⎥ =⎢ � ⎥ Ci W ⎥ ⎢ dVgs 2L ⎦ ⎣ (3) 2.1 2D Numerical Simulation The two-dimensional cross-section along the channel of the (a-IGZO TFT) structure used in this work is shown in Fig. 1. It is designed to match with the actual (TFT) used in this study. The structure consists of a 20 nm thick (a-IGZO) channel layer, a heavily doped n-type poly-silicon substrate that also acts as a gate, a 100 nm thick gate insulator ­(SiO2) layer, and drain and source Ohmic contacts that are the low resistance (Au/Ti) 40∕5 nm thick stacked layer. The channel width (W) is 180 μm, and the channel length (L) is 30 μm, respectively. Contacts between (S/D) electrodes and the (a-IGZO) layer are both assigned as Schottky in this work. The (S/D) metal work function (𝜙m)= 4.33eV) and the electron affin( ity of (a-IGZO) 𝜒a−IGZO were included in the calculation. Both thermionic ( emission ) and tunneling current were considered . 𝜒a−IGZO is estimated from a simple linear relation between electron affinities of its three elementary compounds. ) ( ) ( ) ( 𝜒a−IGZO = a 𝜒In2 O2 + b 𝜒Ga2 O3 + c 𝜒ZnO (4) ( ) where a, b, and c are molar percentages. 𝜒a−IGZO was calculated to be 4.16 eV. Throughout this paper, the Schottky contact model is used as a default in numerical simulation. The channel is made of (a-IGZO) which is an amorphous n-type semiconductor. Disordered materials [like a-Si and (a-IGZO)] contain a large number of defect states continuously distributed within the bandgap of the material. The (2) The (SS) was also extracted from the subthreshold region data at the maximum slope point using Fig. 4 transfer 13 characteristics in the logarithmic scale. Saturation mobility is defined based on how quickly an electron can move through the active layer, which is obtained by the following equation: Fig. 1 Structure of an a-IGZO TFT in a bottom gate (top-contact) staggered configuration Transactions on Electrical and Electronic Materials density of states is a combination of exponentially decaying band tail states and Gaussian distributions of mid-gap states. The conduction band-tail (CB) states and the valence bandtail (VB) states are given by exponential function decay: ) ( E − Ec A gct (E) = gta exp (5) Ea ) ( Ev − E = gtd exp (6) Ed ( ) where (gta ) and gtd (cm−3 eV −1) are the effective density at (Ec ), (Ea ) is the characteristic slope energy of the conduction ( ) band-tail states, and Ev and (Ed ) are the characteristic slope energy of the valence band-tail states. In addition to tail states, Gaussian-distributed donor-like ( ) A and acceptor-like defect states, gD and , respectively (g ) G G are also considered in the energy gap. ( ( )2 ) E − E D gD (7) G (E) = ggd exp − 𝜎D2 gD vt (E) gAG (E) ( ( = gga exp − )2 ) EA − E 𝜎A2 (8) ( ) ( ) ( ) where ggd and gga are the total density (cm−3 eV(−1), )𝜎D and (𝜎A ) are the standard deviation, and (ED ) and EA are the peak energy of the Gaussian distribution. Usually, the tail state ( Dby donor ) ((a-IGZO) ) density of gap states is formed g gD , donor Gaussian distribution with a )maxi(E) (E) vt G ( mum located at 2.9 eV, and acceptor tail state gAct (E) . Figure 2. Shows the different components of the density of states in (a-IGZO), and their values are presented in Table 1. 3 Results and Discussion Figures 3 and 4 show the optimized fittings to the experimental output and the transfer characteristics of a typical depletion-type (a-IGZO TFT). As depicted in these figures, the output characteristics of (a-IGZO TFTs) is performed for different voltage values ((Vgs ) from 4 to 20 V) and ­(Vds = 0.1 V). Results show that the threshold voltage (­ Vth), the on-current (­ Ion), and the sub-threshold swing (SS) are 1.5 V, 3.8E − 7A, and 0.19 V/ decade, respectively. According to the accuracy of the (2D) simulation results, a study of the influence of each densityof-state (DOS) parameter is taken into account. this study is based on the evaluation of the tail acceptor states effect that is defined by its energy decay ( )(Ea ) and ( )its density (gta ). So, by examining the effect of Ea and gta on the transfer characteristics and the output parameters, only one (DOS) Fig. 2 Proposed DOS model for a-IGZO Table 1 The physical parameters of the a-IGZO TFT used in this work Symbol ( ) −3 Nc cm ( ) Nv cm−3 Eg (eV) 𝜒(eV) 𝜀 ( ) gta cm−3 eV−1 ( ) gtd cm−3 eV−1 ( −3 ) gd cm eV−1 Ea (meV) Ed (meV) ( ) 𝜇n cm2 ∕Vs ( 2 ) 𝜇p cm ∕Vs Description Value Effective DOS in the conduction band 5E18 Effective DOS in the valence band 5E18 Bandgap Electronic affinity Permittivity The density of tail states at E = Ec 3.05 4.16 10 1.55E20 The density of tail states at E = Ev 1.55E20 The peak of the donor-like Gaussian states Conduction-band-tail slope Valence-band-tail slope Free electron mobility 6.5E16 Free hole mobility 0.1 13 120 15 parameter is varied, while the other parameters are kept the same as those listed in Table 1. ( ) Figures 5 and 6 displays tail acceptor density gta and tail acceptor energy (Ea ) of the acceptor tail state, which ranges from 0.7E20 to 2.5E20 and 0.005 to 0.016, respectively. Based on ( these ) figures, it is obvious that the effect of each of gta and Ea does not appear on the subthreshold region data, since the sub-threshold swing (SS) and the threshold voltage ­(Vth)keep the same values which are 1.5 V, and 0.19 V/ decade, respectively. However, according to these previous various values, we notice a great effect in terms of mobility and ­(Ids)on ­(Vgs = 20 V). According to saturation mobility (µsat) values presented in Eq. (6), and constructed from 13 Transactions on Electrical and Electronic Materials Fig. 3 Output characteristics of the experimental a-IGZO TFT along with optimized fittings of simulation Fig. 5 a-IGZO TFT simulated linear region (­Vds = 0.1 V) transfer characteristics curves in logarithm scale and Square root scale for various tail acceptor density ( gta) Real experimental data also showed for reference Fig. 4 Transfer characteristics of the experimental a-IGZO TFT along with optimized fittings of simulation in linear and logarithmic scale Fig. 6 a-IGZO TFT simulated linear region (­Vds = 0.1 V) transfer characteristics curves in logarithm scale and Square root scale for various tail acceptor energy ( Ea) Real experimental data also showed for reference both Figs. 5 and 6, more specifically from the square root of the drain ( )current ( ­()Ids) plot, we note that the higher is the value of gta and Ea , the lower is the mobility. This decline could be due to the unrest created by oxygen disorder, which determines the nature of defects near the Conduction Band Minimum (CBM) states. The peak energy of donor Gaussian distribution (ED) and total density of donor Gaussian distribution ( ggd ) parameters of the Gaussian effect on the transfer characteristics and output parameters are examined. Figure 7 shows that the proposed (DOS) models for (a-IGZO), near the Conduction Band Minimum (CBM) for peak energy of donor Gaussian distribution ( ED ) are varied from 1.8 to 2.9 eV, while donor Gaussian distribution ( ggd ) are kept fixed at 6.5E16 cm−3 eV −1. Based on the results shown in Fig. 8 that illustrates the (a-IGZO TFT) simulated linear region ­(Vds = 0.1 V) transfer characteristics in both logarithm and Square root scale for different values of donor gaussian distribution peak energy, we observe that the position of donor Gaussian distribution affects the (TFT) transfer characteristics, since a slight change is observed at the subthreshold region data, while 13 Transactions on Electrical and Electronic Materials Fig. 7 The Proposed DOS model for a-IGZO for various peak energy of donor Gaussian distribution ( ED) Fig. 8 The a-IGZO TFT simulated linear region (­Vds = 0.1 V) transfer characteristics curves in logarithm scale and Square root scale for various peak energy of donor Gaussian distribution ( ED) a great change has occurred on the donor Gaussian distribution [peak energy of donor Gaussian distribution ( ED )] position. Figure 9 shows that the proposed (DOS) models for (a-IGZO), near the Conduction Band Minimum (CBM) for peak energy of donor Gaussian distribution (ED) are kept fixed at 2.9 eV, while donor Gaussian distribution (ggd) are varied from 3.5E16 cm−3eV−1 to 7E17 cm−3eV−1. The simulation results of the total density donor Gaussian distribution ( ggd ) dependence on transfer characteristics is represented in Fig. 10. As noticed, when donor-like Gaussian density ( ggd ) traps Fig. 9 The Proposed DOS model for a-IGZO for the various total density of donor Gaussian distribution ( ggd) Fig. 10 a-IGZO TFT simulated linear region (­Vds = 0.1 V) transfer characteristics curves in logarithm scale and Square root scale for the various total density of donor Gaussian distribution ( ggd) are increased, the (­ Ids/Vgs) is shifted to negative gate bias direction. And it also tends to be a conductor. However, when donor-like Gaussian density attains 7E17 cm−3 eV −1 , the simulated transfer characteristics curves that correspond to the experimental one. To sum up, we can conclude that the tail acceptor states defects near the conduction band is the reason behind the mobility as well as the gaussian donor distribution near conduction band degradation, which has by its turn a major impact on changes that occur in the subthreshold region data [threshold voltage ­(Vth) and subthreshold swing (SS)]. 13 Transactions on Electrical and Electronic Materials 4 Conclusions This work is described in detail, the design and the optimization of an (a-IGZO TFT) numerical fit outputs characteristic which is based on experimental data using the well-known (TCAD) simulator. Furthermore, this paper is reported the investigation of the near band defects effect on the designed (a-IGZO TFT) current–voltage (I/V) characteristics, by proposing a novel modeling approach. This later is called the density of state models (DOS) and its study is based on its two principal components elements which are tail acceptor density ( gta ) and tail acceptor energy ( Ea ) on one hand, and donor gaussian energy ( ED ) and donor gaussian distribution ( ggd ) on the other hand. Results prove that both, tail acceptor density and energy, ( gta ) and ( Ea ), respectively; contribute to the degradation of both mobility and gaussian donor distribution, which leads to the subthreshold region data [threshold voltage ­(Vth), subthreshold swing (SS), and on-state/off-states current ratio ­(Ion/Ioff)] changes. Data Availability The data that support the findings of this study are available from the corresponding author upon reasonable request. References 1. Y. Wang et al., Integrated a-IGZO TFT gate driver with programmable output for AMOLED display. Dig. Tech. Pap. SID Int. Symp. 49(1), 1377–1380 (2018) 2. Y. Chen, D. Geng, J. Jang, Integrated active-matrix capacitive sensor using a-IGZO TFTs for AMOLED. IEEE J. Electron Devices Soc. 6(1), 214–218 (2018) 3. M. H. Hung et al., Ultra low voltage 1-V RFID tag implement in a-IGZO TFT technology on plastic, in 2017 IEEE Int. Conf. RFID, RFID 2017, (2017), pp. 193–197 4. J.Y. Pyo, W.J. Cho, In-plane-gate a-IGZO thin-film transistor for high-sensitivity pH sensor applications. Sensors Actuators, B Chem. 276(August), 101–106 (2018) 5. R.N. Bukke, C. Avis, M.N. Naik, J. Jang, Remarkable increase in field effect mobility of amorphous IZTO thin-film transistors with purified ZrOx gate insulator. IEEE Electron Device Lett. 39(3), 371–374 (2018) 6. K. Ebata, S. Tomai, Y. Tsuruma, T. Iitsuka, S. Matsuzaki, K. Yano, High-mobility thin-film transistors with polycrystalline In-Ga-O channel fabricated by DC magnetron sputtering. Appl. Phys. Express 5(1), 1–4 (2012) 13 7. A. Marroun, N.A. Touhami, T.E. El Hamadi, M. El Bakkali, Highperformance indium–gallium–zinc oxide thin-film transistors based on anodic aluminum oxide. Procedia Manuf. 32, 729–733 (2019) 8. S. Hu et al., Effect of ITO serving as a barrier layer for Cu electrodes on performance of a-IGZO TFT. IEEE Electron Device Lett. 39(4), 504–507 (2018) 9. J.Y. Noh, D.M. Han, W.C. Jeong, J.W. Kim, S.Y. Cha, Development of 55” 4 K UHD OLED TV employing the internal gate IC with high reliability and short channel IGZO TFTs. J. Soc. Inf. Disp. 26(1), 36–41 (2018) 10. A.J. Kronemeijer et al., Dual-gate self-aligned IGZO TFTs monolithically integrated with high-temperature bottom moisture barrier for flexible AMOLED. Dig. Tech. Pap. SID Int. Symp. 49(1), 1577–1580 (2018) 11. R. Yao et al., High-performance flexible oxide TFTs: optimization of a-IGZO film by modulating the voltage waveform of pulse DC magnetron sputtering without post treatment. J. Mater. Chem. C 6(10), 2522–2532 (2018) 12. K. Nomura, T. Kamiya, E. Ikenaga, H. Yanagi, K. Kobayashi, H. Hosono, Depth analysis of subgap electronic states in amorphous oxide semiconductor, a-In-Ga-Zn-O, studied by hard x-ray photoelectron spectroscopy. J. Appl. Phys. 109(7), 073726 (2011) 13. W.H. Han, Y.J. Oh, K.J. Chang, J.S. Park, Electronic structure of oxygen interstitial defects in amorphous In–Ga–Zn–O semiconductors and implications for device behavior. Phys. Rev. Appl. 3(4), 1–8 (2015) 14. S. Choi et al., Effect of oxygen content on current stress-induced instability in bottom-gate amorphous InGaZnO thin-film transistors. Materials (Basel) 12(19), 3149 (2019) 15. T. Kamiya, K. Nomura, H. Hosono, Present status of amorphous In-Ga-Zn-O thin-film transistors. Sci. Technol. Adv. Mater. 11(4), 044305 (2010) 16. A. Marroun, N. A. Touhami, T.-E. El Hamadi, and M. El Bakkali, High-performance indium-gallium-zinc oxide thin-film transistors based on anodic aluminum oxide, in Procedia Manufacturing, vol. 32 (2019) 17. J. Wu, Y. Chen, D. Zhou, Z. Hu, H. Xie, C. Dong, Sputtered oxides used for passivation layers of amorphous InGaZnO thin film transistors. Mater. Sci. Semicond. Process. 29, 277–282 (2015) 18. D. S. Software, Atlas User’ s Manual, no. 408. (2014) 19. M. Adaika, A. Meftah, N. Sengouga, M. Henini, Numerical simulation of bias and photo stress on indium-gallium-zinc-oxide thin film transistors. Vacuum 120, 59–67 (2015) Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.