Characterization of Single Defects in Ultrascaled MoS2 Field-Effect

5 days ago - MoS2 has received a lot of attention lately as a semiconducting channel material for electronic devices, in part due to its large band ga...
4 downloads 0 Views 2MB Size
Subscriber access provided by UNIV OF NEWCASTLE

Article 2

Characterization of Single Defects in Ultra-Scaled MoS Field-Effect Transistors Bernhard Stampfer, Feng Zhang, Yury Yuryevich Illarionov, Theresia Knobloch, Peng Wu, Michael Waltl, Alexander Grill, Joerg Appenzeller, and Tibor Grasser ACS Nano, Just Accepted Manuscript • DOI: 10.1021/acsnano.8b00268 • Publication Date (Web): 07 Jun 2018 Downloaded from http://pubs.acs.org on June 12, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

Characterization of Single Defects in Ultra-Scaled MoS2 Field-Effect Transistors Bernhard Stampfer,∗,† Feng Zhang,‡ Yury Yuryevich Illarionov,†,¶ Theresia Knobloch,† Peng Wu,‡ Michael Waltl,† Alexander Grill,† Joerg Appenzeller,∗,‡ and Tibor Grasser∗,† Institute for Microelectronics (TU Wien), Gusshausstrasse 27–29, 1040 Vienna, Austria, Purdue University, 1205 West State Street, West Lafayette, Indiana 47907, USA, and Ioffe Physical-Technical Institute, Polytechnicheskaya 26, 194021 St-Petersburg, Russia E-mail: [email protected]; [email protected]; [email protected]

Abstract

contain random telegraph noise, which is due to the transfer of charge between the channel of the transistors and individual defects, visible only due to the large impact of a single elementary charge on the local electrostatics in these small devices. Using hidden Markov models for statistical analysis, we extract the charge capture and emission times of a number of defects. By comparing the bias-dependence of the measured capture and emission times to the prediction of theoretical models, we provide simple rules to distinguish oxide traps from adsorbates on these back-gated devices. In addition, we give simple expressions to estimate the vertical and energetic positions of the defects. Using the methods presented in this work, it is possible to locate the sources of performance and reliability limitations in 2D devices and to probe defect distributions in oxide materials with 2D channel materials.

MoS2 has received a lot of attention lately as a semiconducting channel material for electronic devices, in part due to its large band gap as compared to other 2D materials. Yet, the performance and reliability of these devices is still severely limited by defects which act as traps for charge carriers, causing severely reduced mobilities, hysteresis and long-term drift. Despite their importance, these defects are only poorly understood. One fundamental problem in defect characterization is that due to the large defect concentration only their average response to bias changes can be measured. Based on such averaged data, a detailed analysis of their properties and identification of particular defect types is difficult. In order to overcome this limitation, we here characterize single defects on MoS2 devices by performing measurements on ultra-scaled transistors (∼65×50 nm) which contain only few defects. These single defects are characterized electrically at varying gate biases and temperatures. The measured currents

Keywords random telegraph noise, MoS2 , SiO2 , transistor, single defects, charge trapping, dichalcogenides



To whom correspondence should be addressed † Institute for Microelectronics (TU Wien), Gusshausstrasse 27–29, 1040 Vienna, Austria ‡ Purdue University, 1205 West State Street, West Lafayette, Indiana 47907, USA ¶ Ioffe Physical-Technical Institute, Polytechnicheskaya 26, 194021 St-Petersburg, Russia

ACS Paragon Plus Environment

1

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Molybdenum disulfide (MoS2 ) is one of the most promising semiconducting transition metal dichalcogenides which is considered for applications in beyond-CMOS electronic devices. The direct electronic bandgap of singlelayer MoS2 can be as high as 2.6 eV, 1,2 which makes this material an interesting candidate for digital device applications. In addition, there is some understanding that MoS2 field-effect transistors (FETs) are suitable for high-frequency applications 3 and circuit integration. 4–6 In particular, numerous realizations of MoS2 FETs can be found in recent literature reports. 7–14 However, these studies mostly deal with the analysis of the performance of these devices, as well as the exploration of fabrication techniques allowing them to realize their theoretical performance potential predicted by simulations. 15 One of the most important performance limitations is due to defects which can arise from both non-optimized device processing and fundamental properties of insulators with their unique defect bands. 16–18 These defects can exchange charges with the channel, thus strongly affecting the device performance and reliability. 18,19 Although these problems are not often discussed in the current 2D literature, they are of utmost importance for industrial integration of these new technologies and the transition from device prototypes to commercial 2D devices. The reliability of previously investigated MoS2 FETs, as well as of 2D devices in general, is typically reduced by charge trapping in oxide traps 20–23 with very broad distributions of time constants 24 and trapping sites on top of the channel, e.g. adsorbates and water molecules. 19,25 Charge trapping decreases the mobility, and results in a hysteresis of the gate transfer characteristics, 18,19,25–28 as well as in partially recoverable threshold voltage shifts which appear whenever a bias is applied to the gate, 10,19,29,30 conventionally known as bias-temperature instabilities (BTI). 31–36 Although some understanding of these issues for different 2D technologies has been achieved in our previous works, 18–20,23 all these studies have been performed on large area device prototypes with micrometer dimensions. Since the

Page 2 of 14

defect densities in these devices are considerable (we obtained values of ∼1012 cm−2 in our studies mentioned before, which corresponds to 106 traps for e.g. a 10 µm × 10 µm device), the observed drifts of the transistor characteristics are the collective response of a large number of defects. However, a considerably improved understanding of the physical mechanisms of charge trapping and the nature of the involved traps can be obtained via the characterization of individual defects. Such a characterization is possible in ultra-scaled devices which contain merely a few defects within the channel area. 36 As the channel area of a device is reduced, the number of defects in the devices decreases, but at the same time their individual influence on the channel of the device increases. 37 This leads to an increase in the noise level of the drain current as compared to bigger devices. In particular, charging and discharging of single defects can be observed as steps in the drain current. This issue is known as random telegraph noise (RTN) 38 and can be observed only in scaled devices. Statistical characterization of RTN signals caused by every particular defect allows to extract the capture and emission time constants (τc and τe , respectively), trap level, activation energy and vertical position of this defect, 39 thus giving insights into the understanding of the microscopic defect properties. Typically, RTN analysis is possible if only a handful of defects are present. This implies that for a given defect density D, the device area has to be scaled A = 5/D. Based on our previous studies, where we extracted a D of about 1012 /cm2 , an area of 50 nm × 50 nm would be required to reach this single defect limit (see Figure S1 in the Supporting Information (SI)). Therefore we fabricated MoS2 FETs and report sub-100 nm devices which contain only individual discrete defects within the channel area. By performing statistical analysis of a number of RTN traces measured on different devices, we extract defect levels and vertical positions of these defects and discriminate between oxide traps and adsorbates on top of the MoS2 channel. This presents a considerable break-

ACS Paragon Plus Environment

2

Page 3 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Paragon Plus Environment

Page 4 of 14

Page 5 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

increases as the energy level of the defect shifts below the Fermi level. Once in its charged state, the defect causes a threshold voltage shift due to the presence of the electric charge close to the channel, affecting its conductivity. From silicon devices it is known that defects are also visible in stationary drain current measurements if they are close to the Fermi level at the applied gate bias. As such a defect randomly charges and discharges, the measured drain current shows discrete steps to a lower or higher value respectively, an effect called RTN. 38 A number of RTN traces with defects behaving in a different manner are shown in Figure 2bd. The time constants of the defects depend on their energetic and spatial position relative to the channel, the temperature, and the number of carriers in the channel. Thus, extracting their time constants under various conditions allows us to acquire important information on the defects and their surroundings. In Figure 2b we show a defect with time constants that are independent on gate bias. In contrast, Figure 2c shows a defect with a strong gate bias dependence. At a gate bias of −5 V, the bias dependent defect is mostly unoccupied, as indicated by the higher current state. Already at −4 V, the drain current is mostly in the low state, as the average capture time decreases and the emission time increases. The bias range where the defect shows RTN corresponds to the bias range where trapping is visible in the ID –VG measurements, although the range might be shifted depending on the sweep rate and direction of the ID –VG measurement. This microscopic behavior is what is visible in larger devices as hysteresis, caused by the superpositioned effects of many different traps. A more complicated case called anomalous RTN (aRTN) is shown in Figure 2d, where normal RTN behavior is interrupted by random periods of inactivity.

extract the capture and emission time for each bias point is to measure the duration of the high- and low-current states, and average the recorded times as indicated in Figure 2b. This method, however, is quite susceptible to the noise observed on the very low currents measured and typically only successful for simple RTN signals with a single active trap. To extract the time constants of the defects in our more complicated signals, we used hidden Markov models (HMMs) and the BaumWelch algorithm, which are commonly used also in speech processing and the analysis of DNA. 43 This method uses a Markov chain as a model for the defect, as shown in the insets of Figure 2b and d. The algorithm then iteratively optimizes the time constants of the model to most likely represent the measured data (see the methods section for details). This allowed us to extract the time constants of a number of defects. Four representative examples, termed A to D, are shown in Figure 3: Three of the defects (A, C, D) showed normal RTN behavior which can be described by a two-state Markov model. In contrast, defect B showed random periods of inactivity (Figure 2d), indicating that it has an additional meta-stable state between the stable charged and discharged states. This is called aRTN and is known from measurements in Si/SiO2 devices. 44 Anomalous RTN is usually explained by a configurational change at the defect site. 36 Of the four defects presented, two show a bias dependence of their characteristic capture and emission times, while the other two exhibit no gate bias dependence. The latter is not observed in Si/SiO2 devices and most likely due to adsorbates on top of the device. For the interpretation of the gate bias dependence we start with a trap located in the oxide between the backgate and the channel (i.e. an oxide defect) or just under the channel (dangling bonds or processing residue). Applying a gate bias will result in a change of the energetic position of the trap relative to the energies of the carriers in the channel, which will in turn affect the time constants. A defect located within the MoS2 (e.g. an S vacancy), at the edge of the channel (e.g. etching induced

Extraction of Time Constants In order to gain more insight into the physical properties of the defects, it is necessary to extract the time constants from the measured traces. The most straight forward method to

ACS Paragon Plus Environment

5

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Paragon Plus Environment

Page 6 of 14

Page 7 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Paragon Plus Environment

Page 8 of 14

Page 9 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

Table 1: Extracted trap parameters, calculated with the analytical expressions 1 and 2 (left), and from TCAD device simulation (right). Energies are relative to the conduction band edge Type Defect 16 . 25 26 28 a

(A)b (B) (C) (D)

Approximation E1′ ,approx zapprox 0.003 eV . 0.010 eV .−0.018 eV −0.843 eV

−0.8 nm ≈8.0 nm ≈8.0 nm −3.2 nm

E1 a

ε1 ′ 1 a

— −0.149 eV — —

— 0.99 eV — —

Fit Parameter E 1′ R1 ′ 2 — −0.017 eV −0.016 eV −0.640 eV

— 0.80 0.69 0.52

S1′ 2 ~ω — 1.60 1.38 0.33

zsimulated — 8.5 nm 8.0 nm −1.7 nm

Parameters for three-state defects only b No TCAD fit

Ar: 10 sccm; pressure: 3 Pa; RF source power: 50 W; RF bias power: 50 W; time: 17 s). Next, Ni (65 nm) electrodes were deposited acting as source/drain contacts. This step defined the channel length (L ≈ 50 nm) of the back gated three-terminal structures. Flake thicknesses varied between 5 nm and 15 nm.

drain current in dynamic (ID –VG ) and static (RTN) measurements, and extracted time constants from the measured traces. We found that part of the defects in these devices show time constants independent of gate bias, unlike the defects commonly found in silicon devices. We argue that this is likely due to the position of these defects in the devices, and give equations to estimate spatial and energetic positions. We compared the Shockley-Read-Hall and non-radiative multi phonon trap models to find that the simple SRH model is unable to explain the temperature and bias dependences of the measured data and – just like in Si technologies – the NMP model should be used. Using detailed TCAD simulations we have extracted physical parameter sets of the measured defects. From the obtained defect parameters, we made claims as to the physical nature of the studied defects. Using the presented methods, it is possible to pinpoint defects in 2D-devices, and thus learn more about the feasibility of certain oxide and channel material combinations for microelectronic devices.

Measurement Basic electrical characterization (Figure 1bc) of the devices was performed at roomtemperature in air using a commercial parameter analyzer (Agilent 4156C). RTN measurements were performed in the vacuum chamber of a LakeShore probestation which is able to reach a minimum pressure of 5 × 10−6 torr. The vacuum chamber was connected to a dewar with liquid nitrogen, while the temperature was varied between 100 K and 373 K and controlled by a LakeShore temperature controller. Electrical defect characterization was performed using measurement equipment built at TU Wien, which allows to perform a wide range of current measurements down to the microsecond range. 50 Depending on the time constants of the observed defects, measurement time intervals for each single trace were varied from 0.1 s for faster traps to 1000 s for slower traps, with the number of measurement points adjusted accordingly. In order to obtain enough data for statistical analysis, at each bias point we measured 50 – 100 traces for smaller time intervals and 5 – 20 traces for larger time intervals. Where possible, defects were characterized at different temperatures, typically ranging within an 40 – 80 K interval.

Methods Fabrication MoS2 (SPI Supplies) flakes were exfoliated onto a 20 nm silicon dioxide (SiO2 ) substrate with underlying highly doped silicon using standard scotch tape techniques. The 20 nm SiO2 layer was used in the following as the gate dielectric. After the flake transfer, e-beam lithography was used to define the channel width (W ≈ 65 nm) of the active device region, followed by plasma dry etching (SF6 : 10 sccm;

ACS Paragon Plus Environment

9

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Parameter extraction

Page 10 of 14

ting parameters. Following this, we used the NMP trap model together with calibrated device structures to perform the single-trap simulations of the capture and emission times. Further details of the simulations are discussed in the supporting information.

To extract the capture and emission times of individual defects from the recorded data, we statistically analyzed traces in the bias range where RTN was visible. For each bias point and temperature we trained HMMs composed of two- or three state traps to the traces with software developed at TU Wien using the BaumWelch algorithm. 43 The Baum-Welch algorithm is used on the drain current measurements to give the most likely set of parameters (transition times, step heights) for the measured data given the Markov model of the trap. The Markov model itself consists of a number of nodes (the hidden states of the defect, e.g. 1, 1′ , 2), the allowed transitions between these states (1 ⇌ 1′ ⇌ 2) and the observables of the hidden states (e.g. states 1, 1′ → high current level, state 2 → low current level, measurement noise σ).

Acknowledgement The authors thank for the financial support through the FWF grant n◦ I2606-N30. Supporting Information Available: A PDF file is available with additional information on: Single-defect limit for different 2D technologies; SEM image of a device; Dependence of capture and emission times on trap position; Device modeling; Adsorbates on the MoS2 surface; Estimation of trap position and energy; Detailed results for individual defects; This material is available free of charge via the Internet at http://pubs.acs.org/.

References

Simulations

1. Klots, A.; Newaz, A.; Wang, B.; Prasai, D.; Krzyzanowska, H.; Lin, J.; Caudel, D.; Ghimire, N.; Yan, J.; Ivanov, B. Probing Excitonic States in Suspended TwoDimensional Semiconductors by Photocurrent Spectroscopy. Sci. Rep. 2014, 4, 6608.

The current flowing in a thin-body Schottkybarrier transistor, in the below or near threshold regime – which is where the defects in our study are located – is mainly determined by the transmission of the Schottky barriers at the source and drain contacts. 41 Thus, to model the characteristics of these devices and to translate the gate voltage axis into an energy scale, we used the analytical Schottky-barrier model from Penumatcha et al. 51 as the basis to extract the relevant device parameters and band movement from the measured ID –VG curves. Moreover, since the transmission probability from source to drain for the ultra-short channels used in this study is impacted little by scattering inside the MoS2 channel, the only additional correction to the above model in the device onstate made here is through the introduction of the density of states (DOS) dependent quantum capacitance CQ . 52 In this way not only a more precise extraction of energy levels of the defects near threshold is achieved, but this approach also eliminates the need to speculate about the details of the impact of scattering inside the FET channel and introduces no additional fit-

2. Rasmussen, F.; Thygesen, K. tional 2D Materials Database: Structure of Transition-Metal genides and Oxides. J. Phys. 2015, 119, 13169–13183.

ComputaElectronic DichalcoChem. C

3. Krasnozhon, D.; Lembke, D.; Nyffeler, C.; Leblebici, Y.; Kis, A. MoS2 Transistors Operating at Gigahertz Frequencies. Nano Lett. 2014, 14, 5905–5911. 4. Radisavljevic, B.; Whitwick, M.; Kis, A. Integrated Circuits and Logic Operations Based on Single-Layer MoS2 . ACS Nano 2011, 5, 9934–9938. 5. Wang, H.; Lili, Y.; Lee, Y.-H.; Shi, Y.; Hsu, A.; Chin, M.; Li, L.-J.; Dubey, M.; Kong, J.; Palacios, T. Integrated Circuits Based on Bilayer MoS2 Transistors. Nano Lett. 2012, 12, 4674–4680.

ACS Paragon Plus Environment

10

Page 11 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

6. Wachter, S.; Polyushkin, D.; Bethge, O.; Mueller, T. A Microprocessor Based on a Two-Dimensional Semiconductor. arXiv.org, e-Print Arch., Condens. Matter 2016, 1612.00965.

Characterization of Top-Gated Molybdenum Disulfide Field-Effect-Transistors with High-k Dielectrics. Microelectron. Eng. 2017, 178, 190–193. 15. Yoon, Y.; Ganapathi, K.; Salahuddin, S. How Good Can Monolayer MoS2 Transistors Be? Nano Lett. 2011, 11, 3768–3773.

7. Radisavljevic, B.; Radenovic, A.; Berivio, J.; Giacometti, V.; Kis, A. Single-layer MoS2 Transistors. Nat. Nanotechnol. 2011, 6, 147–150.

16. Degraeve, R.; Cho, M.; Govoreanu, B.; Kaczer, B.; Zahid, M.; Van Houdt, J.; Jurczak, M.; Groeseneken, G. Trap Spectroscopy by Charge Injection and Sensing (TSCIS): A Quantitative Electrical Technique for Studying Defects in Dielectric Stacks. IEEE Int. Electron Devices Meet. 2008; pp 1–4.

8. Fiori, G.; Szafranec, B.; Iannaccone, G.; Neumaier, D. Velocity Saturation in FewLayer MoS2 Transistor. Appl. Phys. Lett. 2013, 103, 233509. 9. Lee, G.-H.; Yu, Y.-J.; Cui, X.; Petrone, N.; Lee, C.-H.; Choi, M.; Lee, D.-Y.; Lee, C.; Yoo, W.; Watanabe, K. Flexible and Transparent MoS2 Field-Effect Transistors on Hexagonal Boron Nitride-Graphene Heterostructures. ACS Nano 2013, 7, 7931– 7936.

17. Franco, J.; Kaczer, B.; Eneman, G.; Roussel, P.; Grasser, T.; Mitard, J.; Ragnarsson, L.; Cho, M.; Witters, L.; Chiarella, T. Superior NBTI Reliability of SiGe Channel pMOSFETs: Replacement Gate, FinFETs, and Impact of Body Bias. IEEE Int. Electron Devices Meet. 2011; pp 18–5.

10. Cho, K.; Park, W.; Park, J.; Jeong, H.; Jang, J.; Kim, T.-Y.; Hong, W.-K.; Hong, S.; Lee, T. Electric Stress-Induced Threshold Voltage Instability of Multilayer MoS2 Field Effect Transistors. ACS Nano 2013, 7, 7751–7758.

18. Illarionov, Y.; Knobloch, T.; Waltl, M.; Rzepa, G.; Posphischil, A.; Polyushkin, D.; Furchi, M.; Mueller, T.; Grasser, T. Energetic Mapping of Oxide Traps in MoS2 Field-Effect Transistors. 2D Mater. 2017, 4, 025108.

11. Lopez-Sanchez, O.; Lembke, D.; Kayci, M.; Radenovic, A.; Kis, A. Ultrasensitive Photodetectors Based on Monolayer MoS2 . Nat. Nanotechnol. 2013, 8, 497–501.

19. Illarionov, Y.; Rzepa, G.; Waltl, M.; Knobloch, T.; Grill, A.; Furchi, M.; Mueller, T.; Grasser, T. The Role of Charge Trapping and MoS2 /SiO2 and MoS2 /hBN Field-Effect Transistors. 2D Mater. 2016, 3, 035004.

12. English, C.; Shine, G.; Dorgan, V.; Saraswat, K.; Pop, E. Improved Contacts to MoS2 Transistors by Ultra-High Vacuum Metal Deposition. Nano Lett. 2016, 16, 3824–3830.

20. Illarionov, Y.; Smith, A.; Vaziri, S.; Ostling, M.; Mueller, T.; Lemme, M.; Grasser, T. Bias-Temperature Instability in Single-Layer Graphene Field-Effect Transistors. Appl. Phys. Lett. 2014, 105, 143507.

13. Nourbakhsh, A.; Zubair, A.; Sajjad, R.; Tavakkoli KG, A.; Chen, W.; Fang, S.; Ling, X.; Kong, J.; Dresselhaus, M.; Kaxiras, E. MoS2 Field-Effect Transistor with Sub-10 nm Channel Length. Nano Lett. 2016, 16, 7798–7806.

21. Guo, Y.; Wei, X.; Shu, J.; Liu, B.; Yin, J.; Guan, C.; Han, Y.; Gao, S.; Chen, Q. Charge Trapping at the Mo2 -SiO2 Interface and its Effects on the Characteristics

14. Bolshakov, P.; Zhao, P.; Azcatl, A.; Hurley, P.; Wallace, R.; Young, C. Electrical

ACS Paragon Plus Environment

11

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

of MoS2 Metal-Oxide-Semiconductor Field Effect Transistors. Appl. Phys. Lett. 2015, 106, 103109.

Page 12 of 14

29. Park, W.; Lee, Y.; Kim, J.; Lee, S.; Kang, C.; Cho, C.; Lim, S.; Jung, U.; Hong, W.; Lee, B. Reliability Characteristics of MoS2 FETs. Ext. Abstr. Conf. Solid State Devices Mater. 2013; pp 684–685.

22. Park, Y.; Baac, H.; Heo, J.; Yoo, G. Thermally Activated Trap Charges Responsible for Hysteresis in Multilayer MoS2 FieldEffect Transistors. Appl. Phys. Lett. 2016, 108, 083102.

30. Yang, S.; Park, S.; Jang, S.; Kim, H.; Kwon, J.-Y. Electrical Stability of Multilayer MoS2 Field-Effect Transistor under Negative Bias Stress at Various Temperatures. Phys. Status Solidi RRL 2014, 8, 714–718.

23. Illarionov, Y.; Waltl, M.; Rzepa, G.; Kim, J.-S.; Kim, S.; Dodabalapur, A.; Akinwande, D.; Grasser, T. Long-Term Stability and Reliability of Black Phosphorus Field-Effect Transistors. ACS Nano 2016, 10, 9543–9549.

31. Schroder, D.; Babcock, J. Negative Bias Temperature Instability: Road to Cross in Deep Submicron Silicon Semiconductor Manufacturing. J. Appl. Phys. 2003, 94, 1– 18.

24. Grasser, T.; Goes, W.; Wimmer, Y.; Schanovsky, F.; Rzepa, G.; Waltl, M.; Rott, K.; Reisinger, H.; Afanas’ev, V.; Stesmans, A. On the Microscopic Structure of Hole Traps in pMOSFETs. IEEE Int. Electron Devices Meet. 2014; pp 21.1.1–2 1.1.4.

32. Huard, V.; Denais, M.; Parthasarathy, C. NBTI Degradation: From Physical Mechanisms to Modelling. Microelectron. Reliab. 2006, 46, 1–23.

25. Late, D. J.; Liu, B.; Matte, H. S. S. R.; Dravid, V. P.; Rao, C. N. R. Hysteresis in Single-Layer MoS2 Field Effect Transistors. ACS Nano 2012, 6, 5635–5641, PMID: 22577885.

33. Huard, V. Two Independent Components Modeling for Negative Bias Temperature Instability. IEEE Int. Reliab. Phys. Symp. Proc. 2010; pp 33–42. 34. Ang, D.; Teo, Z.; Ho, T.; Ng, C. Reassessing the Mechanisms of NegativeBias Temperature Instability by Repetitive Stress/Relaxation Experiments. IEEE Trans. Device Mater. Reliab. 2011, 11, 19– 34.

26. Lee, Y.; Kang, C.; Jung, U.; Kim, J.; Hwang, H.; Chung, H.-J.; Seo, S.; Choi, R.; Lee, B. Fast Transient Charging at the Graphene/SiO2 Interface Causing Hysteretic Device Characteristics. Appl. Phys. Lett. 2011, 98, 183508.

35. Grasser, T.; Kaczer, B.; G¨os, W.; Reisinger, H.; Aichinger, T.; Hehenberger, P.; Wagner, P.-J.; Franco, J.; Toledano-Luque, M.; Nelhiebel, M. The Paradigm Shift in Understanding the Bias Temperature Instability: From ReactionDiffusion to Switching Oxide Traps. IEEE Trans. Electron Devices 2011, 58, 3652– 3666.

27. Qiu, H.; Pan, L.; Yao, Z.; Li, J.; Shi, Y.; Wang, X. Electrical Characterization of Back-gated Bi-layer MoS2 Fieldeffect Transistors and the Effect of Ambient on Their Performances. Appl. Phys. Lett. 2012, 100, 123104. 28. Cho, A.-J.; Yang, S.; Park, K.; Namgung, S.; Kim, H.; Kwon, J.-Y. Multi-Layer MoS2 FET with Small Hysteresis by Using Atomic Layer Deposition Al2 O3 as Gate Insulator. ECS Solid State Lett. 2014, 3, Q67–Q69.

36. Grasser, T. Stochastic Charge Trapping in Oxides: From Random Telegraph Noise to Bias Temperature Instabilities. Microelectron. Reliab. 2012, 52, 39–70.

ACS Paragon Plus Environment

12

Page 13 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

37. Kaczer, B.; Roussel, P.; Grasser, T.; Groeseneken, G. Statistics of Multiple Trapped Charges in the Gate Oxide of Deeply Scaled MOSFET Devices-Application to NBTI. IEEE Electron Device Lett. 2010, 31, 411– 413.

Chrzan, D. C.; Javey, A. Measuring the Edge Recombination Velocity of Monolayer Semiconductors. Nano Lett. 2017, 17, 5356–5360, PMID: 28814079. 46. Marcus, R. A. Electron Transfer Reactions in Chemistry. Theory and Experiment. Rev. Mod. Phys. 1993, 65, 599–610.

38. Kirton, M.; Uren, M. Noise in Solid-State Microstructures: A New Perspective on Individual Defects, Interface States and LowFrequency (1/f) Noise. Adv. Phys. 1989, 38, 367–468.

47. Grasser, T.; Reisinger, H.; Wagner, P.J.; Goes, W.; Schanovsky, F.; Kaczer, B. The Time Dependent Defect Spectroscopy (TDDS) for the Characterization of the Bias Temperature Instability. IEEE Int. Reliab. Phys. Symp. Proc. 2010; pp 16–25.

39. Grill, A.; Stampfer, B.; Waltl, M.; Im, K.S.; Lee, J.-H.; Ostermaier, C.; Ceric, H.; Grasser, T. Characterization and Modeling of Single Defects in GaN/AlGaN fin-MISHEMTs. IEEE Int. Reliab. Phys. Symp. Proc. 2017; pp 3B–5.

48. Knobloch, T.; Rzepa, G.; Illarionov, Y. Y.; Waltl, M.; Schanovsky, F.; Jech, M.; Stampfer, B.; Furchi, M. M.; M¨ uller, T.; Grasser, T. Physical Modeling of the Hysteresis in MoS2 Transistors. Proc. Eur. Solid-State Device Res. Conf. 2017; pp 284– 287.

40. Das, S.; Chen, H.; Penumatcha, A.; Appenzeller, J. High Performance Multilayer MoS2 Transistors with Scandium Contacts. Nano Lett. 2013, 13, 100–105.

49. Wimmer, Y.; El-Sayed, A.-M.; G¨os, W.; Grasser, T.; Shluger, A. L. Role of Hydrogen in Volatile Behaviour of Defects in SiO2-based Electronic Devices. Proc. R. Soc. A. 2016; p 20160009.

41. Appenzeller, J.; Zhang, F.; Das, S.; Knoch, J. Transition Metal Dichalcogenide Schottky Barrier Transistors. 2D Mater. Nanoelectron. 2016, 17, 207. 42. Grasser, T.; Waltl, M.; Wimmer, Y.; Goes, W.; Kosik, R.; Rzepa, G.; Reisinger, H.; Pobegen, G.; El-Sayed, A.; Shluger, A. Gate-Sided Hydrogen Release as the Origin of ”Permanent” NBTI Degradation: From Single Defects to Lifetimes. IEEE Int. Electron Devices Meet. 2015; pp 20–1.

50. Waltl, M.; Rzepa, G.; Grill, A.; Goes, W.; Franco, J.; Kaczer, B.; Witters, L.; Mitard, J.; Horiguchi, N.; Grasser, T. Superior NBTI in High-k SiGe Transistors - Part I: Experimental. IEEE Trans. Electron Devices 2017, 64, 2092–2098. 51. Penumatcha, A. V.; Salazar, R. B.; Appenzeller, J. Analysing Black Phosphorus Transistors Using an Analytic Schottky Barrier MOSFET Model. Nat. Commun. 2015, 6 .

43. Rabiner, L.; Juang, B. An Introduction to Hidden Markov Models. IEEE ASSP Magazine 1986, 3, 4–16. 44. Uren, M. J.; Kirton, M. J.; Collins, S. Anomalous Telegraph Noise in Small-Area Silicon Metal-Oxide-Semiconductor FieldEffect Transistors. Phys. Rev. B 1988, 37, 8346–8350.

52. Knoch, J.; Riess, W.; Appenzeller, J. Outperforming the Conventional Scaling Rules in the Quantum-Capacitance Limit. IEEE Electron Device Lett. 2008, 29, 372–374.

45. Zhao, P.; Amani, M.; Lien, D.-H.; Ahn, G. H.; Kiriya, D.; Mastandrea, J. P.; Ager, J. W.; Yablonovitch, E.; ACS Paragon Plus Environment

13

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Paragon Plus Environment

Page 14 of 14