betaFeedback

WorksPhysics and Astronomy

Transistors based on Novel 2-D Monolayer Semiconductors Bi2O2Se, InSe, and MoSi2N4 for Enhanced Logic Density Scaling

Abstract

Making ultra-short gate-length transistors significantly contributes to scaling the contacted gate pitch. This, in turn, plays a vital role in achieving smaller standard logic cells for enhanced logic density scaling. As we push the boundaries of miniaturization, it is intriguing to consider that the ultimate limit of contacted gate pitch could be reached with remarkable 1 nm gate-length transistors. Here, we identify InSe, Bi2O2Se, and MoSi2N4 as potential two-dimensional semiconductors for 1 nm transistors with low contact resistance and outstanding interface properties. We employ a fully self-consistent ballistic quantum transport model starting from first-principle calculations. Our simulations show that the interplay between electrostatics and quantum tunneling influences the performance of these devices over the device design space. MoSi2N4 channels have the best immunity to quantum tunneling, and Bi2O2Se channel devices have the best electrostatics. We show that for a channel length of 12 nm, all the devices can deliver I_ONON/I_OFFOFF > 10^3 , suitable for electronic applications, and Bi2O2Se is the best-performing channel material.

Introduction

FETs based on 2-D channel materials are more immune to short-channel effects (SCEs) and offer better device scaling [1], [2]. In ultra-thin-body devices, they show superior gate control and better mobility than bulk semiconductors such as Si, Ge, and III-V. In recent years, experimental 2-D FETs have shown ideal performance. In addition, one can integrate a 2-D channel with a 1-D metallic nanowire, nanotube, or 2-D graphene as the gate. This integration facilitates 1 nm and sub-1 nm gate lengths in 2-D FETs [3, 4, 5, 6, 7, 8]. Considering the interface quality of the gate stack, metal contact with 2-D semiconductors, and quality of the grown channel, we study a few potential contenders for the channel material in ultra-short 2-D FETs. They are monolayers of MoSi2N4\mathrm{MoSi}_{2}\mathrm{N}_{4}, Bi2O2Se\mathrm{Bi}_{2}\mathrm{O}_{2}\mathrm{Se}, and InSe.

MoSi2N4\mathrm{MoSi}_{2}\mathrm{N}_{4} was synthesized in 2020, showing excellent physical, mechanical, thermal, electronic, and metal contact properties [9, 10, 11, 12]. It outperforms most other 2-D semiconductors and shows excellent ohmic contact with CMOS-compatible metals [10]. Bi2O2Se\mathrm{Bi}_{2}\mathrm{O}_{2}\mathrm{Se} is a ternary semiconducting compound with outstanding stability and a moderate bandgap [13], [14]. High-kk native-crystalline oxide dielectric, Bi2SeO5\mathrm{Bi}_{2}\mathrm{SeO}_{5}, can be formed by layer-by-layer oxidation of Bi2O2Se\mathrm{Bi}_{2}\mathrm{O}_{2}\mathrm{Se}. At wafer-scale, Bi2O2Se\mathrm{Bi}_{2}\mathrm{O}_{2}\mathrm{Se} FETs can be fabricated using this native oxide as gate oxide with EOT below sub-0.5 nm [15]. The performance of wafer-scale Bi2O2Se\mathrm{Bi}_{2}\mathrm{O}_{2}\mathrm{Se} based FinFETs with Bi2SeO5\mathrm{Bi}_{2}\mathrm{SeO}_{5} as gate insulators is competitive to Si and Ge-based FinFETs. These Bi2O2Se\mathrm{Bi}_{2}\mathrm{O}_{2}\mathrm{Se} FinFETs show low contact resistance, RC∼470 Ω μmR_{\mathrm{C}}\sim470\ \Omega\ \mu\mathrm{m} [16]. InSe FETs have shown ideal performance close to the ballistic limit down to 10 nm gate length with a power supply voltage of 0.5 V and low contact resistance, RC∼65 Ω μmR_{\mathrm{C}}\sim65\ \Omega\ \mu\mathrm{m} [17]. Wafer-scale InSe meets the standard for electronic-grade semiconductor material [18]. These are ideal atomic thin channel materials for next-generation logic design and monolithic three-dimensional (M3D) integration with state-of-the-art Si devices.

Here, we explore the potential of ∼1\sim1 nm gate-length transistors using novel 2-D semiconducting monolayers, highlighting their excellent performance characteristics and role in advancing electronic device miniaturization. Our research highlights the role of geometrical and material parameters—such as the nanowire gate, spacer, gate dielectric, and channel length—within integrated 2-D channel FETs featuring metallic nanowires. Considering the favorable figures for 2-D semiconductors for the transistor’s channel, we choose monolayers of MoSi2N4\mathrm{MoSi}_{2}\mathrm{N}_{4}, Bi2O2Se\mathrm{Bi}_{2}\mathrm{O}_{2}\mathrm{Se}, and InSe as channels in our research. The possibility of integrating a 2-D channel with a 1-D metallic nanowire as a gate alleviates the lithography limitation on gate patterning [3, 4, 5, 6]. This configuration allows one to realize the scaling potential of 2-D FETs, thanks to their natural atomic-scale dimensions. The key findings of our research are the following:

  • The fringing field from the nanowire gate enhances gate efficiency, which in turn improves gate control over the channel for a given gate oxide. When the spacer material

is changed from air to SiO2_2, gate efficiency improves by 75%, leading to better switching characteristics for the devices.

  • Slight performance improvements are observed when using CaF2_2 as both the spacer and gate oxide, compared to SiO2_2 (as both the spacer and gate oxide) with the same EOT. This is mainly due to the increased physical thickness of the CaF2_2 gate oxide, which reduces the benefits of having a higher-kk spacer (see Figure 3 and Table I). Although a higher-kk spacer (with greater permittivity than CaF2_2) can provide some advantages, it also introduces increased cross-talk, which is undesirable. Moreover, reducing the gate oxide thickness proves beneficial for a given spacer.

Comparison of SiO₂ and CaF₂ spacers and gate dielectrics

Figure 3. The relative comparison between (a) the SiO₂ spacer with a SiO₂ gate dielectric and (b) the CaF₂ spacer with a CaF₂ gate dielectric, both having the same EOT of 1 nm, shows that the increased TOXT_{\mathrm{OX}} in (b) results in only a slight improvement in gate control compared to (a).

  • As the gate oxide thickness increases, the gate’s control over the channel weakens due to reduced gate oxide capacitance and decreased fringing capacitance from the gate. This results in a logarithmic relationship between the subthreshold slope and gate oxide thickness, contrasting with conventional FETs [19, 20].

  • Increasing the dimensions of the gate uniformly improves switching metrics for a given channel length, and the reverse is also true. However, altering the gate shape from square to triangular degrades the device’s performance due to increased direct source-to-drain tunneling. MoSi2_2N4_4 demonstrates greater resistance to this type of tunneling. For instance, the subthreshold swing and on-state current deteriorate by approximately 14.5% and 53%, respectively, when changing the nanowire gate structure from square to triangular. In contrast, Bi2_2O2_2Se and InSe devices experience significant leakage due to severe direct tunneling, attributed to their low carrier effective mass.

  • Ultra-scaled FETs with these promising 2-D semiconducting channels achieve ION/IOFF>103I_{\mathrm{ON}}/I_{\mathrm{OFF}} > 10^3 with SS<130SS < 130 mV/decade. Bi2_2O2_2Se stands out, closely followed by InSe and MoSi2_2N4_4.

Methodology

To model the full-band transport in the devices, an accurate model Hamiltonian for the channel is required [21]. For this, the electronic properties of channel material are characterized

from first-principles by combining the power of density-functional theory (DFT) [22] and maximally localized Wannier functions (MLWFs) [23]. The optimized structural parameters and bandgap values agree well with the literature. All these are indirect bandgap semiconductors, whose bandgap value lies in the range 1.3-1.9 eV: InSe ∼1.52\sim1.52 eV, Bi2_2O2_2Se ∼1.34\sim1.34 eV, and MoSi2_2N4_4 ∼1.84\sim1.84 eV. Model Hamiltonian validation against DFT results involves calculating eigenenergies along high symmetry points. The band structure from our model demonstrates good agreement with DFT, particularly around the bandgap (for example, see Figure 1 for Bi2_2O2_2Se and InSe).

Comparison of band structures calculated from DFT and tight-binding Hamiltonian

Figure 1. Comparison of band structure calculated from DFT and tight-binding Hamiltonian (in MLWFs basis) along high symmetry paths in the Brillouin zone for (a) Bi2_2O2_2Se and (b) InSe. The band structure from MLWFs Hamiltonian shows good agreement with DFT in the vicinity of conduction band minima and valance band maxima. Fermi energy level is shifted to zero energy.

The charge density in the channel is obtained by estimating the Green’s functions in the non-equilibrium Green’s function (NEGF) formalism [24]. The channel charge is included in Poisson’s equation to determine the converged physical observables of the system [25]. Green’s function is obtained by uniformly sampling the transverse wavevectors with 30 sampling points. The source-to-drain current is calculated by Landauer-Büttiker approach [26]. The device structure is shown in Figure 2 (a), which is analogous to experimental structures in [4, 6]. The side view of the atomic structure of MoSi2_2N4_4, InSe, and Bi2_2O2_2Se is also shown in Figure 2. Figure 3 shows the relative comparison of gate stacks between two different gate DE, both having the same EOT of 1 nm. Considering the ultra-low contact resistance and stability of 2-D devices contacted with semimetal, we mimic the degenerate state for the semiconductor in contact with semimetal by chemically doping the source and drain regions in our simulations [17, 27, 28, 29]. These results should be regarded as the upper limit for device performance, as non-ideal effects are not considered here. All device simulations are carried out at room temperature.

Schematic of top-gated monolayer 2-D FETs with different gate structures

Figure 2. (a) Schematic of top-gated monolayer 2-D FETs with different gate structures like square, rectangle, isosceles, and triangle. The channel comprises of monolayer MoSi2_2N4_4. SiO2_2 is used as BOX oxide with a thickness of TBOXT_{\mathrm{BOX}}. The gate is of metal and separated from the channel by TOXT_{\mathrm{OX}}. Side view of the atomic structure of (b) InSe and (c) Bi2_2O2_2Se.

Results and Discussion

To capture the switching behavior in these ultra-scaled logic devices, we begin with the monolayer MoSi₂N₄ channel. The gate dimensions are 1 nm × 1 nm, and equivalent oxide thickness (EOT) is 1 nm. The transfer characteristics on the semilogarithmic scale are shown in Figure 4 (a) for the air spacer with SiO₂ gate dielectric, the SiO₂ spacer with SiO₂ gate dielectric, and the calcium fluoride (CaF₂) spacer with CaF₂ gate dielectric with LChL_{\mathrm{Ch}} = 12 nm. Ensuring a high-quality integration of 2-D channels with high-kk oxides is paramount. Epitaxial CaF₂ can establish a quasi-van der Waals contact with 2-D semiconductors [30, 31], resulting in excellent interface quality. Also, it can serve as an ultrathin gate insulator with a high dielectric constant (ϵ\epsilon = 8.43) for 2-D devices with an equivalent oxide thickness (EOT) down to 1 nm and minimal leakage current (ION/IOFF>107I_{\mathrm{ON}}/I_{\mathrm{OFF}} > 10^{7}) [32]. For EOT values of 0.5 nm and 1 nm, 2-D devices with CaF₂ gate stack show lower leakage than other high-kk oxides, like HfO₂ and La₂O₃ [32]. Conversely, defects can adversely impact metal-oxide-semiconductor field-effect transistors (MOSFETs) by causing issues such as charge trapping and de-trapping, bias temperature instabilities (BTI) (driven by charge de-trapping), and mobility degradation (due to coulomb scattering from trapped charges in the oxide) [30].

Transfer characteristics and surface potential profiles

Figure 4. (a) Transfer characteristics of ultra-short 2-D FETs comprised of MoSi₂N₄ for air spacer with SiO₂ gate dielectric, SiO₂ spacer with SiO₂ gate dielectric, and CaF₂ spacer with CaF₂ gate dielectric. LCh∼12L_{\mathrm{Ch}} \sim12 nm and TG×LG=1T_{\mathrm{G}} \times L_{\mathrm{G}} = 1 nm × 1 nm. The surface plot of the potential profile in the source, channel, and drain for (b) air spacer, (c) SiO₂ spacer, and (d) CaF₂ spacer at VDSV_{\mathrm{DS}} = 0.5 V and VGSV_{\mathrm{GS}} = −0.5 V. An increase in the dielectric constant of the spacer results in better gate efficiency and, therefore, a larger effective channel length, LCh,effL_{\mathrm{Ch,eff}}. The device with an air spacer has the worst gate efficiency.

Role of Gate Oxide and Spacer

We observe that effective channel length (LeffL_{\mathrm{eff}}) increases (see Figure 4 (b-Figure 4d)), and switching characteristics improve with an increase in the spacer’s permittivity. This improvement is attributed to increased parasitic capacitance arising from the fringing field from gate [33], Cpar∼ϵspacer×ln⁡(1+TG/Tox)C_{\mathrm{par}} \sim\epsilon_{\mathrm{spacer}} \times\ln(1 + T_{\mathrm{G}}/T_{\mathrm{ox}}), where ϵspacer\epsilon_{\mathrm{spacer}} is the permittivity of spacer. The device with air spacer shows poor switching characteristics with gate efficiency < 0.3 and subthreshold slope, SS∼236SS \sim236 mV/decade with the ON to the OFF current ratio, ION/IOFF>102I_{\mathrm{ON}}/I_{\mathrm{OFF}} > 10^{2}. Gate efficiency is defined as −Δϕb/ΔVGS-\Delta\phi_{\mathrm{b}}/\Delta V_{\mathrm{GS}} in the subthreshold regime, where ϕb\phi_{\mathrm{b}} is the source-to-channel barrier height. The simulations are carried out at VDS=0.5V_{\mathrm{DS}} = 0.5 V, where VDSV_{\mathrm{DS}} denotes the drain to source voltages. The VOFFV_{\mathrm{OFF}} voltage is the gate voltage, VGSV_{\mathrm{GS}}, that produces a drain current IDSI_{\mathrm{DS}} equal to a specified IOFF∼10−1 μI_{\mathrm{OFF}} \sim10^{-1}\ \muA/μ\mum. The IONI_{\mathrm{ON}} of a FET is measured at the ON-state voltage VON=VOFF+VDDV_{\mathrm{ON}} = V_{\mathrm{OFF}} + V_{\mathrm{DD}}, where VDD=0.5V_{\mathrm{DD}} = 0.5 V is the power supply voltage. Replacing

the spacer with higher permittivity improves the switching characteristics. For example, SiO₂ spacer improves the gating efficiency by 75%, therefore better switching characteristics (SS∼129SS \sim129 mV/decade) and ION/IOFFI_{\mathrm{ON}}/I_{\mathrm{OFF}} (> 10310^{3}). However, small performance improvements are observed in the case of CaF₂ as both spacer and gate oxide, compared to SiO₂ (as both spacer and gate oxide). This is primarily due to its increased gate oxide physical thickness, which diminishes the impact of having higher-kk spacer (see Figure 3 and Table I). Building on this analysis, we further demonstrate the impact of spacers with CaF₂ gate oxide. The results indicate that while a higher-kk spacer (with greater permittivity than CaF₂) can offer advantages, it comes at the cost of increased cross-talk, which is undesirable. Additionally, reducing the gate oxide thickness proves advantageous within the given spacer configuration.

Further, we assessed the performance of ultra-short devices with various spacers, ranging from low-kk to high-kk materials, using CaF₂ as a high-kk gate dielectric (see Figure 5), which is ideally suited for 2-D channel materials. Performance improvements are noted when k>8k > 8, but this range is less favorable due to heightened electronic cross-talk. To analyze the impact of physical gate oxide thickness, we consider low-kk spacer with permittivity of 3 and CaF₂ as high-kk gate oxide. We perform the analysis with different gate oxide

Transfer characteristics of ultra-short 2-D FETs showing the role of spacer

Figure 5. Transfer characteristics of ultra-short 2-D FETs, showing the role of spacer. The channel is comprised of MoSi2_2N4_4, and CaF2_2 is used as the high-kk gate dielectric. LCh∼12L_{\mathrm{Ch}} \sim12 nm and TG×LG=1T_{\mathrm{G}} \times L_{\mathrm{G}} = 1 nm ×\times 1 nm. Small performance gains are observed when k>8k > 8, though this range is less favorable due to increased electronic cross-talk.

Gate OxideSpacerSSSS (mV/decade)IONI_{\mathrm{ON}} (μ\muA/μ\mum)ION/IOFFI_{\mathrm{ON}}/I_{\mathrm{OFF}}
SiO₂Air23618.451.84 × 10210^{2}
SiO₂SiO₂128.98106.671.07 × 10310^{3}
CaF₂CaF₂124.90117.621.18 × 10310^{3}

Table 1. SUBTHRESHOLD SLOPE (SSSS), ON-STATE CURRENT (IONI_{\mathrm{ON}}), AND ON CURRENT TO THE OFF CURRENT RATIO (ION/IOFFI_{\mathrm{ON}}/I_{\mathrm{OFF}}) FOR DIFFERENT COMBINATIONS OF THE GATE OXIDES AND SPACERS.

thicknesses (TOXT_{\mathrm{OX}}), ranging from 3.3 nm (EOT ∼\sim 1.5 nm) down to 1.1 nm (EOT ∼\sim 0.5 nm). As the oxide thickness (TOXT_{\mathrm{OX}}) increases for a given gate oxide, gate dimensions, and spacer configuration, the gate’s control over the channel weakens. This occurs due to a combination of effects from the gate oxide capacitance and fringing capacitance caused by the spacer. The gate oxide capacitance decreases when the gate oxide thickness increases, reducing the gate’s control. Additionally, the fringing capacitance from the spacer plays a role in gate control in short-channel devices. An increase in TOXT_{\mathrm{OX}} decreases parasitic capacitance, further weakening the gate’s control, and conversely, reducing TOXT_{\mathrm{OX}} increases parasitic capacitance, enhancing the gate’s control over the channel. To show the impact of TOXT_{\mathrm{OX}} quantitatively, we plot the transfer characteristics with different TOXT_{\mathrm{OX}} ranging from 3.3 nm down to 1.1 nm (see Figure 6). This TOXT_{\mathrm{OX}} variation is equivalent to EOT’s variation from 1.5 nm down to 0.5 nm. The channel is comprised of MoSi2_2N4_4, CaF2_2 is used as gate dielectric, and aBN is used as low-kk spacer. The SSSS is plotted with TG/TOXT_{\mathrm{G}}/T_{\mathrm{OX}} in inset of Figure 6. Unlike conventional FETs [19], [20], the SSSS versus TOXT_{\mathrm{OX}} trend is logarithmic with TOXT_{\mathrm{OX}} due to the role of fringing in determining the gate control. Further study considers SiO2_2 to be both a spacer and gate oxide. The gate oxides and spacers are considered perfect insulators.

Transfer characteristics of ultra-short 2-D FETs showing the role of gate oxide thickness scaling

Figure 6. Transfer characteristics of ultra-short 2-D FETs, showing the role of gate oxide thickness scaling. The channel is comprised of MoSi2_2N4_4, CaF2_2 is high-kk gate dielectric, and amorphous boron nitride (aBN) is low-kk spacer. LCh∼12L_{\mathrm{Ch}} \sim12 nm and TG×LG=1T_{\mathrm{G}} \times L_{\mathrm{G}} = 1 nm ×\times 1 nm. In the inset, SSSS is plotted with TG/TOXT_{\mathrm{G}}/T_{\mathrm{OX}}. Unlike conventional FETs, this trend is logarithmic with TOXT_{\mathrm{OX}}.

Role of Channel Materials

For 2-D semiconductors, the centroid capacitance is often assumed to be infinite [34], therefore CS∼CQC_{\mathrm{S}} \sim C_{\mathrm{Q}}, where CSC_{\mathrm{S}} is the capacitance of the channel material and CQC_{\mathrm{Q}} is the quantum capacitance of the channel material. Therefore, SSSS is proportional to (1+CQ/COX)(1 + C_{\mathrm{Q}}/C_{\mathrm{OX}}) in the negligible tunneling regime [35]. Thus, a lower quantum capacitance channel would be beneficial in reducing the thermionic SSSS [36]. This is evident from the subthreshold slope of the thermionic current, SSthermalSS_{\mathrm{thermal}}. It is the best for the Bi2_2O2_2Se channel (∼98\sim98 mV/decade) and slightly degrades for InSe (∼100\sim100 mV/decade) compared to Bi2_2O2_2Se. As the quantum capacitance values are approximately similar for InSe (me∼0.205m0m_{\mathrm{e}} \sim0.205m_0 and gv=1g_{\mathrm{v}} = 1) and Bi2_2O2_2Se (me∼0.197m0m_{\mathrm{e}} \sim0.197m_0 and gv=1g_{\mathrm{v}} = 1), both being less than one-fourth of MoSi2_2N4_4 (me∼0.48m0m_{\mathrm{e}} \sim0.48m_0 and gv=2g_{\mathrm{v}} = 2). For MoSi2_2N4_4, the subthreshold slope is the worst, close to 125 mV/decade. The thermionic current component is plotted in Figure 7 (a), providing a zoomed view of the subthreshold regime. However, at a channel length of LCh=12L_{\mathrm{Ch}} = 12 nm, the switching behavior is influenced by both direct tunneling leakage and electrostatics. The contribution of direct tunneling from source-to-drain in the off-state current, IOFFI_{\mathrm{OFF}}, is significant for InSe (∼54%\sim54\%) and Bi2_2O2_2Se (∼59%\sim59\%) but lower for MoSi2_2N4_4 (∼12%\sim12\%). The strength of InSe and Bi2_2O2_2Se are better electrostatic control in the devices comprised of them. Even in the presence of large direct tunneling, InSe and Bi2_2O2_2Se demonstrate better control over the switching behavior compared to MoSi2_2N4_4 for LCh=12L_{\mathrm{Ch}} = 12 nm (see Figure 7).

Benchmarking of transfer characteristics and ION vs. ION/IOFF for three promising 2-D semiconductors

Figure 7. (a) Benchmarking of transfer characteristics and (b) IONI_{\mathrm{ON}} vs. ION/IOFFI_{\mathrm{ON}}/I_{\mathrm{OFF}} for three promising 2-D semiconductors as a channel in ultra-short FET architecture. SiO2_2 is used as gate oxide with EOT = 1 nm, spacer, and BOX oxide with TBOXT_{\mathrm{BOX}} = 10 nm. The channel length is 12 nm, TG×LGT_{\mathrm{G}} \times L_{\mathrm{G}} = 1 nm ×\times 1 nm, and VDSV_{\mathrm{DS}} = 0.50 V. Bi2_2O2_2Se device shows the best switching characteristic and IONI_{\mathrm{ON}} for a given ION/IOFFI_{\mathrm{ON}}/I_{\mathrm{OFF}}.

Devices comprised of these promising 2-D semiconducting channels can deliver ION/IOFF>103I_{\mathrm{ON}}/I_{\mathrm{OFF}} > 10^3 with SS<130SS < 130 mV/decade. Bi2_2O2_2Se is the best-performing channel material (SS∼109SS \sim109 mV/decade and ION∼186I_{\mathrm{ON}} \sim186 μ\muA/μ\mum) and the performance of the InSe (SS∼112SS \sim112 mV/decade and ION∼153I_{\mathrm{ON}} \sim153 μ\muA/μ\mum) is close to Bi2_2O2_2Se, followed by MoSi2_2N4_4. The ION/IOFFI_{\mathrm{ON}}/I_{\mathrm{OFF}} of all these three channel materials are one order higher than 1 nm MoS2_2 FET with a circular metallic gate of diameter 1 nm with 10 nm channel length (SS∼160SS \sim160 mV/decade and ION/IOFF∼400I_{\mathrm{ON}}/I_{\mathrm{OFF}} \sim400) [37]. Increasing the LChL_{\mathrm{Ch}} degrades the control of the gate over the channel and, on the contrary, reduces the impact of direct tunneling (because of increased tunneling barrier height and width) and vice versa (see Figure 9 (a)). This device configuration comprises these novel 2-D semiconductors and can deliver better switching performance than other ultra-short device configurations like FETs with core-shell nanowire gates [3], [7].

Benchmarking of switching characteristics and circuit-level characteristics

Figure 9. (a) Benchmarking of switching characteristics (SS verss ION/IOFFI_{\mathrm{ON}}/I_{\mathrm{OFF}}) for the devices. (b) Circuit level characteristics (τ\tau versus EDP) for a 20-stage inverter comprised of these ultra-short 2-D channel devices.

In the quest of performance improvement, applying 1% tensile strain along the channel direction, which is reasonable in experiments with flexible substrates [38], [39], leads to minor change in performance of the FETs compared to pristine channel FETs. As, it can lower the CQC_{\mathrm{Q}} by 3.8%, -0.25%, and -0.005% in MoSi2_2N4_4, Bi2_2O2_2Se, and InSe, respectively, than a pristine channel. The decrease in bandgap is less than 6% compared to pristine monolayers, which is suitable for mitigating intra-band tunneling leakage. Another possible solution could be utilizing the non-linear dielectric response of dielectric-ferroelectric (DE-FE) [40] stacked or FE-antiferroelectric (FE-AFE) [41] stacked materials in the gate-stack. This is a promising approach to enhance the performance of devices by amplifying the surface potential’s response to changes in gate voltage.

Role of Gate's dimensions and Performance Metrics

The role of gate geometry and dimensions is also studied for better assessment of device performance over design space. We find that a uniform increment in TGT_{\mathrm{G}} and LGL_{\mathrm{G}} improves the switching metrics for a given channel length and vice versa. For example, a change in gate dimensions from 0.5 ×\times 0.5 to 2 ×\times 2, improves the gate efficiency by ∼\sim 45%, SSSS by ∼\sim 30%, and IONI_{\mathrm{ON}} by ∼\sim 215% in MoSi2_2N4_4 devices. Increasing gate height from 0.1 nm to 2 nm with LGL_{\mathrm{G}} of 1 nm enhances the gate efficiency by ∼\sim 34%, SSSS by ∼\sim 31%, and IONI_{\mathrm{ON}} by ∼\sim 189%. The gate geometry variations, from square to triangular, are characterized by an angle theta (in degrees). For MoSi2_2N4_4, the SSSS and IONI_{\mathrm{ON}} are deteriorated by ∼\sim 14.5% and 53%, respectively, while varying the gate structure from square (theta = 90) to triangular (theta = 63.44). For theta >> 72 (78), the degradation in IONI_{\mathrm{ON}} is less than 32% (23%), see Figure 8. For Bi2_2O2_2Se and InSe devices, severe direct tunneling results in severe leakage for theta << 73.30 due to low carrier effective mass.

The role of gate dimensions in the performance of ultra-short channel devices

Figure 8. The role of gate dimensions in the performance of ultra-short channel devices.

Finally, critical circuit-level performance metrics are estimated for a chain of inverters with a logic depth of 20, as shown in Figure 9. We have estimated the delay and energy per operation. These metrics have been normalized by the node capacitance and the total capacitance of the logic design. This

normalization approach is reasonable in sub-10 nm technology, as interconnect capacitance becomes the dominant component of total capacitance [42], [43]. For these promising channel materials, the normalized delay (τ\tau) and energy-delay product (EDP) for a chain of 20 inverters are impressively low, measuring less than 95 and 1.5, respectively. To put this into perspective, these circuit-level metrics outperform 1 nm MoS2_2 FET with a circular metallic gate of diameter 1 nm with 10 nm channel length (the τ\tau is around 250, and the EDP is approximately 5 [37]).

Conclusion

In the pursuit of identifying promising 2-D semiconductors for cutting-edge electronic applications, we have identified MoSi2_2N4_4, InSe, and Bi2_2O2_2Se as highly promising candidates. By considering the feasibility of integrating metallic nanowires with 2-D semiconductors, we have investigated their suitability as channel materials within an innovative device structure. Our goal is to harness their unique properties and capabilities for ultra-scaled FETs with ultra-scaled gate lengths (approximately 1 nm). We have found that the Bi2_2O2_2Se FET exhibits the most favorable switching metrics and circuit-level performance. InSe FETs perform closely to Bi2_2O2_2Se, with MoSi2_2N4_4 following closely behind.

Beyond its applications in logic circuits, this novel device structure has the potential to be extended into a multi-gate configuration model [6]. This extension opens up exciting possibilities for achieving precise and localized control over band structures, which in turn enables the design of more integrated, multifunctional, and highly controllable nanodevices. Moreover, this versatile device topology can also find utility in the development of ultra-low-power spiking neurons, particularly for neuromorphic applications [44]. This represents an exciting direction where the technology can contribute to the advancement of brain-inspired computing.

Acknowledgment

The authors thank the HPC facility, IIT Kanpur, Kanpur, India, and the National Supercomputing Mission (NSM) for providing the computational resources. KN is thankful to Ashok for the insightful discussions.

References

References

  1. [1]F. Zhuo, J. Wu, B. Li, M. Li, C. L. Tan, Z. Luo, H. Sun, Y. Xu, and Z. Yu, “Modifying the power and performance of 2-dimensional MoS₂ field effect transistors,” Research, vol. 6, p. 0057, 2023, doi: 10.34133/research.0057.DOI
  2. [2]K. Nandan, A. Agarwal, S. Bhowmick, and Y. S. Chauhan, “Two-dimensional semiconductors based field-effect transistors: review of major milestones and challenges,” Frontiers in Electronics, vol. 4, 2023, doi: 10.3389/felec.2023.1277927.
  3. [3]W. Cao, W. Liu, and K. Banerjee, “Prospects of ultra-thin nanowire gated 2d-fets for next-generation cmos technology,” in 2016 IEEE International Electron Devices Meeting (IEDM), 2016, doi: 10.1109/IEDM.2016.7838419. pp. 14.7.1–14.7.4.DOI
  4. [4]S. B. Desai, S. R. Madhvapathy, A. B. Sachid, J. P. Llinas, O. Wang, G. H. Ahn, G. Pitner, M. J. Kim, J. Bokor, C. Hu, H.-S. P. Wong, and A. Javey, “MoS₂ transistors with 1-nanometer gate lengths,” Science, vol. 354, no. 6308, pp. 99–102, 2016, doi: 10.1126/science.aah4698.DOI
  5. [5]F. Wu, H. Tian, Y. Shen, Z. Hou, J. Ren, G. Gou, Y. Sun, Y. Yang, and T.-L. Ren, “Vertical MoS₂ transistors with sub-1-nm gate lengths,” Nature, vol. 603, no. 7900, pp. 259–264, Mar 2022, doi: 10.1038/s41586-021-04323-3.DOI
  6. [6]Y. Xiao, X. Zhou, J. Huo, T. Sun, B. Feng, and L. Liu, “Locally thinned, core–shell nanowire-integrated multi-gate MoS₂ transistors for active control of extendable logic,” ACS Applied Materials & Interfaces, vol. 15, no. 1, pp. 1563–1573, 2023, doi: 10.1021/acsami.2c17788.DOI
  7. [7]X. Liu, R. Liang, G. Gao, C. Pan, C. Jiang, Q. Xu, J. Luo, X. Zou, Z. Yang, L. Liao, and Z. L. Wang, “MoS₂ negative-capacitance field-effect transistors with subthreshold swing below the physics limit,” Advanced Materials, vol. 30, no. 28, p. 1800932, 2018, doi: 10.1002/adma.201800932.
  8. [8]W. Jolie and T. Michely, “1d metals for 2d electronics,” Nature Nanotechnology, pp. 1–2, 2024.DOI
  9. [9]Y.-L. Hong, Z. Liu, L. Wang, T. Zhou, W. Ma, C. Xu, S. Feng, L. Chen, M.-L. Chen, D.-M. Sun, X.-Q. Chen, H.-M. Cheng, and W. Ren, “Chemical vapor deposition of layered two-dimensional MoSi₂N₄ materials,” Science, vol. 369, no. 6504, pp. 670–674, 2020, doi: 10.1126/sci- ence.abb7023.DOI
  10. [10]Q. Wang, L. Cao, S.-J. Liang, W. Wu, G. Wang, C. H. Lee, W. L. Ong, H. Y. Yang, L. K. Ang, S. A. Yang, and Y. S. Ang, “Efficient ohmic contacts and built-in atomic sublayer protection in mosi₂n₄ and wsi₂n₄ monolayers,” npj 2D Materials and Applications, vol. 5, no. 1, p. 71, Aug 2021, doi: 10.1038/s41699-021-00251-y.DOI
  11. [11]K. Nandan, B. Ghosh, A. Agarwal, S. Bhowmick, and Y. S. Chauhan, “Two-dimensional hmosi₂n₄: An excellent 2-d semiconductor for field-effect transistors,” IEEE Transactions on Electron Devices, pp. 1–8, 2021, doi: 10.1109/TED.2021.3130834.DOI
  12. [12]K. Nandan, S. Bhowmick, Y. S. Chauhan, and A. Agarwal, “Designing power-efficient transistors using narrow-bandwidth materials from the mA₂Z₄ (m = Mo, Cr, Zr, Ti, Hf; a = Si, Ge; z = N, P, As) monolayer series,” Phys. Rev. Appl., vol. 19, p. 064058, Jun 2023, doi: 10.1103/PhysRevApplied.19.064058.DOI
  13. [13]T. Li and H. Peng, “2D Bi₂O₂Se: An emerging material platform for the next-generation electronic industry,” Accounts of Materials Research, vol. 2, no. 9, pp. 842–853, 2021, doi: 10.1021/accountsmr.1c00130.DOI
  14. [14]R. Quhe, J. Liu, J. Wu, J. Yang, Y. Wang, Q. Li, T. Li, Y. Guo, J. Yang, H. Peng et al., “High-performance sub-10 nm monolayer bi₂o₂se transistors,” Nanoscale, vol. 11, no. 2, pp. 532–540, 2019.DOI
  15. [15]Y. Zhang et al., “A single-crystalline native dielectric for two-dimensional semiconductors with an equivalent oxide thickness below 0.5 nm,” Nature Electronics, vol. 5, no. 10, pp. 643–649, Oct 2022, doi: 10.1038/s41928-022-00824-9.DOI
  16. [16]C. Tan, M. Yu, J. Tang, X. Gao, Y. Yin, Z. Zhang, J. Wang, X. Gao, C. Zhang, X. Zhou, L. Zheng, H. Liu, K. Jiang, F. Ding, and H. Peng, “2D fin field-effect transistors integrated with epitaxial high-k gate oxide,” Nature, vol. 616, no. 7955, pp. 66–72, Apr 2023, doi: 10.1038/s41586-023-05797-z.DOI
  17. [17]J. Jiang, L. Xu, C. Qiu, and L.-M. Peng, “Ballistic two-dimensional InSe transistors,” Nature, vol. 616, no. 7957, pp. 470–475, Apr 2023, doi: 10.1038/s41586-023-05819-w.DOI
  18. [18]S. Song, S. Jeon, M. Rahaman, J. Lynch, D. Rhee, P. Kumar, S. Chakravarthi, G. Kim, X. Du, E. W. Blanton, K. Kisslinger, M. Snure, N. R. Glavin, E. A. Stach, R. H. Olsson, and D. Jariwala, “Wafer-scale growth of two-dimensional, phase-pure InSe,” Matter, vol. 6, no. 10, pp. 3483–3498, 2023, doi: 10.1016/j.matt.2023.07.012.DOI
  19. [19]A. AlMutairi and Y. Yoon, “Device performance assessment of monolayer hfse2: A new layered material compatible with high- k hfo₂,” IEEE Electron Device Letters, vol. 39, no. 11, pp. 1772–1775, 2018, doi: 10.1109/LED.2018.2867957.DOI
  20. [20]Y. Yoon, K. Ganapathi, and S. Salahuddin, “How good can be monolayer mos2 transistors be?” Nano letters, vol. 11, no. 9, pp. 3768–3773, 2011.
  21. [21]K. Nandan, A. Agarwal, S. Bhowmick, and Y. S. Chauhan, “Performance investigation of p-fets based on highly air-stable monolayer pentagonal pdse₂,” IEEE Transactions on Electron Devices, vol. 68, no. 12, pp. 6551–6557, 2021, doi: 10.1109/TED.2021.3119552.DOI
  22. [22]G. Kresse and J. Furthmüller, “Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set,” Phys. Rev. B, vol. 54, pp. 11169–11186, Oct 1996, doi: 10.1103/PhysRevB.54.11169.DOI
  23. [23]A. A. Mostofi, J. R. Yates, G. Pizzi, Y.-S. Lee, I. Souza, D. Vanderbilt, and N. Marzari, “An updated version of wannier90: A tool for obtaining maximally-localised wannier functions,” Computer Physics Communications, vol. 185, no. 8, pp. 2309–2310, 2014, doi: 10.1016/j.cpc.2014.05.003.
  24. [24]S. Datta, Quantum transport: atom to transistor. Cambridge university press, 2005.DOI
  25. [25]S. Bruzzone, G. Iannaccone, N. Marzari, and G. Fiori, “An open-source multiscale framework for the simulation of nanoscale devices,” IEEE Transactions on Electron Devices, vol. 61, no. 1, pp. 48–53, 2014, doi: 10.1109/TED.2013.2291909.DOI
  26. [26]M. Büttiker, Y. Imry, R. Landauer, and S. Pinhas, “Generalized many-channel conductance formula with application to small rings,” Phys. Rev. B, vol. 31, pp. 6207–6215, May 1985, doi: 10.1103/PhysRevB.31.6207.DOI
  27. [27]P.-C. Shen, C. Su, Y. Lin, A.-S. Chou, C.-C. Cheng, J.-H. Park, M.-H. Chiu, A.-Y. Lu, H.-L. Tang, M. M. Tavakoli et al., “Ultralow contact resistance between semimetal and monolayer semiconductors,” Nature, vol. 593, no. 7858, pp. 211–217, 2021.
  28. [28]W. Li, X. Gong, Z. Yu, L. Ma, W. Sun, S. Gao, Ç. Köroğlu, W. Wang, L. Liu, T. Li et al., “Approaching the quantum limit in two-dimensional semiconductor contacts,” Nature, vol. 613, no. 7943, pp. 274–279, 2023.DOI
  29. [29]T. Wu and J. Guo, “Multiscale modeling of semimetal contact to two-dimensional transition metal dichalcogenide semiconductor,” Applied Physics Letters, vol. 121, no. 2, 2022.DOI
  30. [30]Y. Y. Illarionov et al., “Transistors for 2D nanoelectronics: the gap to bridge,” Nature Communications, vol. 11, no. 1, p. 3385, 2020.DOI
  31. [31]Y. Y. Illarionov, A. G. Banshchikov, D. K. Polyushkin, S. Wachter, T. Knobloch, M. Thesberg, M. I. Vexler, M. Waltl, M. Lanza, N. S. Sokolov et al., “Reliability of scalable mos2 fets with 2 nm crystalline caf2 insulators,” 2D Materials, vol. 6, no. 4, p. 045004, 2019.DOI
  32. [32]Y. Y. Illarionov, A. G. Banshchikov, D. K. Polyushkin, S. Wachter, T. Knobloch, M. Thesberg, L. Mennel, M. Paur, M. Stöger-Pollach, A. Steiger-Thirsfeld et al., “Ultrathin calcium fluoride insulators for two-dimensional field-effect transistors,” Nature Electronics, vol. 2, no. 6, pp. 230–235, 2019.DOI
  33. [33]W. Zhao et al., “Field-based capacitance modeling for sub-65-nm on-chip interconnect,” IEEE Transactions on Electron Devices, vol. 56, no. 9, pp. 1862–1872, 2009, doi: 10.1109/TED.2009.2026162.
  34. [34]N. Ma and D. Jena, “Carrier statistics and quantum capacitance effects on mobility extraction in two-dimensional crystal semiconductor field-effect transistors,” 2D Materials, vol. 2, no. 1, p. 015003, 2015.DOI
  35. [35]A. Rahman, J. Guo, S. Datta, and M. Lundstrom, “Theory of ballistic nanotransistors,” IEEE Transactions on Electron Devices, vol. 50, no. 9, pp. 1853–1864, 2003, doi: 10.1109/TED.2003.815366.
  36. [36]F. Wu, H. Tian, Z. Yan, Y. Shen, J. Ren, Y. Yang, and T.-L. Ren, “Transistor subthreshold swing lowered by 2-d heterostructures,” IEEE Transactions on Electron Devices, vol. 68, no. 1, pp. 411–414, 2021, doi: 10.1109/TED.2020.3040350.DOI
  37. [37]M. Perrucchini, E. G. Marin, D. Marian, G. Iannaccone, and G. Fiori, “Physical insights into the operation of a 1-nm gate length transistor based on mos2 with metallic carbon nanotube gate,” Applied Physics Letters, vol. 113, no. 18, p. 183507, 2018, doi: 10.1063/1.5045281.
  38. [38]I. M. Datye, A. Daus, R. W. Grady, K. Brenner, S. Vaziri, and E. Pop, “Strain-enhanced mobility of monolayer mos2,” Nano Letters, vol. 22, no. 20, pp. 8052–8059, 2022, doi: 10.1021/acs.nanolett.2c01707.DOI
  39. [39]C. Song, F. Fan, N. Xuan, S. Huang, G. Zhang, C. Wang, Z. Sun, H. Wu, and H. Yan, “Largely tunable band structures of few-layer InSe by uniaxial strain,” ACS Applied Materials & Interfaces, vol. 10, no. 4, pp. 3994–4000, 2018, doi: 10.1021/acsami.7b17247.DOI
  40. [40]A. K. Shah, M. Si, P. D. Ye, and S. K. Gupta, “Polarization switching in Hf0.5Zr0.5O2-dielectric stack: The role of dielectric layer thickness,” Applied Physics Letters, vol. 119, no. 12, p. 122903, 09 2021, doi: 10.1063/5.0056448.
  41. [41]S. S. Cheema and Others, “Ultrathin ferroic hfo2–zro2 superlattice gate stack for advanced transistors,” Nature, vol. 604, no. 7904, pp. 65–71, Apr 2022, doi: 10.1038/s41586-022-04425-6.DOI
  42. [42]T. Agarwal, I. Radu, P. Raghavan, G. Fiori, A. Thean, M. Heyns, and W. Dehaene, “Effect of material parameters on two-dimensional materials based tfets: An energy-delay perspective,” in ESSCIRC Conference 2016: 42nd European Solid-State Circuits Conference, 2016, doi: 10.1109/ESSCIRC.2016.7598241. pp. 55–58.DOI
  43. [43]A. Naseer, K. Nandan, A. Agarwal, S. Bhowmick, and Y. S. Chauhan, “Di-metal chalcogenides: A new family of promising 2-d semiconductors for high-performance transistors,” IEEE Transactions on Electron Devices, vol. 70, no. 5, pp. 2445–2452, 2023, doi: 10.1109/TED.2023.3261831.DOI
  44. [44]K. Thakar, B. Rajendran, and S. Lodha, “Ultra-low power neuromorphic obstacle detection using a two-dimensional materials-based subthreshold transistor,” npj 2D Materials and Applications, vol. 7, no. 1, p. 68, Sep 2023, doi: 10.1038/s41699-023-00422-z.DOI

Paper details

vlnc:W_snj8wnyj
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents