Next Article in Journal
Dynamic Optimal Power Flow of Active Distribution Network Based on LSOCR and Its Application Scenarios
Next Article in Special Issue
An Improved Structure Enabling Hole Erase Operation When Using an IGZO Channel in a 3D NAND Flash Structure to Which COP (Cell-On-Peri) Structure Is Applied
Previous Article in Journal
Acoustic Feedback Cancellation Algorithm for Hearing Aids Based on a Weighted Error Adaptive Filter
Previous Article in Special Issue
Low-Frequency Noise Modeling of 4H-SiC Metal-Oxide-Semiconductor Field-Effect Transistors
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Investigation of Electro-Thermal Performance for TreeFET from the Perspective of Structure Parameters

1
College of Electronics and Information Engineering, Shanghai University of Electric Power, Shanghai 200090, China
2
Radiawave Technologies Corporation Limited, Shenzhen 518172, China
*
Author to whom correspondence should be addressed.
Electronics 2023, 12(7), 1529; https://doi.org/10.3390/electronics12071529
Submission received: 23 February 2023 / Revised: 20 March 2023 / Accepted: 22 March 2023 / Published: 24 March 2023
(This article belongs to the Special Issue Advanced CMOS Devices)

Abstract

:
In this work, the electro-thermal properties of TreeFET, which combines vertically stacked nanosheet (NS) and fin-shaped interbridge (IB) channels, are investigated in terms of interbridge width (WIB), nanosheet space (SNS) and nanosheet width (WNS) by TCAD simulation. Electrical characteristics such as electron density distributions, on/off-state current (ION, IOFF), subthreshold swing (SS) and self-heating effects (SHE) such as lattice temperature and thermal resistance (Rth) are systematically studied to optimize the performance of TreeFET. The result shows that a smaller WIB mitigates the short-channel effects and increases the electron concentration in NS channels but increases thermal resistance. A larger SNS increases the on-state current while compensating for the gate drive loss and mitigating the thermal coupling effect between NS channels but results in longer conduction paths of carriers and heat, which hinders further improvements. Moreover, a suitable WNS is required to lessen the decline of gate controllability induced by IB channels. Hence, suitable geometry parameters should be selected to achieve a compromise between thermal and electrical performance.

1. Introduction

In order to satisfy the scaling-down demand and to mitigate the short channel effect (SCE), silicon-based multi-gate devices such as FinFETs, GAAFETs, Forksheets and CFETs have been proposed [1,2,3,4,5]. In pursuit of low power dissipation (LP) and high performance (HP) targets as the technology nodes scale down, GAAFETs with unique gate structures are expected to become mainstream devices [6,7,8]. Stacked nanosheet GAAFETs (NSFETs) are one of the most compelling candidates for GAAFETs due to their excellent gate controllability, design flexibility with variable WNS and high drive current per footprint [9,10,11,12]. To further alleviate the impact of SCE on NSFETs, the overlap structure is used by introducing a barrier between the channel and source/drain, which reduces IOFF [13].
To improve performance beyond NSFETs, a novel TreeFET device that combines FinFETs and stacked nanosheets was proposed [14]. By using interbridge channels, TreeFETs provide a higher drive current than NSFETs without the additional footprint. Furthermore, TreeFETs are compatible with NSFETs manufacturing process by simply not etching the sacrificial layers completely to form the interbridge, and Chien-Te Tu et al. successfully demonstrated the feasibility and outstanding performance of GeSi-Ge TreeFET in the experiment [15]. Modified devices based on TreeFETs have been reported, such as Si channel tree-shaped junctionless NSFET, Si channel tree-type reconfigurable FETs and Si-SiGe channel Fishbone FETs [16,17,18]. Due to the confined geometry structure of TreeFET, where the NS and IB channels are fully surrounded by the low thermal conductivity gate dielectric, SHE should be thoroughly investigated [19,20,21]. In addition, the increased power density aggravates the heat dissipation of TreeFET.
Since TreeFETs are potential devices for design requirements, it should be of great interest to investigate and co-optimize the electro-thermal performance from the perspective of structural parameters in terms of WIB, SNS and WNS. In this work, the device structure, simulation setup and calibration efforts with the experimental data are given in Section 2. In Section 3, the influence of WIB, SNS and WNS on the electro-thermal performance is thoroughly investigated. Finally, the conclusion is given in Section 4.

2. Device Structure and Simulation Calibration

2.1. Device Structure

Figure 1 shows the structure of the Si-channel TreeFET, and its parameters are listed in Table 1. The simulation was based on the Sentaurus TCAD and performed under the condition of VGS = VDS = 0.7 V. Nanosheet thickness (TNS) was fixed at 5 nm. The source/drain was modeled as a cuboid. TiN with flexible and controllable work function was used as the gate metal [22]. The source/drain contact resistance was set to 1 × 10−9 Ω cm−2 [23]. Under the condition of the same EOT (Equivalent oxide thickness), devices with high-k materials suffer from poor heat dissipation due to their small physical thickness and reduced carrier mobility; SiO2 was chosen as the gate dielectric [24,25]. In order to reduce the impact of random doping fluctuations in the channel on the electrical performance of the device, the doping concentration of the channel and source/drain region is 1 × 1015 cm−3 and 1 × 1021 cm−3, respectively, and bulk region doping with 5 × 1018 cm−3 was used to suppress the subthreshold leakage [26].
The thermal conductivity of different regions, according to the published work, is shown in Table 2 [27]. SHE was investigated in terms of lattice temperature and Rth. The ambient temperature was assumed to be 300 K for the simulated devices under normal operations. Affected by phonon boundary scattering and interface traps, the hetero-interfacial-thermal resistance (HITR) of Si/SiO2 was set as 2 × 10−4 cm2K/W [28,29]. Referring to the thermal resistance of BEOL at different VIA densities in 14 nm FinFET, the initial boundary thermal resistance at the gate, source and drain were 2 × 10−6 cm2K/W, and the conventional measurement configuration was set up, and the substrate was set up as the main heat dissipation path [30].

2.2. Calibration

In order to ensure the accuracy of the simulation and calculation, the density gradient model was adopted for the quantum confinement effect. The Philips unified mobility model was used to account for the effects of electron–hole scattering, lattice scattering and ionized impurity scattering. Thin-layer mobility model and enhanced Lombardi model were adopted to account for the effects of phonon boundary scattering and surface roughness scattering. The extended Canali model was considered to account for degraded mobility under the effects of high-field velocity saturation. Moreover, the Shockley–Read–Hall recombination model with doping dependence is adopted, and the thermodynamic model and diffusion–drift model were considered for SHE.
The experimental data used for calibration come from the IBM Stacked Nanosheet Gate-All-Around Transistor, and the calibration data are shown in Figure 2 [9]. For I–V calibration to reflect the true performance measurement of the data, the work function of the metal gate and the saturation velocity was adjusted to match the experimental data in the subthreshold region and saturation region. Note that the TreeFET without IB channels (NSFET) was used in the calibration. After the Id-Vg curve was accurately matched with the experimental data, the IB channels were added for subsequent experiments.

3. Investigation of Electrical and Thermal Characteristics

3.1. Impact of IB Width

The impact of WIB on the electrical characteristics with different SNS at WNS = 20 nm is illustrated in Figure 3. ION and IOFF with various WIB are shown in Figure 3a,b, respectively. At each fixed SNS, it is observed that ION rises with the decrease in WIB, and IOFF increases with WIB increasing. The wider channel width reduces the potential barrier height in the conduction band energy of the channel, resulting in an increase in IOFF. The ON/OFF current ratio (defined as log10 (ION/IOFF)) and SS under different WIBs are shown in Figure 3c,d. ION/IOFF is larger, with improved SS for smaller WIB. A narrow WIB helps to improve the inversion in the channels, which results in the improvement of the gate control ability. However, TreeFET with smaller WIB suffers more surface scattering and quantum confinement effects, which impedes performance improvement.
Figure 4 shows the electron density of channel cross sections with different WIB at 3/5/7 nm. Note that the trend of electron density in channel cross sections with different widths is similar for different SNS; SNS = 15 nm is selected to demonstrate the variation in electron density. As indicated in Figure 4, although the wider WIB has a larger channel cross-sectional area, which reduces the equivalent resistance in the IB channel region, the distance between the side gate and the connections of NSs and IBs increases, and the gate at the corners needs to control both IBs and NSs, resulting in a decrease in the gate control capability and the electron density of the inversion layer in connections and corners. Moreover, it was found that the electron density of the center of the middle NS is seriously degraded due to the connection of IBs, compared with the top and bottom NSs.
TreeFET with WIB = 3 nm has better gate control ability, but its IB channels turn on later than TreeFET with WIB = 5 nm due to the quantum confinement effect and more surface roughness scattering. The threshold voltage (Vth) is extracted by the constant current criterion (Icon = 100 nA). Figure 5 shows the electron density of channel cross sections with different WIB = 3/5 nm when SNS = 15 nm at different overdrive voltages (VOV, VOV = VGS − Vth). As shown in Figure 5a, when VOV = 0.1 V, the electron density of NS channels in TreeFET with WIB = 3 nm is larger than that of IB channels. Moreover, the electron density of IB channels is lower compared to TreeFET with WIB = 5 nm, which implies that TreeFET with smaller WIB has a higher threshold voltage. Note that the connection of middle NS at WIB = 3 nm has a larger electron density than that of WIB = 5 nm. As the gate voltage increases to VOV = 0.5 V in Figure 5b, more electron concentrates in all channels, and the electron density of 3 nm IB channels exceeds that of 5 nm IB channels because of better gate control.
The total line electron density (Ned) is obtained by integrating the electron density over the entire channel region. As shown in Figure 6, the Ned is compared with different WIBs. As WIB increases from 3 nm to 5 nm, Ned drops by 19.4%, and the electron density in the IB channel region decreases from 27.6% to 25.3% of the total channel region. Therefore, to achieve good electrical performance, it is necessary to maintain a WIB of 5 nm or less.

3.2. Impact of NS Space

Figure 7 shows the impact of SNS on the electrical characteristics with different WIB at WNS = 20 nm. As shown in Figure 7a,b, both ION and IOFF increase with the increase in SNS at fixed WIB, which is mainly attributed to the increased cross-sectional area of the channel region. Note that the increasing trend of ION slows down with the increase in SNS and Slope1 > Slope2 > Slope3. It can be seen from Figure 7c that when the WIB is smaller than 5 nm, the ON/OFF current ratio shows an upward trend with the increase in SNS, and when the WIB is larger than 5 nm, the trend is the opposite. This phenomenon can be attributed to the fact that the increase in SNS makes the increase in IOFF much higher than ION when WIB is larger than 5 nm. Figure 7d shows that SS decreases with the increase in SNS due to the better gate control ability in IBs under the condition of higher SNS.
Figure 8 shows the electron density of channel cross sections with different SNS at 10/20/40 nm at VGS = VDS = 0.7 V, and Figure 9a shows the Ned is compared when WNS = 20 nm. Figure 10 illustrates the electric field of channel cross sections at SNS = 10/20/40 nm, and the inset shows the short and long carrier paths at different SNS, respectively. A median WIB = 5 nm is chosen in Figure 8, Figure 9 and Figure 10. It can be observed that with the increase in SNS, more electrons concentrate in the top NS and both IBs, and Ned rises slightly in the middle NS while remaining almost invariable in the bottom NS and even drops at SNS = 40 nm. The IB channels have a larger contact area with the spacer and the source/drain (S/D) as SNS increases, which has smaller extension resistance, contact resistance, surface scattering and a better gate control capability, which leads to a larger Ned. Figure 9b illustrates the proportion of IB current to the total current and total NS current. Note that the total current is the sum of the total NS current and total IB current. The increase in SNS makes the channel area increase and resistances decrease. It can be observed that the IB current ratio is 17.4%/33.8%/54.1% at SNS = 10/20/40 nm, respectively. The total NS current rises by 14% with the increase in SNS from 10 nm to 20 nm but decreases by 11.5% from 20 nm to 40 nm. However, as shown in Figure 10, as SNS continues to increase, the S/D parasitic series resistance (RS/D) increases due to the increased height of the S/D region, the voltage of the bottom channel is divided, and the top channel has a larger potential difference than the bottom channel and has a higher electric field intensity [31]. This is the reason why the Ned of the upper part in IB is higher than the lower part, and the Ned of the bottom NS is reduced at SNS = 40 nm. Moreover, the channel with a long distance from the S/D contact further hinders its performance improvement due to its longer carrier path and increased RS/D, resulting in the uptrends of ION in Figure 7a and IB current ratio in Figure 9b slowing down with the increase in SNS and causing the total NS current to decrease at SNS = 40 nm.

3.3. Trade-off of Electro-Thermal Performance

The temperature distribution of N-type NSFET and TreeFET is shown in Figure 11, and Tpeak is defined as the peak temperature of the channel cross section. It can be seen from Figure 11a,b that Tpeak locates at the channel region near the drain side. The electric field intensity at the drain side is larger at VGS = VDS = 0.7 V. Moreover, confined by the gate dielectric material with low thermal conductivity (KSiO2 = 1.4 WK−1 m−1) and Si/SiO2 thermal boundary resistance, maximum lattice temperature (TL, MAX) is usually at the boundary between the channel and drain. Therefore, the TreeFET, as a similar structure to the NSFET, which has a larger ION and more heat generation per footprint, suffers severe self-heating effects.
Figure 12a,b show TL, MAX and Rth (defined by Rth = (TL, MAX − Tambient)/(VDS × IDS), Tambient = 300 K) of different SNS at WIB = 3/4/5 nm. It can be seen that TL, MAX increases with the increase in SNS and the decrease in WIB, which is mainly due to the increase in ION. As the WIB decreases, the thermal resistance increases due to the intensified phonon-boundary scattering, while as the WIB increases, the cross-sectional area of the IB increases, which improves the heat dissipation towards BEOL and increases the heat flux exchange among channels.
As SNS increased, Rth rapidly decreased and then increased. Part of the heat is conducted through S/D to the BEOL; the substrate is the main heat dissipation path, and the other part is conducted through the gate dielectric layer, metal gate and spacer area between the channel layer. Thus, a severe thermal coupling effect exists [32,33,34]. The thermal coupling effect is weakened with the increase in SNS due to the increase in IB, which connects NSs and conducts heat. Moreover, the increasing temperature degrades the thermal conductivity of Silicon material. Heat is mainly dissipated through the bulk and substrate. As the SNS continues to increase, the overall thermal conduction path of the channel becomes longer, which causes worse heat dissipation and results in more total heat generating in the top channel and the increase in Rth, as shown in Figure 11c and Figure 12b, respectively.
Figure 13 shows the thermal properties of the DC power of the device. Figure 13a shows the curve of ΔT and DC power of TreeFET under different IB parameters, where ΔT = TL, MAX − Tambient, P = VDS × IDS, and the slope represents the thermal resistance of the device at this power. It can be observed in Figure 13a that TreeFET with a larger WIB and SNS has a smaller thermal resistance. Moreover, with the increase in power, lattice temperature rises, and the phonon scattering intensifies, which leads to the degradation of the thermal conductivity of silicon material. Figure 13b–e show the thermal distribution of TreeFET with different WIB and SNS at the same power of 80 μw. As shown in Figure 13b,c, with the same WIB = 3 nm, TreeFET with SNS = 5 nm has a higher lattice temperature. Additionally, as the SNS increases to 20 nm, the thermal coupling effect is weakened, and the self-heating effect is improved, which results in a lower lattice temperature. As shown in Figure 13c,d, when the WIB increases from 3 nm to 5 nm, the heat dissipation capability of the device is enhanced, the thermal resistance is reduced, and the lattice temperature is decreased. Note that with the increase in SNS, the Tpeak of the device gradually moves towards the center of the device. Figure 13d,e show that as the SNS increases from 20 nm to 40 nm, the average heat dissipation path of the channel increases, and the heat dissipation capacity degrades.
In summary, a wide WIB is favorable for heat dissipation capability, and a narrow WIB has a better gate control capability but suffers from a greater quantum confinement effect. A suitable SNS is key to achieving good electro-thermal performance, but a too-large SNS also leads to an increase in thermal resistance and a low space utilization where an additional nanosheet can be stacked as an alternative. Therefore, SNS should be 20–30 nm to reach the compromise of electro-thermal performance.

3.4. Impact of NS Width

NSFET has a continuous range of active widths for additional design flexibility. TreeFET retains the unique advantages of NSFET, which can maximize the effective channel width (Weff) in the same footprint.
The simulations with various WIB and SNS under the conditions of different WNS (WNS = 10/20/30/40 nm) are investigated. All devices require narrower and higher IB to maintain a good gate control capability. WSW = WIB/WNS is defined as the specific width. Figure 14 shows the schematic diagram of the specific width and NS total line electron density degradation. As shown in Figure 14a, the electron concentration in the NS channels is more sensitive to the change in WIB with a smaller WNS and a larger WSW. TreeFET with a larger WSW and a smaller WNS at the same WIB has a larger proportion of electron concentration degradation in the NS channels shown in Figure 14b.
Figure 15 shows ION, total IB electron density, total NS electron density and Rth with different SNS under the condition of different WNS = 10/20/30/40 nm and WIB = 4 nm. Note that ION and total NS electron density are normalized by the WNS. As shown in Figure 15a,b, the total IB electron density is less affected by WNS and TreeFET, with smaller WNS having larger ION per footprint due to better gate control capability and additional IB channels. Figure 15c shows the impact of the normalized total NS electron density with the different IB channels. As SNS increases, the degradation of electron density improves. For WNS = 10 nm, although it has the highest total NS electron density without IB, the presence of the IB channels severely impairs the gate control capability of NS channels, leading to a sharp drop in total NS electron density. As shown in Figure 15d, the trend of thermal resistance with SNS and WNS is similar to that shown in Figure 12b. With the increase in WNS, the cross-sectional areas of the channel regions increase, which enhances the heat dissipation capacity of the device and decreases Rth. The heat dissipation capacity of the device tends to be gradually saturated, which is less affected by the IB channel and SNS. The thermal resistance Rth of the device decreases with the increase in WNS, and the heat dissipation capability of the device gradually tends to be saturated, which is less affected by variation in WIB and SNS.
Figure 16 shows the cutline in the center of the middle NS channel and the distribution of electron density along the cutline with different WNS under SNS = 20 nm. It can be seen that since the middle NS channel is connected to two IB channels, the electron density at the center of the NS channel is significantly reduced due to the reduction in the gate control capability. Moreover, the peak of electron density in the center of the NS channel of the device with WNS = 10 nm is smaller compared with TreeFET with WNS = 20/40 nm due to the larger WSW, which also explains the phenomenon of Figure 15c. Therefore, devices with different WNSs also require a suitable SNS to maintain the electro-thermal properties.
In summary, TreeFET has a higher normalized current per footprint when the WNS is small, but the IB channels seriously deteriorate the gate control ability of the NS channel. Therefore, WNS should be at least greater than 10 nm, and WSW should be small. When the WNS is larger, the thermal resistance is lower, and the gate control capability of NS channels is weakened, which has little influence on the IB channels. WNS provides a choice to make trade-offs between thermal properties and electrical performance.

4. Conclusions

In this work, TreeFET with different geometry parameters is systematically investigated by TCAD simulation. The investigation of the electro-thermal properties of the device is discussed by adjusting WIB, SNS and WNS. The results show that a smaller WIB and a higher SNS improve the SCE, which avoids the gate control capability degradation of the NS channel. On the other hand, a higher SNS increases ION and improves the heat dissipation capability, but longer carrier paths for bottom-side channels and heat conduction paths for middle-side channels hinder further improvements in electro-thermal performance. In addition, a suitable WNS and WSW are required to maintain high electron density in NS channels. To pursue better electro-thermal performance, WIB should be 5 nm or less, SNS should be 20–30 nm, and WNS should be at least larger than 10 nm. These characteristics indicate that the TreeFET can become a promising alternative device for further performance optimizations, and the systematic investigation lays a foundation for design requirements. The reasonable geometric parameters are expected to be selected flexibly in different applications.

Author Contributions

Conceptualization, X.P.; methodology, X.P.; software, X.P.; validation, X.P.; formal analysis, X.P.; investigation, X.P.; data curation, X.P.; writing—original draft, X.P.; writing—review and editing, X.P., J.L. and W.L.; visualization, X.P. and J.L.; supervision, W.L. and Q.L. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (Grant Nos. 52177185 and 62174055).

Data Availability Statement

The data and code are available from the corresponding authors upon reasonable request.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Hisamoto, D.; Lee, W.-C.; Kedzierski, J.; Takeuchi, H.; Asano, K.; Kuo, C.; Anderson, E.; Tsu-Jae, K.; Bokor, J.; Chenming, H. FinFET-a self-aligned double-gate MOSFET scalable to 20 nm. IEEE Trans. Electron Devices 2000, 47, 2320–2325. [Google Scholar] [CrossRef]
  2. Yakimets, D.; Eneman, G.; Schuddinck, P.; Bao, T.H. Vertical GAAFETs for the Ultimate CMOS Scaling. IEEE Trans. Electron Devices 2015, 62, 1433–1439. [Google Scholar] [CrossRef]
  3. Singh, N.; Agarwal, A.; Bera, L.K.; Liow, T.Y.; Yang, R.; Rustagi, S.C.; Tung, C.H.; Kumar, R.; Lo, G.Q.; Balasubramanian, N.; et al. High-performance fully depleted silicon nanowire (diameter/spl les/5 nm) gate-all-around CMOS devices. IEEE Electron Device Lett. 2006, 27, 383–386. [Google Scholar] [CrossRef]
  4. Weckx, P.; Ryckaert, J.; Litta, E.D.; Yakimets, D.; Matagne, P.; Schuddinck, P.; Jang, D.; Chehab, B.; Baert, R.; Gupta, M.; et al. Novel forksheet device architecture as ultimate logic scaling device towards 2 nm. In Proceedings of the 2019 IEEE International Electron Devices Meeting (IEDM), San Francisco, CA, USA, 7–11 December 2019; pp. 36.5.1–36.5.4. [Google Scholar] [CrossRef]
  5. Ryckaert, J.; Schuddinck, P.; Weckx, P.; Bouche, G.; Vincent, B.; Smith, J.; Sherazi, Y.; Mallik, A.; Mertens, H.; Demuynck, S.; et al. The Complementary FET (CFET) for CMOS scaling beyond N3. In Proceedings of the 2018 IEEE Symposium on VLSI Technology, Honolulu, HI, USA, 18–22 June 2018; pp. 141–142. [Google Scholar] [CrossRef]
  6. Yang, B.; Buddharaju, K.D.; Teo, S.H.G.; Singh, N.; Lo, G.Q.; Kwong, D.L. Vertical Silicon-Nanowire Formation and Gate-All-Around MOSFET. IEEE Electron Device Lett. 2008, 29, 791–794. [Google Scholar] [CrossRef]
  7. Kuhn, K.J. Considerations for Ultimate CMOS Scaling. IEEE Trans. Electron Devices 2012, 59, 1813–1828. [Google Scholar] [CrossRef]
  8. Ritzenthaler, R.; Mertens, H.; Pena, V.; Santoro, G.; Chasin, A.; Kenis, K.; Devriendt, K.; Mannaert, G.; Dekkers, H.; Dangol, A.; et al. Vertically Stacked Gate-All-Around Si Nanowire CMOS Transistors with Reduced Vertical Nanowires Separation, New Work Function Metal Gate Solutions, and DC/AC Performance Optimization. In Proceedings of the 2018 IEEE International Electron Devices Meeting (IEDM), San Francisco, CA, USA, 1–5 December 2018; pp. 21.5.1–21.5.4. [Google Scholar] [CrossRef]
  9. Loubet, N.; Hook, T.; Montanini, P.; Yeung, C.-W.; Kanakasabapathy, S.; Guillom, M.; Yamashita, T.; Zhang, J.; Miao, X.; Wang, J.; et al. Stacked nanosheet gate-all-around transistor to enable scaling beyond FinFET. In Proceedings of the 2017 Symposium on VLSI Technology, Kyoto, Japan, 5–8 June 2017; pp. T230–T231. [Google Scholar] [CrossRef]
  10. Bae, G.; Bae, D.-I.; Kang, M.; Hwang, S.M.; Kim, S.S.; Seo, B.; Kwon, T.Y.; Lee, T.J.; Moon, C.; Choi, Y.M.; et al. 3nm GAA Technology featuring Multi-Bridge-Channel FET for Low Power and High Performance Applications. In Proceedings of the 2018 IEEE International Electron Devices Meeting (IEDM), San Francisco, CA, USA, 1–5 December 2018; pp. 28.7.1–28.7.4. [Google Scholar] [CrossRef]
  11. Kim, S.; Kim, M.; Ryu, D.; Lee, K.; Kim, S.; Lee, J.; Lee, R.; Kim, S.; Lee, J.-H.; Park, B.-G. Investigation of Electrical Characteristic Behavior Induced by Channel-Release Process in Stacked Nanosheet Gate-All-Around MOSFETs. IEEE Trans. Electron Devices 2020, 67, 2648–2652. [Google Scholar] [CrossRef]
  12. Kim, S.; Guillorn, M.; Lauer, I.; Oldiges, P.; Hook, T.; Na, M. Performance trade-offs in FinFET and gate-all-around device architectures for 7nm-node and beyond. In Proceedings of the 2015 IEEE SOI-3D-Subthreshold Microelectronics Technology Unified Conference (S3S), Rohnert Park, CA, USA, 5–8 October 2015; pp. 1–3. [Google Scholar] [CrossRef]
  13. Trivedi, V.; Fossum, J.G.; Chowdhury, M.M. Nanoscale FinFETs with gate-source/drain underlap. IEEE Trans. Electron Devices 2005, 52, 56–62. [Google Scholar] [CrossRef]
  14. Ye, H.-Y.; Liu, C.W. On-Current Enhancement in TreeFET by Combining Vertically Stacked Nanosheets and Interbridges. IEEE Electron Device Lett. 2020, 41, 1292–1295. [Google Scholar] [CrossRef]
  15. Tu, C.-T.; Hsieh, W.-H.; Huang, B.-W.; Chen, Y.-R.; Liu, Y.-C.; Tsai, C.-E.; Chueh, S.-J.; Liu, C.W. Experimental Demonstration of TreeFETs Combining Stacked Nanosheets and Low Doping Interbridges by Epitaxy and Wet Etching. IEEE Electron Device Lett. 2022, 43, 682–685. [Google Scholar] [CrossRef]
  16. Valasa, S.; Tayal, S.; Thoutam, L.R. An Intensive Study of Tree-Shaped JL-NSFET: Digital and Analog/RF Perspective. IEEE Trans. Electron Devices 2022, 69, 6561–6568. [Google Scholar] [CrossRef]
  17. Sun, Y.; Li, X.; Liu, Z.; Liu, Y.; Li, X.; Shi, Y. Vertically Stacked Nanosheets Tree-Type Reconfigurable Transistor with Improved ON-Current. IEEE Trans. Electron Devices 2022, 69, 370–374. [Google Scholar] [CrossRef]
  18. Cao, L.; Zhang, Q.; Luo, Y.; Gu, J.; Gan, W.; Lu, P.; Yao, J.; Xu, H.; Zhao, P.; Luo, K.; et al. Novel Channel-First Fishbone FETs With Symmetrical Threshold Voltages and Balanced Driving Currents Using Single Work Function Metal Process. IEEE Trans. Electron Devices 2022, 69, 5971–5977. [Google Scholar] [CrossRef]
  19. Kim, H.; Son, D.; Myeong, I.; Kang, M.; Jeon, J.; Shin, H. Analysis on Self-Heating Effects in Three-Stacked Nanoplate FET. IEEE Trans. Electron Devices 2018, 65, 4520–4526. [Google Scholar] [CrossRef]
  20. Myeong, I.; Song, I.; Kang, M.J.; Shin, H. Self-Heating and Electrothermal Properties of Advanced Sub-5-nm Node Nanoplate FET. IEEE Electron Device Lett. 2020, 41, 977–980. [Google Scholar] [CrossRef]
  21. Tsen, C.; Chung, C.-C.; Liu, C.W. Self-Heating Mitigation of TreeFETs by Interbridges. IEEE Trans. Electron Devices 2022, 69, 4123–4128. [Google Scholar] [CrossRef]
  22. Liu, Y.; Kijima, S.; Sugimata, E.; Masahara, M.; Endo, K.; Matsukawa, T.; Ishii, K.; Sakamoto, K.; Sekigawa, T.; Yamauchi, H.; et al. Investigation of the TiN Gate Electrode with Tunable Work Function and Its Application for FinFET Fabrication. IEEE Trans. Nanotechnol. 2006, 5, 723–730. [Google Scholar] [CrossRef]
  23. Niimi, H.; Liu, Z.; Gluschenkov, O.; Mochizuki, S.; Fronheiser, J.; Li, J.; Demarest, J.; Zhang, C.; Liu, B.; Yang, J.; et al. Sub-10−9 Ωcm−2 n-Type Contact Resistivity for FinFET Technology. IEEE Electron Device Lett. 2016, 37, 1371–1374. [Google Scholar] [CrossRef]
  24. Balestra, F.; Ghibaudo, G. Physics and performance of nanoscale semiconductor devices at cryogenic temperatures. Semicond. Sci. Technol. 2017, 32, 23002. [Google Scholar] [CrossRef]
  25. Park, J.-Y.; Lee, B.-H.; Chang, K.S.; Kim, D.U.; Jeong, C.; Kim, C.-K.; Bae, H.; Choi, Y.-K. Investigation of Self-Heating Effects in Gate-All-Around MOSFETs With Vertically Stacked Multiple Silicon Nanowire Channels. IEEE Trans. Electron Devices 2017, 64, 4393–4399. [Google Scholar] [CrossRef]
  26. Mertens, H.; Ritzenthaler, R.; Hikavyy, A.; Kim, M.S.; Tao, Z.; Wostyn, K.; Chew, S.A.; de Keersgieter, A.; Mannaert, G.; Rosseel, E.; et al. Gate-all-around MOSFETs based on vertically stacked horizontal Si nanowires in a replacement metal gate process on bulk Si substrate. In Proceedings of the 2016 IEEE Symposium on VLSI Technology, Honolulu, HI, USA, 14–16 June 2016; pp. 1–2. [Google Scholar] [CrossRef]
  27. Liu, R.; Li, X.; Sun, Y.; Shi, Y. A Vertical Combo Spacer to Optimize Electrothermal Characteristics of 7-nm Nanosheet Gate-All-Around Transistor. IEEE Trans. Electron Devices 2020, 67, 2249–2254. [Google Scholar] [CrossRef]
  28. Venkateswarlu, S.; Nayak, K. Hetero-Interfacial Thermal Resistance Effects on Device Performance of Stacked Gate-All-Around Nanosheet FET. IEEE Trans. Electron Devices 2020, 67, 4493–4499. [Google Scholar] [CrossRef]
  29. Pop, E.; Dutton, R.; Goodson, K. Thermal analysis of ultra-thin body device scaling [SOI and FinFet devices]. In Proceedings of the IEEE International Electron Devices Meeting 2003, Washington, DC, USA, 8–10 December 2003; pp. 36.6.1–36.6.4. [Google Scholar] [CrossRef]
  30. Yan, J.-Y.; Jan, S.-R.; Peng, Y.-J.; Lin, H.H.; Wan, W.K.; Huang, Y.H.; Hung, B.; Chan, K.T.; Huang, M.; Yang, M.; et al. Thermal resistance modeling of back-end interconnect and intrinsic FinFETs, and transient simulation of inverters with capacitive loading effects. In Proceedings of the 2016 IEEE International Electron Devices Meeting (IEDM), San Francisco, CA, USA, 3–7 December 2016; pp. 35.6.1–35.6.4. [Google Scholar] [CrossRef]
  31. Venkateswarlu, S.; Badami, O.; Nayak, K. Electro-Thermal Performance Boosting in Stacked Si Gate-all-Around Nanosheet FET with Engineered Source/Drain Contacts. IEEE Trans. Electron Devices 2021, 68, 4723–4728. [Google Scholar] [CrossRef]
  32. Cai, L.; Chen, W.; Du, G.; Zhang, X.; Liu, X. Layout Design Correlated with Self-Heating Effect in Stacked Nanosheet Transistors. IEEE Trans. Electron Devices 2018, 65, 2647–2653. [Google Scholar] [CrossRef]
  33. Cai, L.; Chen, W.; Du, G.; Kang, J.; Zhang, X.; Liu, X. Investigation of self-heating effect on stacked nanosheet GAA transistors. In Proceedings of the 2018 International Symposium on VLSI Technology, Systems and Application (VLSI-TSA), Hsinchu, Taiwan, 16–19 April 2018; pp. 1–2. [Google Scholar] [CrossRef]
  34. Liu, R.; Li, X.; Sun, Y.; Li, F.; Shi, Y. Thermal Coupling Among Channels and Its DC Modeling in Sub-7-nm Vertically Stacked Nanosheet Gate-All-Around Transistor. IEEE Trans. Electron Devices 2021, 68, 6563–6570. [Google Scholar] [CrossRef]
Figure 1. (a) Three-dimensional schematic of TreeFET; (b) cut plane I cross section along channel length (X-axis); (c) cut plane II (Y-axis).
Figure 1. (a) Three-dimensional schematic of TreeFET; (b) cut plane I cross section along channel length (X-axis); (c) cut plane II (Y-axis).
Electronics 12 01529 g001
Figure 2. Id-Vg calibration of 3-stacked nanosheet structure with the SHE considered.
Figure 2. Id-Vg calibration of 3-stacked nanosheet structure with the SHE considered.
Electronics 12 01529 g002
Figure 3. Impact of WIB: (a) ION; (b) IOFF; (c) ION/IOFF; (d) SS of the TreeFET with different SNS at VGS = VDS = 0.7 V and WNS = 20 nm at VGS = VDS = 0.7 V.
Figure 3. Impact of WIB: (a) ION; (b) IOFF; (c) ION/IOFF; (d) SS of the TreeFET with different SNS at VGS = VDS = 0.7 V and WNS = 20 nm at VGS = VDS = 0.7 V.
Electronics 12 01529 g003
Figure 4. Electron density of channel cross sections with WIB = 3/5/7 nm (form left to right) at SNS = 15 nm at VGS = VDS = 0.7 V.
Figure 4. Electron density of channel cross sections with WIB = 3/5/7 nm (form left to right) at SNS = 15 nm at VGS = VDS = 0.7 V.
Electronics 12 01529 g004
Figure 5. Electron density of channel cross sections with WIB = 3/5 nm (from left to right) at SNS = 15 nm at (a) VOV = 0.1 V and (b) VOV = 0.5 V.
Figure 5. Electron density of channel cross sections with WIB = 3/5 nm (from left to right) at SNS = 15 nm at (a) VOV = 0.1 V and (b) VOV = 0.5 V.
Electronics 12 01529 g005
Figure 6. Total line electron density (Ned) of different parts of channels with different WIB at VGS = VDS = 0.7 V.
Figure 6. Total line electron density (Ned) of different parts of channels with different WIB at VGS = VDS = 0.7 V.
Electronics 12 01529 g006
Figure 7. Impact of SNS: (a) ION; (b) IOFF; (c) ION/IOFF; and (d) SS of the TreeFET with different WIB at VGS = VDS = 0.7 V.
Figure 7. Impact of SNS: (a) ION; (b) IOFF; (c) ION/IOFF; and (d) SS of the TreeFET with different WIB at VGS = VDS = 0.7 V.
Electronics 12 01529 g007
Figure 8. Electron density of channel cross sections with SNS = 10/20/40 nm (from left to right) at WIB = 5 nm at VGS = VDS = 0.7 V.
Figure 8. Electron density of channel cross sections with SNS = 10/20/40 nm (from left to right) at WIB = 5 nm at VGS = VDS = 0.7 V.
Electronics 12 01529 g008
Figure 9. (a) Total line electron density (Ned) of different parts of channels with different SNS; (b) the IB current ratio=the total IB current/ the total current; total NS current.
Figure 9. (a) Total line electron density (Ned) of different parts of channels with different SNS; (b) the IB current ratio=the total IB current/ the total current; total NS current.
Electronics 12 01529 g009
Figure 10. Electric field of channel cross sections with SNS = 10/20/40 nm (from left to right) at WIB = 5 nm and VGS = VDS = 0.7 V. The inset shows the short and long carrier paths at different SNS, respectively.
Figure 10. Electric field of channel cross sections with SNS = 10/20/40 nm (from left to right) at WIB = 5 nm and VGS = VDS = 0.7 V. The inset shows the short and long carrier paths at different SNS, respectively.
Electronics 12 01529 g010
Figure 11. (a) Thermal distribution of 3-stacked nanosheet FET/TreeFET (SNS = 10/20 nm) at WIB = 5 nm; (b) lattice temperature distribution curve along channel length of middle NS; (c) thermal distribution of channel cross sections with WIB = 4 nm, SNS = 40 nm at VGS = VDS = 0.7 V.
Figure 11. (a) Thermal distribution of 3-stacked nanosheet FET/TreeFET (SNS = 10/20 nm) at WIB = 5 nm; (b) lattice temperature distribution curve along channel length of middle NS; (c) thermal distribution of channel cross sections with WIB = 4 nm, SNS = 40 nm at VGS = VDS = 0.7 V.
Electronics 12 01529 g011
Figure 12. Impact of SNS on (a) TL, MAX; (b) Rth of the TreeFET with different WIB at VGS = VDS = 0.7 V.
Figure 12. Impact of SNS on (a) TL, MAX; (b) Rth of the TreeFET with different WIB at VGS = VDS = 0.7 V.
Electronics 12 01529 g012
Figure 13. (a) DC power-ΔT curve; thermal distribution of TreeFETs at 80 μW with (b) SNS = 5 nm, WIB = 3 nm; (c) SNS = 20 nm, WIB = 3 nm; (d) SNS = 20 nm, WIB = 5 nm; (e) SNS = 40 nm, WIB = 5 nm.
Figure 13. (a) DC power-ΔT curve; thermal distribution of TreeFETs at 80 μW with (b) SNS = 5 nm, WIB = 3 nm; (c) SNS = 20 nm, WIB = 3 nm; (d) SNS = 20 nm, WIB = 5 nm; (e) SNS = 40 nm, WIB = 5 nm.
Electronics 12 01529 g013
Figure 14. (a) Schematic diagram of the specific width WSW; (b) NS total line electron density degradation, 0 represents a nanosheet without IB.
Figure 14. (a) Schematic diagram of the specific width WSW; (b) NS total line electron density degradation, 0 represents a nanosheet without IB.
Electronics 12 01529 g014
Figure 15. Impact of SNS: (a) normalized ION; (b) total IB line electron density; (c) normalized total NS line electron density; (d) Rth of the TreeFET with different WNS at WIB = 4 nm and VGS = VDS = 0.7 V.
Figure 15. Impact of SNS: (a) normalized ION; (b) total IB line electron density; (c) normalized total NS line electron density; (d) Rth of the TreeFET with different WNS at WIB = 4 nm and VGS = VDS = 0.7 V.
Electronics 12 01529 g015
Figure 16. (a) Cutline of middle NS; (b) electron density along NS channel with different WNS at SNS = 20 nm and WIB = 4 nm when VGS = VDS = 0.7 V.
Figure 16. (a) Cutline of middle NS; (b) electron density along NS channel with different WNS at SNS = 20 nm and WIB = 4 nm when VGS = VDS = 0.7 V.
Electronics 12 01529 g016
Table 1. Device parameters.
Table 1. Device parameters.
SymbolParameterValue
LGGate length12 nm
TNSNanosheet thickness5 nm
WNSNanosheet width10–40 nm
SNSNanosheet space5–40 nm
WIBInterbridge width3–7 nm
LSDSource/drain length13 nm
LSPSpacer length5 nm
TBulkBulk thickness100 nm
EOTEquivalent oxide thickness1 nm
CGPContacted gate pitch48 nm
NchChannel doping1015 cm−3
NSDSource/drain doping1021 cm−3
NBulkBulk doping5 × 1018 cm−3
Table 2. Thermal properties.
Table 2. Thermal properties.
MaterialThermal Conductivity (WK−1 m−1)
Gate metal (TiN)19.2
Bulk region (Si)148
Source/Drain region (Si)16.61
Channel region (Si)8.07
Spacer (Si3N4)18.5
SiO21.4
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Liu, W.; Pan, X.; Liu, J.; Li, Q. Investigation of Electro-Thermal Performance for TreeFET from the Perspective of Structure Parameters. Electronics 2023, 12, 1529. https://doi.org/10.3390/electronics12071529

AMA Style

Liu W, Pan X, Liu J, Li Q. Investigation of Electro-Thermal Performance for TreeFET from the Perspective of Structure Parameters. Electronics. 2023; 12(7):1529. https://doi.org/10.3390/electronics12071529

Chicago/Turabian Style

Liu, Weijing, Xinfu Pan, Jiangnan Liu, and Qinghua Li. 2023. "Investigation of Electro-Thermal Performance for TreeFET from the Perspective of Structure Parameters" Electronics 12, no. 7: 1529. https://doi.org/10.3390/electronics12071529

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop