Cloud native EDA tools & pre-optimized hardware platforms
Study by IBM
Lanzillo et al., Comp. Mat. Sci. 158, 398 (2019)
____________________________________________________
IBM Research & IBM T. J. Watson used QuantumATK to evaluate the impact of oxygen contamination, i.e. oxidation of barrier Ta and liner Ru metals, on vertical resistance of Cu/Ta/Ru/Cu interfaces found in back-end-of-the-line interconnect stacks. DFT+NEGF simulations were performed to obtain density of states, electron transmission, reflection coefficients, and total resistance for each interface. This works reveals that oxidation of the barrier (Ta->TaO, Ta2O5) /liner (Ru->RuO2) materials increases vertical resistance and that it varies by as much as an order of magnitude depending on the degree of oxidation, and also the metal-to-metal contact area, i.e. the interconnect pitch size.
Study by IBM and Rensselaer Polytechnic Institute
Zhou et al., AIP Adv. 8, 055127 (2018).
____________________________________________________
IBM Research and Rensselaer Polytechnic Institute used QuantumATK to evaluate aluminum (Al) as an alternate conductor to Cu for the back-end-of-line (BEOL) interconnect technology. DFT+NEGF simulations were done to calculate electron transmission, conductance for different nanowire sizes either oxidized or not, resistivity and reflection coefficients due to grain boundary scattering, and vertical resistance across the Al/Ta(N)/Al interfaces. This work suggests that even though Al has a high ballistic conductance, Al is not a suitable Cu-replacement conductor due to larger grain boundary scattering and vertical resistance.
“These results suggest that the resistance of Al interconnect structures at fine dimensions is likely larger than that of Cu interconnects, despite semiclassical models predicting a smaller size effect in Al than in Cu."
Study by GlobalFoundries
Dixit et al., SISPAD 2018
____________________________________________________
GlobalFoundries used QuantumATK to investigate various scattering mechanisms (impurity scattering, interface resistance across different Ru/liner interfaces, grain boundary (GB) scattering) in ruthenium (Ru) interconnects leading to increased resistivity. DFT+NEGF simulations were carried out to calculate transmission spectra, resistance/resistivity and grain boundary reflection coefficients. In addition to providing insight into resistivity due to GB scattering, this study shows that carbon and oxygen substitutional defects lead to larger resistivity than vacancy or interstitial defects. Calculations also predict that the Ru/TiN (liner) interface has the smallest contact resistance and, therefore, that TiN is the most suitable liner for Ru interconnects.
“Our study thus provide a deeper understanding of the GB resistivity in polycrystalline Ru and have implications towards engineering the metal interconnect resistance in modern IC’s .”
Study by IBM
Lanzillo et al., IITC 2018.
____________________________________________________
IBM Research and IBM Systems and Technology employed QuantumATK to investigate the impact of materials choice on interconnect via resistance and device performance. DFT+NEGF simulations were performed to obtain Cu/Ta(N,O)/Cu via resistance as a function of via contact area. The study predicts that by changing from TaN to Ta one could reduce via resistance by ~26% (corresponding to 2 % device performance enhancement at 3nm), whereas oxidation of Ta layer could lead to 300% increase in via resistance (corresponding to 16% performance degradation). Interconnect line resistance’s dependence on line geometry variation and impact on device performance was estimated using the finite-element methodology.
“This work demonstrates the extreme sensitivity of device performance on material selection and geometry at fine dimensions. ”
Study by GlobalFoundries & IBM
Lanzillo et al., Appl. Phys. Lett. 112, 163107 (2018)
____________________________________________________
GlobalFoundries, IBM Research & IBM T. J. Watson used QuantumATK to evaluate the impact of replacing a thin TaN barrier with Ta in Cu/Ta(N)/Co/Cu and Cu/Ta(N)/Ru/Cu vias found in back-end-of-line interconnect stacks. DFT+NEGF simulations (with spin-polarization) were performed in order to obtain k-resolved transmission spectra, resistance/resistivity and atom-resolved projected density of states. The study shows that changing from TaN to Ta results in a significant reduction of the vertical resistance through the via structures and that the electron scattering at the Cu/Ta(N) interface is the dominant contribution to the total vertical resistance. Furthermore, the resistance trends predicted by QuantumATK and experiments are in a good agreement.
“This work further develops a fundamental understanding of electron transport and material characteristics in nano-sized interconnects.”
Study by GlobalFoundries & IBM
Lanzillo et al., J. Appl. Phys. 123, 154303 (2018)
____________________________________________________
GlobalFoundries and IBM Research employed QuantumATK to estimate the impact of defect and grain boundary scattering, and thus scaling, on specific resistivity of tungsten (W), which is often used as a middle-of-line (MOL) conductor. DFT+NEGF simulations were carried out in order to understand the electron transport properties and obtain the total specific resistivity, normalized specific resistivities due to various defects, specific resistivities and grain boundary reflection coefficients for nine representative high-symmetry twin grain boundaries. This work demonstrates that the impact of grain boundary scattering is roughly an order of magnitude larger than that of defect scattering. The calculated grain boundary reflection coefficients are in a good agreement with those measured experimentally.
“These results provide crucial insights for understanding the impact of scaling of W-based contacts between active devices and back-end-of-line interconnects in next-generation semiconductor technology.”
Study by GlobalFoundries
Singh et al., IEEE International Reliability Physics Symposium (IRPS) 2018
____________________________________________________
GlobalFoundries used QuantumATK to predict thermal properties of Cu/TaN/Co/Cu interfaces found in back-end-of-line interconnect stacks. DFT+NEGF simulations were performed to obtain transmission functions and electrical and heat currents as a function of electron energy. This study predicts a reduction in electrical and heat currents in Cu/TaN/Co/Cu vias, compared to bulk Cu, due to interface scattering. Results obtained with atomistic simulations are then used in large scale finite element TCAD simulations for further modeling of self-heating in interconnects.
“A bottom-up finite-element simulation methodology informed by atomistic simulations was developed to quantitatively resolve self-heating mechanisms without resorting to fitting or empirical models.”
Study by IBM Research
Lanzillo, J. Appl. Phys. 121, 175104 (2017)
____________________________________________________
IBM Research used QuantumATK to investigate four novel metal (Pt, Rh, Ir, and Pd) nanowires as back-end-of-line (BEOL) interconnects. DFT cohesive energy (1) simulations were carried out to evaluate the structural integrity and resistance to electromigration, while DFT+NEGF electron transport simulations were performed to obtain conductance (2) for different sizes, orientations of nanowires affected by surface scattering and grain boundary reflection coefficients & specific resistivity (3) for different grain boundaries due to grain boundary scattering. This work shows that properties (1) and (2) of Pt, Rh, and Ir nanowires are superior to Cu. However, Cu outperforms the other metals in (3). Specific resistivities for Cu grain boundaries are in a good agreement with experimentally measured ones.
“This work demonstrates that several novel metals have the potential to offer superior performance relative to Cu for the next generation interconnect technology”
Study by GlobalFoundries
Dixit et al., J. Phys. D: Appl. Phys. 50, 455103 (2017)
____________________________________________________
Global Foundries used QuantumATK as part of a multi-scale modeling approach to determine the critical thickness of the TiN layer to prevent Co diffusion within interconnects. DFT+NEB (nudged elastic band) simulations were performed to calculate defect formation energies and activation barriers along different minimum energy paths for Co diffusion through crystalline TiN, through amorphous TiN, and Co migration from the Co layer into the TiN metal across the Co/TiN interface. Parameters obtained with QuantumATK were used in kinetic Monte Carlo (KMC) simulations with Sentaurus to estimate the diffusion lengths as a function of temperature and time. This study predicts that a roughly 3 nm thick TiN layer is needed to avoid Co diffusion in middle-of-line (MOL) and back-end-of-line (BEOL) stacks.
“Such a multi-scale modelling approach yields an exact critical thickness of the metal layer sufficient to prevent the Co diffusion in IC interconnects.”
Study by GlobalFoundries & IBM Research
Dixit et al., IEEE Trans. Elec. Dev. 64, 3775 (2017)
____________________________________________________
Global Foundries & IBM Research employed QuantumATK to provide atomic-level insight into the interface between semiconducting Ge and metallic TiGe, used in contacts for sub-10 nm nodes, and give directions for how to minimize the contact resistance at the interface. DFT+NEGF was employed to calculate Schottky barrier height (SBH), I-V curves and contact resistance for the TiGe/Ge interface. The results show that the calculated SBH depends strongly on the phase of TiGe and on the different crystallographic orientations of Ge, highlighting the paramount importance of a detailed atomic-scale description of the interface in this case. The study also demonstrates that by increasing the doping density in the semiconductor, it is possible to reduce the contact resistance and convert the Schottky barrier into Ohmic, which is crucial for sub-10 nm nodes. A number of calculated properties such as the value of the Ge band gap, the SBH, as well as the trends of the current-voltage (I-V) curves and of the contact resistance with doping are in a good agreement with experimentally determined values.
“Such an approach allows for a direct comparison between the theory and experiment and provides useful engineering insights toward the reduction of the contact resistance.”
Study by Imec, University of Antwerp, KULeuven, MathAM-OIL and Synopsys QuantumATK Team
Pourtois et al., ESC Trans. 80, 303 (2017)
____________________________________________________
Imec and collaborators used QuantumATK to investigate the doping-dependence of contact resistance and intrinsic processes that limit the resistance of the TiSi (amorphous)/Si contact. DFT+NEGF was used to calculate I-V curves, local density of states and contact resistance as a function of the doping concentration . This work demonstrates that the intrinsic contact resistivity saturates with the doping concentration. Furthermore, it is shown that as the doping concentration increases, the impact of interface composition decreases and the contact resistance starts to be dominated by intrinsic properties of the metal and the semiconductor, such as their effective masses and Fermi energies. Authors also suggest that metals with a high effective mass, such as, for example, Sc, could be used to maximize the transmission probability of electron injection and thus reduce the metal-semiconductor resistance.
“ In this contribution, we report a fundamental study of the factors that set the contact resistivity between metals and highly doped semiconductors.”
Study by IBM Research & GlobalFoundries
Gluchenkov et al., International Electron Devices Meeting (IEDM) 2016.
____________________________________________________
IBM Research and GlobalFoundries employed QuantumATK as part of a combined multi-scale ab initio + TCAD modeling and experimental approach to investigate the TiGe2/Ge contact. DFT+NEGF was used to calculate local density of states for the interface. This study suggests that introducing interface defects results in Fermi level pinning near the valence band for Ge, thus reducing the Schottky barrier height and the unipolar heterojunction resistances.
“This results in clear Fermi level pinning near the valence band for Ge, consistent with the experimental data”
Study by GlobalFoundries & Synopsys
Park et al., International Electron Devices Meeting (IEDM) 2017
____________________________________________________
GlobalFoundries together with Synopsys used QuantumATK as part of the Sentaurus Materials Workbench (SMW) workflow to study dopant B, P, and As diffusion in Si, SiGe and Ge. DFT with nudged elastic band (NEB) simulations were carried out to calculate defect formation energies and migration barriers. The latter were used as input for generation of model parameters (such as diffusivities of dopant-defect pairs and pairing coefficients in all relevant charge states) for Sentaurus Process. The study provides a recipe for a good workflow to model dopant diffusion in semiconductors: calibrate DFT against measured data and use DFT to obtain TCAD parameters and their relations which are not available from experiments.
“ It is demonstrated that DFT could help to get unknown TCAD parameters and provide valuable insights on the parameter relations supporting experimental data.”
Study by Imec, PLASMANT and Synopsys QuantumATK Team
Sankaran et al., Phys. Rev. B 94, 094424 (2016)
____________________________________________________
Imec and collaborators employed QuantumATK to investigate the impact of the CoFe ferromagnetic layer thickness and the presence or absence of the paramagnetic Ta capping layer on the tunnel magnetoresistance ratio (TMR) in Ta/CoFe/MgO magnetic tunnel junctions (MTJs). DFT+NEGF with spin-polarization was used not only to calculate conductance and TMR, but also density of states, bandstructure and work functions to explain the observed TMR trends. The study reveals strong oscillations of the TMR amplitude with different CoFe thicknesses, when combined with thick MgO tunneling barriers. Furthermore, the TMR oscillates even more in the presence of the Ta capping layer. These results are important for evaluating different MTJ stacks for spintronic memory applications.
“Understanding both the role of the crystalline CoFe ferromagnetic electrode thickness and the impact of the paramagnetic capping layers could help the scientific and engineering communities in their search for high TMR. “
Study by Fujitsu Ltd. and Fujitsu Laboratories Ltd.
Harada et al., Appl. Phys. Expr. 10, 074001 (2017)
____________________________________________________
Fujitsu used QuantumATK to analyze a p-n heterojunction backward diode based on edge-modified graphene nanoribbons (GNRs) for millimeter- or terahertz-wave detection applications. DFT calculations were performed to obtain the electronic structure of AGNRs terminated with hydrogen (H) or fluorine (F). Then, DFT+NEGF was used to calculate current-voltage characteristics for a F-GNR (n-doped)/H-GNR (p-doped) heterojunction diode, revealing an exponential-like current under forward bias and nonlinear current under reverse bias. The calculated junction resistance and curvature coefficient were used to obtain important performance parameters of the detector diode, such as noise equivalent power and the matched voltage sensitivity. Based on the calculated parameters, the authors conclude that GNRs are a promising candidate material for high-frequency diodes.
“In this study, we aimed to clarify the potential capabilities and performances of backward diodes built with GNRs via theoretical methods.”
Study by University of Leuven, IMEC, University of Antwerp, PLASMANT, and TSMC
Lu et al., Appl. Phys. Lett. 108, 043504 (2016)
____________________________________________________
Imec and collaborators used QuantumATK to investigate the impact of scaling of the channel length on the performances of metal-oxide-semiconductor field effect transistors (MOSFETs) based on 2D materials. DFT+NEGF simulations were carried out not only to calculate current-voltage (I-V) characteristics and subthreshold slope, but also density of states plots and electrostatic potential profiles to find the origin of the device performance degradation due to scaling. This work shows that having shorter than 10 nm results in strong device performance degradations due to the electron tunneling process. Furthermore, this process and thus device performance can be modulated by different effective masses of various 2D materials (MoS2, WS2, HfS2, ZrS2, YCl3) and their phases (2H, 1T).
“Our findings thus highlight that the selection criteria of 2D channel materials should not only be driven by improving the carrier mobility but also by reducing the source to drain electron tunneling effect in aggressively scaled MOSFETs.”
Interested in applying QuantumATK software to your research? Test our software or contact us at quantumatk@synopsys.com to get more information on QuantumATK platform for atomic-scale modeling.