Periodic Reporting for period 1 - F2GO (Fluorides for 2D Next-Generation Nanoelectronics)
Reporting period: 2022-09-01 to 2025-02-28
Microelectronic devices have become an inseparable part of our everyday life. In order to keep up with the increasing demand in performance, the development of these essential building blocks is coordinated in the IRDS roadmap. Conventional silicon technology faces the challenge that the silicon channel is difficult to scale below 12nm, as then the mobility of silicon degrades considerably. The recent IRDS roadmap highlights the potential of two-dimensional (2D) materials for scaling electronic devices down to the atomic limit as these materials can maintain a decent mobility even in the monolayer form. The ultimate goal is to achieve ultra-densely packed and ultra-fast devices with ultra-low power consumption for nanoelectronic applications of the 2030s. Various challenges exist, including identification of the 2D material with the highest mobility, reduction of the contact resistances, various fabrication challenges, and the lack of a suitable insulator which is required for the fabrication of field effect transistors (FETs).
Compared to the other challenges, relatively little attention has been paid to the search of insulators compatible with 2D channel materials. We suggested that ionic compounds such as fluorides may offer advantages in device performance and stability. Following our demonstration of ultrathin calcium fluoride (CaF2) as an insulator for 2D FETs, the F2GO project aims to advance 2D nanoelectronics by using fluoride-based insulators for advanced devices, enabling technologies like ultra-scaled non-volatile memory and steep slope devices for CMOS logic.
In the F2GO project, various fluorides are explored for their use in 2D nanoelectronics, starting with CaF2 and SrF2. Contrary to the prototype devices, which used CaF2 epitaxially grown on Si(111) substrates, which resulted in undesired space charge regions in the Si substrate, F2GO investigates the growth of fluorides on metallic substrates (for contacts) and 2D channel materials (for top gates). Once suitable insulator films are available, devices will be fabricated, characterized and modeled. All experimental efforts will be supported by theoretical calculations, starting from the ab initio level up to TCAD. These theoretical efforts will guide the experiments by explaining the observations, provide suggestions for different materials based on predicted permittivities and bandgaps, and eventually lead to better quality devices.
Compared to the other challenges, relatively little attention has been paid to the search of insulators compatible with 2D channel materials. We suggested that ionic compounds such as fluorides may offer advantages in device performance and stability. Following our demonstration of ultrathin calcium fluoride (CaF2) as an insulator for 2D FETs, the F2GO project aims to advance 2D nanoelectronics by using fluoride-based insulators for advanced devices, enabling technologies like ultra-scaled non-volatile memory and steep slope devices for CMOS logic.
In the F2GO project, various fluorides are explored for their use in 2D nanoelectronics, starting with CaF2 and SrF2. Contrary to the prototype devices, which used CaF2 epitaxially grown on Si(111) substrates, which resulted in undesired space charge regions in the Si substrate, F2GO investigates the growth of fluorides on metallic substrates (for contacts) and 2D channel materials (for top gates). Once suitable insulator films are available, devices will be fabricated, characterized and modeled. All experimental efforts will be supported by theoretical calculations, starting from the ab initio level up to TCAD. These theoretical efforts will guide the experiments by explaining the observations, provide suggestions for different materials based on predicted permittivities and bandgaps, and eventually lead to better quality devices.
The most urgent initial target of F2GO was to establish growth procedures for fluorides on metallic and 2D substrates using molecular beam epitaxy (MBE). Various metallic substrates were considered, including Cu, Ag, and Pt which should theoretically allow for commensurate epitaxial growth of CaF2 and SrF2. Various challenges were encountered during growth and most films only had poor surface coverage and in general were of poor quality. Similar observations were made during the deposition on graphene and PtSe2.
To overcome these challenges, other deposition methods were considered, for instance thermal evaporation (TE). TE has a much higher deposition rate compared to MBE and is not performed under ultra-high vacuum (UHV) conditions. Given its simplicity, TE would also be more compatible with industrial fab processing than MBE. In collaboration with Prof. Hongtao Yuan (Nanjing University) first test devices were fabricated. While the polycrystalline films do not have the same purity as MBE grown films, they covered the whole substrate with small surface roughness and were found to result in very promising device performance. Consequently, TE will be more closely explored during the next stages of F2GO, together with magnetron sputtering, another industry-compatible methodology.
In order to allow for a better understanding of the experimental results and to provide guidance for future experiments, e.g. related to the choice of the best fluoride, various theoretical calculations were performed. Using hybrid-DFT methods, we calculated the parameters of a large variety of CaF2 point defects, which are critical for predictive stability and reliability simulations of 2D devices. These DFT parameters where then used in our open-source 1D device simulator Comphy (www.comphy.eu) to calculate transfer characteristics as well as hysteresis of fluoride-based devices. Thereby, F2GO systematically combines experimental growth, device fabrication, computational modeling, and defect analysis to advance fluoride-based electronics, with promising results that point to necessary optimizations for device reliability and industry application potential.
To overcome these challenges, other deposition methods were considered, for instance thermal evaporation (TE). TE has a much higher deposition rate compared to MBE and is not performed under ultra-high vacuum (UHV) conditions. Given its simplicity, TE would also be more compatible with industrial fab processing than MBE. In collaboration with Prof. Hongtao Yuan (Nanjing University) first test devices were fabricated. While the polycrystalline films do not have the same purity as MBE grown films, they covered the whole substrate with small surface roughness and were found to result in very promising device performance. Consequently, TE will be more closely explored during the next stages of F2GO, together with magnetron sputtering, another industry-compatible methodology.
In order to allow for a better understanding of the experimental results and to provide guidance for future experiments, e.g. related to the choice of the best fluoride, various theoretical calculations were performed. Using hybrid-DFT methods, we calculated the parameters of a large variety of CaF2 point defects, which are critical for predictive stability and reliability simulations of 2D devices. These DFT parameters where then used in our open-source 1D device simulator Comphy (www.comphy.eu) to calculate transfer characteristics as well as hysteresis of fluoride-based devices. Thereby, F2GO systematically combines experimental growth, device fabrication, computational modeling, and defect analysis to advance fluoride-based electronics, with promising results that point to necessary optimizations for device reliability and industry application potential.
Using thermal evaporation, first promising devices with ultra-scaled fluoride insulators were fabricated. These devices will be thoroughly characterized, modeled and further refined. Various fluorides other than CaF2 and SrF2 will be considered to further improve device performance.
Recent updates in our gate stack analyzer Comphy also enable physics-based TCAD modeling of fluoride-based 2D-FETs, including detrimental features like hysteresis (clockwise and counter-clockwise) and bias temperature instability (BTI). While it is widely acknowledged that the amount of hysteresis is an indicator for insulator quality, the phenomenon is poorly understood. Published data is based on random and often undocumented experimental conditions like sweep time, sweep voltage range, and temperature. Based on a thorough experimental and theoretical study we demonstrate that the observed hysteresis is extremely sensitive to all these parameters and that published data are essentially random numbers that say little about the quality of the insulator. To overcome this unsatisfactory situation, a standardized hysteresis benchmarking protocol was developed to facilitate comparative analysis across a wide range of device prototypes, incorporating different equivalent oxide thicknesses (EOT). It was demonstrated that by appropriately choosing the bias conditions, the hysteresis normalized by EOT allows for a consistent comparison, and thus insulator quality, across different prototype devices. In addition, the next release of Comphy, which was extensively used and enhanced during this study, is currently being polished for publication (expected Spring 2025, www.comphy.eu). By using Comphy and the standardized hysteresis protocol, other researchers will be able to consistently compare their results with the results obtained in F2GO.
Recent updates in our gate stack analyzer Comphy also enable physics-based TCAD modeling of fluoride-based 2D-FETs, including detrimental features like hysteresis (clockwise and counter-clockwise) and bias temperature instability (BTI). While it is widely acknowledged that the amount of hysteresis is an indicator for insulator quality, the phenomenon is poorly understood. Published data is based on random and often undocumented experimental conditions like sweep time, sweep voltage range, and temperature. Based on a thorough experimental and theoretical study we demonstrate that the observed hysteresis is extremely sensitive to all these parameters and that published data are essentially random numbers that say little about the quality of the insulator. To overcome this unsatisfactory situation, a standardized hysteresis benchmarking protocol was developed to facilitate comparative analysis across a wide range of device prototypes, incorporating different equivalent oxide thicknesses (EOT). It was demonstrated that by appropriately choosing the bias conditions, the hysteresis normalized by EOT allows for a consistent comparison, and thus insulator quality, across different prototype devices. In addition, the next release of Comphy, which was extensively used and enhanced during this study, is currently being polished for publication (expected Spring 2025, www.comphy.eu). By using Comphy and the standardized hysteresis protocol, other researchers will be able to consistently compare their results with the results obtained in F2GO.