Final Report Summary - MOWSES (Nanoelectronics based on two-dimensional dichalcogenides)
Partner TCD, represented by the Nicolosi group, has established wet synthetic methodologies for the production of several new types of semiconducting layered crystals, including GaS and GaSe, InS and InSe , ReSe2 nanosheets and nanotubes. We have developed liquid-phase metholodologies for the exfoliation of such layered crystals, as well as for transition metal dichalcogenides (i.e. MoS2). Furthermore, we have developed a set of practices and processes for semi-industrial-scale deposition methodologies based on ultrasonic spray for the deposition of suitable arge-scale thin films for electronic applications as the enabling step in exploring silicon-like applications as well as flexible electronics. We have developed methodologies for the fabrication of the thinnest Cu interconnects ever reported (3-atom-wide wires) grown on black phosphorous nanosheets. We have implemented methodologies for the atom-by-atom electron microscopy characterization of systems synthetized, exfoliated and fabricated within MoWSeS.
Partner JSI focused on what parameters influence the electronic properties of TMDs. We found that in multilayers excitons efficiently dissociate into charges, which are subsequently trapped at defects or flake edges. The electron dynamics proved remarkably robust against the use of surfactants in the liquid phase exfoliation (LPE). Spectroscopic monitoring of LPE enabled us to disperse monolayer contents exceeding 70% and the direct exfoliation into a polymer solution, yielding fluorescent TMD-polymer composites. In monolayers, exciton dissociation has a much lower efficiency, which we found can be significantly increased by applying a moderate in-plane electric field. A transverse electric field, on the other hand, provides an efficient modulation of the absorption spectrum, enabling an unprecedentedly compact, energy-efficient electromodulator device for integrated photonics.
Partner EPFL was working on the realization of high-performance devices on flexible devices and devices based on nanopatterned TMDC materials. We have successfully fabricated transistor arrays based on CVD-grown monolayer MoS2 showing performance similar to the state-of-the based on exfoliated materials. These transistor arrays were then integrated with deformable substrates. For stretching, PDMS with low elastic modulus value was chosen as the substrate because it could be applied large strain before breaking. We found that devices remained functional up to a 4% of tensile strain with on-off ratio staying at the same order of magnitude and mobility remaining at 67% of its original value. For bending tests, polyimide substrate is used as the flexible substrate. Bottom-gate staggered structure transistors have been implemented. Devices remain well functioning with the on-off ratio reduction from 105 to 104. Electrical measurements on MoS2 in confined geometries e.g. nanoribbons, quantum point contacts or quantum dots were also successfully carried out. The realization of a quantum point contact demonstrating conductance quantization in multiples of 2e2/h was achieved. Subbands are clearly resolved as step-like plateaus in the conductance curve. This is the first demonstration of a quantum point contact in single layer MoS2. The realization of the constrictions is achieved electrostatically due to the large bandgap of MoS2, which, unlike in the case of graphene, does not require etching of the material. Further investigation of the evolution of the conductance in magnetic field can pave the way towards creation of spin polarized states and their application in quantum computing devices.
Partner Jacobs, in collaboration with Partner SCM, has generated the methodological basis for the simulation of electronic properties and transport properties. We are now capable to calculate ballistic transport through junctions composed of any material, which is a significant step forward to the status at the beginning of the project. The code, based on the non-equilibrium Green’s function (NEGF) technique, is already commercially available via partner SCM in ADF2014, and a graphical user interface (GUI) is available since 2017. For all developed methods there is a second, command-line based support-free and warranty-less version available that is be freely distributed.
Among the most striking results of our work we have explained why giant spin-orbit splittings are present in MoWSeS monolayers, but disappear completely in bilayers. One can reintroduce them, however, by applying an electric field, e.g. by a gate voltage. Moreover, we have shown that MoWSeS monolayers are electronically insensitive gate voltages, contrary to bilayers that show a linear band gap reduction as function of the field. We have also investigated straintronic effects [Ghorbani-Asl et al., Phys. Rev. B 87, 235434 (2013)] and the magnetization of MoS2 nanoribbons in presence of external electric and magnetic field [Kou et al., J. Phys. Chem. Lett. 3, 2394 (2012)]. According to WP2 and WP4 the focus of the project was set on the wet-chemical deposition of 2D TMD nanomaterials aiming towards the fabrication of TFTs for investigations on charge transport phenomena. Dip-, spray-, spin-coating and drop-casting processes were investigated. Liquid-exfoliated MoS2-dispersions from Trinity College Dublin (TCD) and deposition of MoS2-flakes from precursor materials ((NH4)2MoS4 in water) were tested. Flakes exceeding 150 µm lateral size and 2-5 ML thicknesswere obtained from precursor solution-based dip-coating process by taking Landau-Levich3 mechanism into account. Drop-casting and spray-coating were found to be suitable techniques to achieve large surface coverage of the samples coated with dispersed MoS2-flakes. Furthermore, the impact of the annealing conditions on the quality of the flakes was determined. The samples were characterizes by means of AFM, SEM, optical microscopy and Raman spectroscopy.
One of the most striking examples for the power of the NEGF-DFTB approach is the simulation of logical elements, as for example diodes. We have carried out such simulations for noble metal dichalcogenide mono- and bilayers.
We have simulated the structural and electronic properties of alloyed TMDC, were both transition metal (Mo, W, Nb) and chalcogen (S, Se) were varied. We found that alloys of Group 6 TMDC in almost all compositions are thermodynamically stable and have very similar electronic structure as averages of their pure phases.
We studied spin-orbit interactions as a very interesting means to tailor the spin and valley properties in TMDC. Giant spin orbit splittings in Group 6 TMDC monolayer have been known and studied intensively. Due to symmetry increase, i.e. the presence of inversion symmetry with inversion centres inbetween the layers, spin-orbit splittings are absent in the bilayers. Those can be re-introduced, though, by an external electric field. Thus, we have demonstrated a possibility to switch valley spin polarization on and off.
In order to arrive to the state of predictive theory, calculations need to be closely compared to experimental results. This has been possible for strain-dependent electronic properties, in particular for the band gap. We could demonstrate very close agreement for the strain-dependent decrease of the band gap in MoS2 and MoSe2.
Partner Weizman’s (group of Reshef Tenne) goal was to develop new knowhow which will convey these nanoparticles towards electronic applications. To advance this goal we have established three objectives: 1. Synthesize new nanotubes from different 2D compounds, which may offer new electrical properties; 2. To control the work function of WS2 and MoS2 nanotubes (INT)/fullerene-like (IF) nanoparticles via controlled doping with electrnically active impurities such as Re and Nb; 3. To fabricate electronic devices based on single nanotube. In fact we have made major scientific and technological progress in all three fronts, which is reflected by the publications/patent applications and presentations in different international conferences. Concomitantly, three students who were on the payload of the project and few others who made contributions in this important aspects of our work received highly valuable training which will help them to become independent researchers wherever they chose to go. Therefore I can definitely state that the contribution of our team to this project was valuable and fulfilled our expectations.
The goal behind SCM’s work in MoWSeS was to carry out theoretical developments and to implement them into software products, enabling atomistic simulations on the electronic
structure of 2D MoWSeS. In particular, we aimed to:
a) Improve the computational performance of the DFTB, LDA+U and NEGF methods in order to allow the study of systems of the size of ~10,000 atoms as needed for this project.
b) Study the dependence of electronic structure and electron mobility on the presence of dopants/defects.
c) Model optical properties of MoWSeS nanomaterials using time-dependent DFT (TDDFT) and vibrational spectra.
A code for computing quantum conductance based on non-equilibrium Green’s function and on the basis of DFTB-Hamiltonians was developed and made available to the community via SCM’s ADF2014 release. The code has also been used successfully for modelling changes in the electronic properties of TMD materials under mechanical deformation.
A new TDDFT algorithm to calculate both absorption UV-VIS spectra and Circular Dichroism (CD) spectra from the dynamical polarizability has also been implemented into the ADF2017 release, allowing the simulation of spectra for 2D (surface) models.
The time-dependent current-density functional theory (TD-CDFT) approach, already available for one- and three-dimensional materials, was extended to model optical response properties, like the dielectric function or the electric susceptibility, for two-dimensional, non-conducting materials. Therefore, a new approach to evaluate integration weights for single-orbital transitions in reciprocal space and an approximation for the unit cell volume in real-space depending on the ground state density was developed and implemented in SCM’s periodic DFT module, BAND.
The new method was applied to one-, two- and three-dimensional MoS2 materials. Thereby, the comparability of optical response properties for one- and two-dimensional, as well as two- and three-dimensional materials was verified. Moreover, the anomalous behaviour of the imaginary part of the dielectric function with respect to the number of MoS2 layers, ranging from one to ten layers, was reproduced qualitatively. This was previously not possible by means of a time-dependent density functional approach.
The computational performance was enhanced by introducing an eigenvalue difference dependent selection rule to neglect single-orbital transitions and at the same time a transition number dependent switching between integral evaluation in a Bloch expanded atomic orbital or a Bloch expanded crystal orbital representation. This approximation can speed-up the calculation times by a factor of up to 10. The optical response properties do depend on the energy range of the transitions, their real part experiencing a constant shift over the sampled frequency interval.
Partner Evonik developed two novel methods for deposition and growth of MoS2 based nanoscale transistor arrays for application in flexible electronics. Fabrication techniques are suitable for up-scaling to larger substrates and keep the fabrication cost low by avoiding expensive vacuum processing steps. We have investigated two independent techniques: the first uses exfoliated MoS2 flake dispersions, and the second focuses on bottom-up growth of MoS2 films through solution phase of molybdenum precursors.
Top-down approach by using liquid-phase exfoliated MoS2 flakes obtained from Trinity College Dublin (TCD) were further optimized at Evonik facilities for various coating/printing techniques (such as spin-coating, dip-coating, spray-coating, drop-casting, ink-jet printing and slot-die coating). Solvent systems and additional additives were carefully chosen to be suitable for deposition method, and at the same time to avoid aggregation of exfoliated flakes and formation of crystallites. In the next step, modified formulations with MoS2 flakes were successfully deposited on flexible substrates (i.e. PI and PEN foils) as an active material for fabrication of thin film transistors. Pre-treatment of substrate surface was necessary to control the drying behaviour of the printed ink. Electrical characterization of fabricated devices with several monolayer thick MoS2 film showed only minor conductivity directly after layer deposition. By depositing thicker films, large improvement in electrical characteristics was observed, but without visible field effect in those devices.
Alternatively, bottom-up approach using solution type Mo-precursor annealed in presence of sulfur was performed to form MoS2 thin films. Uniform and homogenous films with thickness down to 4 nm were obtained by combination of spin-coating and high temperature annealing. The fabrication process is wafer scalable with precursor concentration defining the thickness of the MoS2 layers, thereby making it suitable for different applications. The obtained MoS2 films were characterized by microscopy (TEM and SEM), Raman and X-ray photoelectron spectroscopy. High crystallinity and film homogeneity over large area with an average grain size of 100 nm were obtained. Thin film transistors fabricated based on solution processed MoS2 films demonstrated p-type semiconducting behaviour. Bottom-gate top-contact transistors exhibited the field effect mobility of 0.03 cm2/Vs. Due to excess sulfur and polycrystalline nature of the grown MoS2 film, electrical performance of fabricated devices was strongly dominated by high grain-to-grain resistance. Further improvements in growth of MoS2 films may increase the field effect mobility of thin film transistors. Together with low-cost and scalable solution-based fabrication processes, this novel deposition method shall promote the application of dichalcogenides in future nanoelectronic devices.
Partner Aixtron studied the direct formation of the 2D TMDC MoS2 via MOVPE. The deposition is realized in an AIXTRON horizontal hot-wall reactor, initially configured for growing nitride semi¬conductors. Due to the 10 × 2" set-up of the reactor, the simultaneous growth on different substrates such as sapphire, p-Si as well as AlN and GaN buffer structures on sapphire is possible. Molybdenum hexacarbonyl (MCO) and di tert-butyl sulphide (DTBS) are used as metal-organic precursors, for molybdenum and sulphur respectively. The developed process leads to a uniform; wafer-scale deposition of MoS2 on various substrate types. The deposited films mainly consists of MoS2 bilayers and exhibit a very high initial nucleation density. With optimization of growth parameters, a crystal growth process closer to thermodynamical equilibrium can be achieved. A set of experiments are conducted to investigate the nucleation of the films and to further tune nucleation density and lateral growth rate. The target is to deposit coherent wafer-scale monolayer MoS2 films for future electronic and optoelectronic applications.
Partner JSI focused on what parameters influence the electronic properties of TMDs. We found that in multilayers excitons efficiently dissociate into charges, which are subsequently trapped at defects or flake edges. The electron dynamics proved remarkably robust against the use of surfactants in the liquid phase exfoliation (LPE). Spectroscopic monitoring of LPE enabled us to disperse monolayer contents exceeding 70% and the direct exfoliation into a polymer solution, yielding fluorescent TMD-polymer composites. In monolayers, exciton dissociation has a much lower efficiency, which we found can be significantly increased by applying a moderate in-plane electric field. A transverse electric field, on the other hand, provides an efficient modulation of the absorption spectrum, enabling an unprecedentedly compact, energy-efficient electromodulator device for integrated photonics.
Partner EPFL was working on the realization of high-performance devices on flexible devices and devices based on nanopatterned TMDC materials. We have successfully fabricated transistor arrays based on CVD-grown monolayer MoS2 showing performance similar to the state-of-the based on exfoliated materials. These transistor arrays were then integrated with deformable substrates. For stretching, PDMS with low elastic modulus value was chosen as the substrate because it could be applied large strain before breaking. We found that devices remained functional up to a 4% of tensile strain with on-off ratio staying at the same order of magnitude and mobility remaining at 67% of its original value. For bending tests, polyimide substrate is used as the flexible substrate. Bottom-gate staggered structure transistors have been implemented. Devices remain well functioning with the on-off ratio reduction from 105 to 104. Electrical measurements on MoS2 in confined geometries e.g. nanoribbons, quantum point contacts or quantum dots were also successfully carried out. The realization of a quantum point contact demonstrating conductance quantization in multiples of 2e2/h was achieved. Subbands are clearly resolved as step-like plateaus in the conductance curve. This is the first demonstration of a quantum point contact in single layer MoS2. The realization of the constrictions is achieved electrostatically due to the large bandgap of MoS2, which, unlike in the case of graphene, does not require etching of the material. Further investigation of the evolution of the conductance in magnetic field can pave the way towards creation of spin polarized states and their application in quantum computing devices.
Partner Jacobs, in collaboration with Partner SCM, has generated the methodological basis for the simulation of electronic properties and transport properties. We are now capable to calculate ballistic transport through junctions composed of any material, which is a significant step forward to the status at the beginning of the project. The code, based on the non-equilibrium Green’s function (NEGF) technique, is already commercially available via partner SCM in ADF2014, and a graphical user interface (GUI) is available since 2017. For all developed methods there is a second, command-line based support-free and warranty-less version available that is be freely distributed.
Among the most striking results of our work we have explained why giant spin-orbit splittings are present in MoWSeS monolayers, but disappear completely in bilayers. One can reintroduce them, however, by applying an electric field, e.g. by a gate voltage. Moreover, we have shown that MoWSeS monolayers are electronically insensitive gate voltages, contrary to bilayers that show a linear band gap reduction as function of the field. We have also investigated straintronic effects [Ghorbani-Asl et al., Phys. Rev. B 87, 235434 (2013)] and the magnetization of MoS2 nanoribbons in presence of external electric and magnetic field [Kou et al., J. Phys. Chem. Lett. 3, 2394 (2012)]. According to WP2 and WP4 the focus of the project was set on the wet-chemical deposition of 2D TMD nanomaterials aiming towards the fabrication of TFTs for investigations on charge transport phenomena. Dip-, spray-, spin-coating and drop-casting processes were investigated. Liquid-exfoliated MoS2-dispersions from Trinity College Dublin (TCD) and deposition of MoS2-flakes from precursor materials ((NH4)2MoS4 in water) were tested. Flakes exceeding 150 µm lateral size and 2-5 ML thicknesswere obtained from precursor solution-based dip-coating process by taking Landau-Levich3 mechanism into account. Drop-casting and spray-coating were found to be suitable techniques to achieve large surface coverage of the samples coated with dispersed MoS2-flakes. Furthermore, the impact of the annealing conditions on the quality of the flakes was determined. The samples were characterizes by means of AFM, SEM, optical microscopy and Raman spectroscopy.
One of the most striking examples for the power of the NEGF-DFTB approach is the simulation of logical elements, as for example diodes. We have carried out such simulations for noble metal dichalcogenide mono- and bilayers.
We have simulated the structural and electronic properties of alloyed TMDC, were both transition metal (Mo, W, Nb) and chalcogen (S, Se) were varied. We found that alloys of Group 6 TMDC in almost all compositions are thermodynamically stable and have very similar electronic structure as averages of their pure phases.
We studied spin-orbit interactions as a very interesting means to tailor the spin and valley properties in TMDC. Giant spin orbit splittings in Group 6 TMDC monolayer have been known and studied intensively. Due to symmetry increase, i.e. the presence of inversion symmetry with inversion centres inbetween the layers, spin-orbit splittings are absent in the bilayers. Those can be re-introduced, though, by an external electric field. Thus, we have demonstrated a possibility to switch valley spin polarization on and off.
In order to arrive to the state of predictive theory, calculations need to be closely compared to experimental results. This has been possible for strain-dependent electronic properties, in particular for the band gap. We could demonstrate very close agreement for the strain-dependent decrease of the band gap in MoS2 and MoSe2.
Partner Weizman’s (group of Reshef Tenne) goal was to develop new knowhow which will convey these nanoparticles towards electronic applications. To advance this goal we have established three objectives: 1. Synthesize new nanotubes from different 2D compounds, which may offer new electrical properties; 2. To control the work function of WS2 and MoS2 nanotubes (INT)/fullerene-like (IF) nanoparticles via controlled doping with electrnically active impurities such as Re and Nb; 3. To fabricate electronic devices based on single nanotube. In fact we have made major scientific and technological progress in all three fronts, which is reflected by the publications/patent applications and presentations in different international conferences. Concomitantly, three students who were on the payload of the project and few others who made contributions in this important aspects of our work received highly valuable training which will help them to become independent researchers wherever they chose to go. Therefore I can definitely state that the contribution of our team to this project was valuable and fulfilled our expectations.
The goal behind SCM’s work in MoWSeS was to carry out theoretical developments and to implement them into software products, enabling atomistic simulations on the electronic
structure of 2D MoWSeS. In particular, we aimed to:
a) Improve the computational performance of the DFTB, LDA+U and NEGF methods in order to allow the study of systems of the size of ~10,000 atoms as needed for this project.
b) Study the dependence of electronic structure and electron mobility on the presence of dopants/defects.
c) Model optical properties of MoWSeS nanomaterials using time-dependent DFT (TDDFT) and vibrational spectra.
A code for computing quantum conductance based on non-equilibrium Green’s function and on the basis of DFTB-Hamiltonians was developed and made available to the community via SCM’s ADF2014 release. The code has also been used successfully for modelling changes in the electronic properties of TMD materials under mechanical deformation.
A new TDDFT algorithm to calculate both absorption UV-VIS spectra and Circular Dichroism (CD) spectra from the dynamical polarizability has also been implemented into the ADF2017 release, allowing the simulation of spectra for 2D (surface) models.
The time-dependent current-density functional theory (TD-CDFT) approach, already available for one- and three-dimensional materials, was extended to model optical response properties, like the dielectric function or the electric susceptibility, for two-dimensional, non-conducting materials. Therefore, a new approach to evaluate integration weights for single-orbital transitions in reciprocal space and an approximation for the unit cell volume in real-space depending on the ground state density was developed and implemented in SCM’s periodic DFT module, BAND.
The new method was applied to one-, two- and three-dimensional MoS2 materials. Thereby, the comparability of optical response properties for one- and two-dimensional, as well as two- and three-dimensional materials was verified. Moreover, the anomalous behaviour of the imaginary part of the dielectric function with respect to the number of MoS2 layers, ranging from one to ten layers, was reproduced qualitatively. This was previously not possible by means of a time-dependent density functional approach.
The computational performance was enhanced by introducing an eigenvalue difference dependent selection rule to neglect single-orbital transitions and at the same time a transition number dependent switching between integral evaluation in a Bloch expanded atomic orbital or a Bloch expanded crystal orbital representation. This approximation can speed-up the calculation times by a factor of up to 10. The optical response properties do depend on the energy range of the transitions, their real part experiencing a constant shift over the sampled frequency interval.
Partner Evonik developed two novel methods for deposition and growth of MoS2 based nanoscale transistor arrays for application in flexible electronics. Fabrication techniques are suitable for up-scaling to larger substrates and keep the fabrication cost low by avoiding expensive vacuum processing steps. We have investigated two independent techniques: the first uses exfoliated MoS2 flake dispersions, and the second focuses on bottom-up growth of MoS2 films through solution phase of molybdenum precursors.
Top-down approach by using liquid-phase exfoliated MoS2 flakes obtained from Trinity College Dublin (TCD) were further optimized at Evonik facilities for various coating/printing techniques (such as spin-coating, dip-coating, spray-coating, drop-casting, ink-jet printing and slot-die coating). Solvent systems and additional additives were carefully chosen to be suitable for deposition method, and at the same time to avoid aggregation of exfoliated flakes and formation of crystallites. In the next step, modified formulations with MoS2 flakes were successfully deposited on flexible substrates (i.e. PI and PEN foils) as an active material for fabrication of thin film transistors. Pre-treatment of substrate surface was necessary to control the drying behaviour of the printed ink. Electrical characterization of fabricated devices with several monolayer thick MoS2 film showed only minor conductivity directly after layer deposition. By depositing thicker films, large improvement in electrical characteristics was observed, but without visible field effect in those devices.
Alternatively, bottom-up approach using solution type Mo-precursor annealed in presence of sulfur was performed to form MoS2 thin films. Uniform and homogenous films with thickness down to 4 nm were obtained by combination of spin-coating and high temperature annealing. The fabrication process is wafer scalable with precursor concentration defining the thickness of the MoS2 layers, thereby making it suitable for different applications. The obtained MoS2 films were characterized by microscopy (TEM and SEM), Raman and X-ray photoelectron spectroscopy. High crystallinity and film homogeneity over large area with an average grain size of 100 nm were obtained. Thin film transistors fabricated based on solution processed MoS2 films demonstrated p-type semiconducting behaviour. Bottom-gate top-contact transistors exhibited the field effect mobility of 0.03 cm2/Vs. Due to excess sulfur and polycrystalline nature of the grown MoS2 film, electrical performance of fabricated devices was strongly dominated by high grain-to-grain resistance. Further improvements in growth of MoS2 films may increase the field effect mobility of thin film transistors. Together with low-cost and scalable solution-based fabrication processes, this novel deposition method shall promote the application of dichalcogenides in future nanoelectronic devices.
Partner Aixtron studied the direct formation of the 2D TMDC MoS2 via MOVPE. The deposition is realized in an AIXTRON horizontal hot-wall reactor, initially configured for growing nitride semi¬conductors. Due to the 10 × 2" set-up of the reactor, the simultaneous growth on different substrates such as sapphire, p-Si as well as AlN and GaN buffer structures on sapphire is possible. Molybdenum hexacarbonyl (MCO) and di tert-butyl sulphide (DTBS) are used as metal-organic precursors, for molybdenum and sulphur respectively. The developed process leads to a uniform; wafer-scale deposition of MoS2 on various substrate types. The deposited films mainly consists of MoS2 bilayers and exhibit a very high initial nucleation density. With optimization of growth parameters, a crystal growth process closer to thermodynamical equilibrium can be achieved. A set of experiments are conducted to investigate the nucleation of the films and to further tune nucleation density and lateral growth rate. The target is to deposit coherent wafer-scale monolayer MoS2 films for future electronic and optoelectronic applications.