Semiconductor device and method for manufacturing the semiconductor device

Abstract

A method for manufacturing a semiconductor device, wherein the semiconductor device comprises a two-dimensional semiconducting material disposed on a dielectric layer. The method comprises cleaning an exposed surface of the two-dimensional semiconducting material by using supercritical carbon dioxide.

Description

DescriptionTitle: Semiconductor Device and Method for Manufacturing the Semiconductor DeviceCross-Reference to Related Applications

[0001] This application claims priority to German Patent Application No.10 2023 107 230.7, filed on 22 March 2023Field of the Invention

[0002] The invention comprises a semiconductor device having a two-dimensional layer, and a method for manufacturing the semiconductor device.Background of the Invention

[0003] Stable and a long-term functioning of semiconductor devices made from a two- dimensional semiconducting material (such as, but not limited to, graphene) is one of the main challenges towards the technological applications of the semiconductor devices. Transport properties of graphene field-effect transistors (GFETs) are influenced by exposure to ambient conditions.

[0004] It is known that properties of substrates and free-standing sheets of the two-dimensional semiconductor materials are influenced due to ambient adsorbates from the ambient conditions. For example, it is known that the transport properties of the GFETs are diminished due to adsorption of ambient adsorbates, such as water, on an exposed surface of the semiconducting material of the GFETs. This is discussed extensively in Melios et al “Water on graphene: review of recent progress”, 2018, 2D Mater. 5 022001. The adsorption of the ambient adsorbates results in an undesirable and an uncontrolled p-type or n-type doping. The undesirable and the uncontrolled doping results in a device-to-device variation of the transport properties and prevents reliable application of the GFETs in, e.g., GFET -based sensors.

[0005] Another possible source diminishing the performance of the GFETs is an adsorption of polymeric residues on the exposed surface of the semiconducting material. The adsorption of the polymeric residues can arise from a transfer of the semiconducting material or from a microfabrication of the semiconductor devices.

[0006] One of commonly used polymers for the transfer or the microfabrication is poly (methyl methacrylate) (PMMA). The PMMA is used as a support layer during the transfer of the semiconducting material and is also used as an e-beam resist in lithography processes.

[0007] Some of the polymeric residues inevitably prevail on the exposed surface of the semiconducting material, while cleaning by using acetone is typically used to dissolve the majority of the PMMA. It is these polymeric residues serve as a dominant source of doping and scattering of charge carriers. The polymeric residues thus degrade electrical properties of the semiconductor devices comprising the semiconducting material.

[0008] Conventional techniques for cleaning the exposed surface of the semiconducting materials, such as plasma cleaning, cannot be employed to clean the semiconducting materials. These conventional techniques would etch the semiconducting materials or introduce defects and unwanted chemical groups onto the surface of the semiconducting materials.

[0009] Other specific techniques for the cleaning have been developed in recent years, such as thermal annealing in vacuum or in a dihydrogen / argon (FL / Ar) environment, current-induced cleaning, laser cleaning, low-density inductively coupled plasma (ICP), ultraviolet (UV) ozone treatment, wet chemical cleaning, and mechanical cleaning via in-situ electron microscopy or atomic force microscopy. Nevertheless, the specific techniques for cleaning suffer from slow processing or from a need for a complicated equipment, a use of harsh chemicals and in some cases a lack of scalability and reproducibility. Therefore, there is a demand for more straightforward approaches that are compatible with large-scale manufacturing of the devices from the semiconducting materials as well as being environmentally friendly.

[0010] It has been reported that the stability of graphene and other 2D materials on the substrate in solvents is a relevant issue. It is known that the graphene materials may delaminate in the presence of solvents for cleaning. This low stability results in damage of the devices upon removal of the resist materials used for the microfabrication. For example, Weber et al, “Direct Growth of Patterned Graphene”, Small 2016, 12, No. 11, 1440-1445, report the local adhesion of the graphene to the substrate is too weak to withstand the forces occurring during the removal of poly(methyl methacrylate) (PMMA) by acetone.

[0011] Two papers discuss graphene delamination and re-lamination after the transfer from the growth substrate to the target substrate and the problem of adhesion to the substrate. These are Langston et al, “Graphene Transfer: A Physical Perspective”, Nanomaterials2021,11,2837. https: / / doi.org / 10.3390 / nanol 1112837, and Kim et al. “Understanding Solvent-Induced Delamination and Intense Water Adsorption in Janus Transition Metal Dichalcogenides for Enhanced Device Performance”, Adv. Funct. Mater. 2024, 34, 2308709. Neither of these papers discuss the use of critical point drying (CPD) to clean the surfaces of the graphene.

[0012] Supercritical fluids (SCF) exhibit a unique combination of liquid-like density and gas-like transport properties. The fluid surface tension of the SCF is equal to zero above the critical point of the SCF. The dissolving capability and the pore penetration capability of the SCF increases due to the low viscosity and the high diffusion coefficients of the SCF. The physico-chemical characteristics of the SCF make the SCF a suitable alternative for organic solvents in a wide range of industrial and laboratory applications. Moreover, the SCFs are commercially employed in extraction and purification technologies, for example in refining of cooking oils or decaffeination of coffee. Carbon dioxide (CO2) is the most commonly used SCF due to the easily achievable critical point state, non-toxicity, non-flammability, and recyclability of the CO2. Supercritical CO2 drying is widely used to dry delicate biological specimens while preserving the original morphology of the biological specimens.

[0013] It also known that the CPD treatment further alleviate stiction problems in microelectromechanical (MEMS) devices.

[0014] US Patent application US 2021 / 0242314 Al discloses a method of manufacturing a graphene-based device. The graphene-based device in US 2021 / 0242314 Al comprises a plurality of layers of graphene. Layers of graphene are disposed on a substrate. The substrate has a plurality of cavities and each layer of graphene traverses at least one of the plurality of the cavities in the substrate. In other words, the layers of graphene are suspended above the cavities. The method in US 2021 / 0242314 Al further comprises depositing a photoresist layer on a surface of the plurality of layers of graphene and subsequently immersing the graphene-based device in a first solvent which is then replaced with a second solvent. The second solvent is a low surface tension solvent, for example isopropyl alcohol. The low surface tension solvent is replaced by liquid CO2 which is then converted into supercritical CO2. The photoresist layer is removed from the surface of the plurality of layers of graphene by using the supercritical CO2. The use of the supercritical CO2 for removing the photoresist layer in this patent application enables the removal of the photoresist without damaging the layers of graphene.

[0015] The international patent application WO 2018 / 017369 discloses a method for manufacturing a three-dimensional (3D) graphene material. The three-dimensional (3D) graphene material in WO 2018 / 017369 comprises n-doped graphene foams, carbon nano- tube / graphene composites, carbon fibres, carbon nanotube fibres, carbon nanoonions / gra- phene composites, phosphorene foams, 3D MX and / or MX2 foams, and combinations thereof. The method in WO 2018 / 017369 comprises mixing a metal powder and a carbon source to form a metal and carbon source mixture. The method comprises converting the carbon source into graphene sheets that cover the metal powder. The method comprises removing the metal source to form the 3D graphene material. The method further comprises drying the 3D graphene material by a critical point dryer using liquid / supercritical CO2. The supercritical CO2 is used to dry the graphene material with minimum changes in volume without damaging structural integrity of the 3D graphene material.

[0016] European Patent Application No. EP 0 386 895 A2 (IBM) teaches a method for the removal of residue from an etched precision surface, such as a bulk semiconductor sample. The surface is exposed to a supercritical fluid or liquid CO2 under appropriate conditions that are sufficient to remove the residue from the precision surface. The etched precision surface has microfabricated 3D structures with features below the plane of the surface such as cavities, trenches, or channels, or raised features such as mesas.

[0017] None of the prior art documents show the use of a supercritical fluid for cleaning the surface of a two-dimensional semiconductor material to overcome the issues of delamination of the two-dimensional semiconductor material from the substrate as well as cleaning the impurities from the surface of the two-dimensional material.Summary of the Invention

[0018] A method for manufacturing a semiconductor device is taught in this disclosure. The semiconductor device comprises a two-dimensional semiconducting material disposed on a dielectric layer, the method comprising cleaning an exposed surface of the semiconducting material by using a supercritical fluid.

[0019] The method of cleaning is also termed “critical point dryer” (CPD) treatment. In one aspect, the supercritical fluid is chosen from supercritical CO2, supercritical ethane, supercritical propane, supercritical ethylene, or supercritical ammonia.

[0020] The CPD treatment is used as a robust cleaning and drying post-processing method to enhance electric properties of the semiconductor device. The CPD treatment enables enhancement of a field-effect mobility and a recovery of the undesirable doping of the semiconductor device. A functioning of the semiconductor device is diminished by an exposure to ambient conditions. The functioning of the semiconductor device can be recovered via the CPD treatment. The CPD treatment of semiconductor devices comprising the semiconducting material is an effective step for achieving better transport characteristics. The CPD treatment can be readily integrated into a microfabrication of the semiconductor device in cleanrooms as well as into a microfabrication of the semiconductor device on a large scale.

[0021] It has also been found that the CPD treatment of the semiconductor devices does not results in the delamination of the 2D materials, such as graphene, from the substrate, as has been reported in the past.

[0022] In a further aspect, the method comprises immersing the semiconductor device in a solvent, replacing the solvent by the supercritical carbon dioxide, and removing one of polymeric residues, ambient adsorbates, or a combination thereof.

[0023] In one aspect, the polymeric residues are chosen from at least one of poly (methyl methacrylate) (PMMA), SU-8 polymer, copolymers comprising methyl methacrylate and methacrylic acid, styrene acrylates, diazonaphthoquinone, off-stoichiometry thiol-ene (OSTE) polymer, hydrogen silsesquioxane (HSQ) or a combination thereof.

[0024] In a further aspect, the ambient adsorbates are water, alkanes, or a combination thereof.

[0025] A semiconductor device is also taught in this disclosure. The semiconductor device comprises a two-dimensional semiconducting material disposed on a dielectric layer, wherein an exposed surface of the two-dimensional semiconducting material is a cleaned surface with supercritical carbon dioxide.

[0026] In one aspect, the semiconducting material is a semiconducting two-dimensional material chosen from at least one of a chalcogenide, phosphorene, graphene, or a combination thereof.

[0027] In one aspect, the chalcogenide is chosen from at least one of InSe, GaSe, GaS, SnS2, SnSe2, or a transition metal di chalcogenide of M0S2, WS2, MoSe2, WSe2, MoTe2, Wte2, PdSe2, PtSe2, PdTe2, PtTe2, PtS2, PdS2 or a combination thereof.

[0028] The two-dimensional semiconducting material is a substantially monolayer material. The two-dimensional semiconducting material may also form bilayers or multilayers.

[0029] This document also teaches use of the semiconductor device as a field effect transistor, a photodetector, a diode, a logic element, a photovoltaic device, an electroluminescent light emitting device, a solution gated field effect transistor, a two-dimensional materialbased sensor, or a saturable absorber.Description of the figures

[0030] Figs. 1 and 2 show a view of a semiconductor device comprising a semiconducting material disposed on a dielectric layer as well as cleaning of an exposed surface of the semiconducting material with supercritical carbon dioxide.

[0031] Fig. 3 shows a flow chart of the method for manufacturing the semiconductor device.

[0032] Figs. 4A to 8 show an effect of the cleaning, i.e., critical point drying (CPD) treatment, with the supercritical carbon dioxide on charge transfer characteristics of the semiconductor device.

[0033] Fig. 4A shows an optical microscope image of a chip with an array of fifteen semiconductor devices. Fig. 4B shows an optical microscope image of an enlarged view of one semiconductor device from Fig. 4A.

[0034] Fig. 5 shows a schematic three-dimensional view of the semiconductor device and a measurement configuration.

[0035] Fig. 6A shows a source-drain current as a function of a back-gate voltage, i.e., Ids- Vgforward sweep, of an exemplary semiconductor device before and after the CPD treatment. Model fitting results are shown with grey-dashed curves.

[0036] Fig. 6B shows an effect of the CPD treatment on a hysteresis of the semiconductor device 10 of GFETs.

[0037] Fig. 7 shows a Dirac point on the left Y-axis and a hysteresis on the right Y-axis of the semiconductor device before and after the CPD treatment.

[0038] Fig. 8 shows box plots and whisker plots of a carrier mobility for the ten semiconductor devices before and after the CPD treatment.

[0039] Figs. 9A to 9D show a surface cleaning of the dielectric layer upon the CPD treatment. Figs. 9A and 9B show a height atomic force microscopy (AFM) topography image ofa surface 10^ 10 m2of the semiconducting material transferred by chemical vapor deposition (CVD) on the dielectric layer. The AFM topography image of Fig. 9A is taken before the CPD treatment and the AFM topography image of Fig. 9B after the CPD treatment. A high-resolution C is X-ray photoelectron spectroscopy (XPS) spectrum of an exposed surface of the semiconducting material obtained before the CPD treatment is shown on Fig. 9C and after the CPD treatment on Fig. 9D. Fitting peaks are assigned to sp2C-C, sp3C-C, C- O, C=O and O-C=O.

[0040] Fig. 9E shows height histograms of the polymeric residues of PMMA calculated from Fig. 9A and 9B. The inset of Fig. 9E shows diameter histograms of the polymeric residues. The light grey and the dark grey colours represent the histograms before and after the CPD treatment, respectively.

[0041] Figs 10A to 10C show Raman spectroscopy data of the semiconductor device before and after the CPD treatment. Fig. 10A shows exemplary Raman spectra taken of the semiconductor device before and after the CPD treatment. The black-dashed lines indicate the position of the G band and of the 2D band of the semiconductor device. The G band is at 1589 cm'1and the 2D band is at 2682 cm'1for the dark grey spectrum, i.e., the semiconductor device before the CPD treatment. Results of fitting parameters are shown in Fig. 10B for the position and in Fig. 10C for the full width at half maximum of the 2D band as a function of the G band position before (circles) and after (hexagons) the CPD treatment.

[0042] Fig. 11 A shows a change in charge transfer curves of the semiconductor device after the CPD treatment and after ambient condition exposure for 1 day, 3 days and 14 days. Grey-dashed curves show model fitting results. Figs 11B and 11C show the mobility and Dirac point changes with the prolonged exposure time and after repeated CPD treatment (CPD repeat).

[0043] Fig. 12A shows a schematic illustration of a semiconductor device of GFET with a channel with doping areas Ai and A2. Fig. 12B shows a zoom-in view of the transfer curve of the electronic device 10 shown in Fig. 11 A, exposure for 14 days in ambient condition with the model results assuming heterogeneous doping of the channel of the semiconducting material 20.

[0044] Figs 13 A shows exemplary Raman spectra taken of the semiconductor device after staying for fourteen days in ambient condition. Fig. 13B shows results of the fitting param-eters position and Fig. 13C shows the full width at half maximum of the 2D band as a function of the G band position for the semiconductor device after staying for fourteen days in ambient condition.Detailed description of the invention

[0045] The invention will now be described on the basis of the figures. It will be understood that the embodiments and aspects of the invention described herein are only examples and do not limit the protective scope of the claims in any way. The invention is defined by the claims and their equivalents. It will be understood that features of one aspect or embodiment of the invention can be combined with a feature of a different aspect or aspects and / or embodiments of the invention.

[0046] Figs. 1 and 2 show a view of a semiconductor device 10 comprising a semiconducting material 20 disposed on a dielectric layer 31, for example silicon dioxide (SiCh). The dielectric layer 31 is located on top of a substrate 32. The substrate 32 is, for example, made of doped silicon (Si) and forms a gate electrode. A source S and a drain D are connected to the semiconducting material 20. The semiconducting material 20 is, for example, a semiconducting two-dimensional material, i.e., a substantially monolayer material. In one aspect, the semiconducting two-dimensional material is chosen from at least one of a chalcogenide, phosphorene, graphene, or a combination thereof. In a further example, the chalcogenide is chosen from at least one of InSe, GaSe, GaS, SnS2, SnSe2, or a transition metal dichalcogenide of M0S2, WS2, MoSe2, Wse2, MoTe2, Wte2, PdSe2, PtSe2, PdTe2, PtTe2, PtS2, PdS2or a combination thereof.

[0047] In one example, the semiconducting material 20 is a substantially monolayer (i.e. two-dimensional) material, such as graphene. The semiconductor device 10 depicted in Figs. 1 and 2 is a graphene field-effect transistor (GFET), but this is not limiting of the invention.

[0048] Figs. 1 and 2 further show a cleaning S300 of an exposed surface 11 of the semiconducting material 20 with a supercritical fluid 12 of supercritical carbon dioxide. Fig. 2 shows that the cleaning S300 results in removing S340 polymeric residues 40, such as poly (methyl methacrylate) (PMMA). Fig. 2 further shows that the cleaning S300 results in removing S340 ambient adsorbates 50, such as water (H2O) or alkanes, for example ethane (C2H6).

[0049] In one example, the semiconductor devices 10 are prepared by growing the semiconducting material 20 by chemical vapour deposition (CVD) and transferred via a PMMA assisted wet transfer onto a p-doped silicon wafer with 300 nm thickness of thermally grown SiCh on top of the silicon wafer. The surface of the p-doped silicon wafer has to be cleaned to remove airborne contaminations and flat to enhance the van der Waal’s interaction between the transferred semiconducting material 20 and the substrate of the p-doped silicon wafer. The cleaning of the surface is carried out using, for example, organic solvents, water, or oxygen or argon plasma. The surface of the substrate has, for example, a mean root square roughness of less than 1 nm.

[0050] In a further aspect, the surface can be functionalised to enhance the van der Waal’s interaction. This can be done by oxygen plasma or short organic molecules to create dipoles in the substrate and thus strengthen the van der Waals interaction with the two-dimensional (2D) semiconductor material 20, such as graphene or monolayers of transition metal dichalcogenides (e.g., M0S2, WS2, MoSe2, or WSe2)

[0051] Fig. 3 shows a flow chart of the method for manufacturing the semiconductor device 10. The method comprises the cleaning of the step S300 of the exposed surface 11 of the semiconducting material 20. The cleaning S300 comprises the use of the supercritical fluid 12.

[0052] In one example, the method further comprises immersing in step S310 the semiconductor device 10 in a solvent 15. The solvent is, for example, acetone. The method further comprises replacing in step S320 the solvent 15 by the supercritical fluid 12.

[0053] In a further example, the method comprises in step S340 removing one of polymeric residues 40, ambient adsorbates 50, or a combination thereof. The polymeric residues 40 are, for example, poly (methyl methacrylate) (PMMA), copolymers comprising methyl methacrylate and methacrylic acid, styrene acrylates or diazonaphthoquinone. The ambient adsorbates are, for example, are a ketone or an ester.

[0054] In other aspect, it is possible to make multilayers of two-dimensional semiconductor materials. This can be done using different layer of two-dimensional semiconductor materials which were grown by any method or exfoliated from bulk material. The monolayers can also be stacked on each other to make bilayers or multilayers, making the so-called van der Waals heterostructures from the same two-dimensional semiconductor material or from monolayers or few layers of different two-dimensional materials to achieve some specificelectronic, optoelectronic, or photonic property. The transfer of the monolayers is carried out using the transfer method described above and the cleaning of the substrate is also set out as above.

[0055] Fig. 4A shows an optical microscope image of a chip area containing fifteen microfabricated two-terminal semiconductor devices. The semiconductor devices 10 have an active area of 190x 100 pm2each, as can be also seen in Fig. 4B. Fig. 4B is a high-resolution optical image of an individual semiconductor device 10.

[0056] A schematic view of the semiconductor device 10 with a measurement configuration is shown in Fig. 5. The semiconductor device 10 has contacts 35 which are located on opposite sides on the semiconducting two-dimensional material 30 and form the source S and the drain D.

[0057] A source-drain current Ids is measured as a function of the back-gate voltage Vgby applying a bias Vds of 10 mV between the source and drain. The source-drain current Ids is measured to collect a field-effect response of each semiconductor device 10 with a channel aspect ratio LAV of 1.9.

[0058] Fig. 6A shows exemplary Ids-Vgtransfer curves of the semiconductor device 10 before and after the CPD treatment. The semiconductor device 10, in the example a GFET device, initially shows a high p-type doping state with a Dirac point at 42 V. The Dirac point shifts to near 0 V after the CPD treatment. Furthermore, the transfer curve becomes steeper, as shown in Fig. 6A. This steeper transfer curve implies an enhancement of the charge carrier mobility of the semiconductor device 10. The electron-hole conduction asymmetry is reduced.

[0059] Fig. 6B shows transfer curves before and after the CPD treatment. A direction of forward sweeps and backward sweeps is indicated with arrows. The hysteresis, i.e., the difference of the Dirac points in the forward gate voltage and the backward gate voltage sweeps, is significantly reduced, as shown in Fig. 6B.

[0060] The difference of the Dirac points is further shown in Fig. 7. Circle / triangle symbols represent the average Dirac point / hysteresis values for ten of the semiconductor devices 10 in Fig. 7. Error bars show a standard deviation for the average Dirac point and the hysteresis values.

[0061] The difference of the Dirac points is, on average, from 33.1 ± 4.9 V to 12.9 ± 3.1 V. The decreased hysteresis in the semiconductor device 10 after the CPD treatment indicates a corresponding decrease in an accumulation of charges in trap centres. The trapped charges are typically caused by water molecules bound on the surface of the dielectric layer 31 made of SiCh. The bound water molecules screen the electric field generated by the back gate.

[0062] Fig. 6A shows a comparison between experimental transport data and results from a phenomenological model. The results from the phenomenological model are shown as dashed curves in Fig. 6A The phenomenological model defines an effective density n of free charge carriers in the semiconducting material 20 as a function of an applied gate voltage Vg. It is assumed that the effective density n of free charge carriers in the semiconducting material 20 is described by:where VCNPis the gate voltage at the charge neutrality point (CNP), CG=is the gatecapacitance per unit area, s0is the permittivity of free space, sris the relative permittivity of the silicon dioxide, e is the elementary electron charge, and n* is the effective residual charge carrier density at the charge neutrality point where disorder in the semiconducting material 20 will lead to electron and hole puddles.

[0063] The conductivity o of the semiconductor device 10 is a function of the effective charge carrier density in the channel of the semiconducting material 20.

[0064] The conductivity o is then described by: o(n) = l / [(nep,)“1+ o”1] (2) where Pi is an independent charge-carrier long-range scattering mobility and osis the charge carrier independent conductivity due to short-range scattering.

[0065] The following equation:is used to express the total resistance R of the semiconductor device 10. L and W in the equation are the length and the width of the active area of the semiconductor device 10. Rcis the contact resistance in the semiconductor device 10.

[0066] The variablesinthe phenomenological model are the fitting parameters of Eq. (2). VCNPis obtained from the experimental transfer curves of Fig. 6A.

[0067] The transfer characteristics of the semiconductor device 10 are recorded over a large gate voltage range, for example, from -90 V to 120 V. The hole conduction curve is fitted with the results from the phenomenological model as shown in the grey-dashed curve in Fig. 6A. The recording over the large gate voltage and the fitting of the hole conduction curve is done to extract the model fitting parameters from the experimental transport data.

[0068] Fig. 8 shows box plots and whisker plots of carrier mobility for the ten semiconductor devices before and after the CPD treatment. The carrier mobility value for each of the semiconductor devices has a different shape on Fig. 8. The highest horizontal lines in the boxes are the first quartiles, i.e., splits off the lowest 25 % of the values from the highest 75 % of the values. The lowest horizontal lines in the boxes are the third quartiles, i.e., splits off the highest 25 % of the values from the lowest 75 % of the values. The band inside the box is the median and the vertical whiskers show the maximum and the minimum carrier mobility values.

[0069] The CPD treatment leads to an average increase in the value of the charge-carrier long-range scattering mobility Pi from (2032 ± 198) cm2 / Vs to (2695 ± 178) cm2 / Vs, as shown in Fig. 8. The increase of in the value of the charge-carrier long-range scattering mobility Pi is consistent with the decrease in the value of the gate voltage at the charge neutrality point VCNP. The value of VCNPdecreases indeed from (42 ± 2.7) V to (1.9 ± 1.4) V. The increase of Pi and the decrease of VCNPcan be attributed to the removal of the charge impurities. Moreover, the effective residual charge carrier density n* decreases in average from a value of (0.50 ± 0.02) x io12cm'2to a value of (0.39 ± 0.03) x io12cm'2. The decrease of the effective residual charge carrier density n* is shown in Table 1. An average change of the model fitting parameters VCNP, Pi and 01+ Rc) upon the CPD treatment for the tensemiconductor devices 10 are further shown in Table 1.Table 1 :

[0070] The decrease of the effective residual charge carrier density n* is consistent with an assumed reduction in the surface density of impurities of the semiconductor devices 10 after the CPD treatment.

[0071] The overall results of the quantitative analysis of Table 1 show that the reduction of charge scattering centres results in an enhancement of the charge transfer characteristics.

[0072] It is suggested that the cleaning S300 of the exposed surface 11 of graphene of the semiconducting material 20 via an effective removal of the polymer residues 40 and other adsorbates during the CPD treatment causes the enhancement of the charge transfer characteristics.

[0073] In a further example, the semiconductor devices 10 are prepared by growing the semiconducting material 20 of graphene by CVD and transferred on the dielectric layer 31 / the substrate 32 of SiC>2 / Si by the same process as for the microfabricated devices.

[0074] The surface morphology and the chemical composition of the semiconductor devices 10 are characterized by AFM and XPS. AFM height topographical images of an area of 10x 10 pm2of the semiconductor device 10 are shown in Fig. 9A before the CPD treatment and in Fig. 9B after the CPD treatment.

[0075] Figs. 9A and 9B show a typical topography of a monolayer of the semiconducting material 20 of graphene grown by CVD and transferred on the dielectric layer 31 / the substrate 32 of SiC>2 / Si wafer. The topography is overlayed with some nanometre-sized dots resulting from the presence of impurities, i.e., polymeric residues 40 of PMMA due to the transfer. As can be seen on Figs. 9A and 9B, the impurities are significantly reduced after the CPD treatment.

[0076] A quantitative image analysis reveals that an impurity coverage is reduced from approximately 13 % to approximately 2 %.

[0077] Fig. 9E shows histogram distributions of height and diameter of the polymeric residues 40. The histogram distributions show a significant reduction of the vertical size of the distributions after the CPD treatment.

[0078] The observed effect of the cleaning S300 of the exposed surface 11 of the semiconducting material 20 can be attributed to an interaction of the supercritical fluid 12 of supercritical carbon dioxide with polymeric residues 40 made of amorphous polymers, such as PMMA. The observed effect can also be attributed to the high diffusivity of the supercriticalcarbon dioxide in the polymeric residues 40 compared to diffusivity of organic solvents. The supercritical CO2 has also been shown to act as an effective plasticizer for glassy polymers by reducing the glass transition temperature (Tg) of the glassy polymer to near room temperature.

[0079] The solubility of the PMMA in the supercritical CO2 is found to be higher compared to the solubility of other crystalline polymers. The solubility of the PMMA in the supercritical CO2 12 is influenced by the thickness and the morphology of the polymer film. The polymeric residues 40 of the nanometre-sized PMMA are dissolved in the supercritical CChand swollen during the CPD treatment. The supercritical CO2 treatment weakens a van der Waals affinity of the polymeric residues 40 to the semiconducting material 20 of graphene. The supercritical CO2 treatment causes the polymeric residues 40 to detach from the exposed surface 11 of the semiconducting material 20.

[0080] Figs. 9C and 9D show high resolution C is XP spectra of a sample of the semiconductor device 10 comprising the semiconducting material 20 of graphene before (Fig. 9C) and after (Fig. 9D) the CPD treatment. The sample is measured at the same lateral position to ensure comparable analysis of the spectra. A spectral deconvolution shows that the peak intensity due to sp2carbon - carbon bonds (sp2C-C) after the CPD treatment increases from 62 ± 2 % to 71 ± 2 %. The peak intensity of the components assigned to sp3-carbon bonds (C-C, C-O, C=O and O-C=O) decreases. In particular, the peak intensity related to the methoxy and to the carboxyl groups of the polymeric residues 40 of PMMA, having binding energies of approximately 286.9 eV and approximately 289.1 eV, reduces from 6 ±1 % to 2 ± 1 % and from 5 ± l % to 3 ± l %.

[0081] Thus, the presence of the polymeric residues 40 on the semiconducting material 20 of graphene after the CPD treatment is reduced, as shown by the XPS analysis of Figs. 9C and 9D, in agreement with the AFM results of Fig. 9A and 9B.

[0082] The reduction of the presence of the polymeric residues 40 on the semiconducting material 20 of graphene after the CPD treatment can also be observed by helium ion microscopy on the dielectric layer 31 of free-standing graphene sheets.

[0083] A series of Raman spectroscopy measurements under the same experimental conditions used for the electrical measurements are conducted to investigate structural and doping characteristics of the semiconductor device 10 of GFET.

[0084] Fig. 10A to IOC show Raman spectra obtained before and after the CPD treatment. Notably, the absence of the D peak near 1350 cm’1in the Raman spectra obtained before and after the CPD treatment suggests that the CPD treatment does not introduce any measurable number of defects in the semiconductor device 10.

[0085] The observed variation in the spectroscopy features in the Raman spectra can be attributed to the polycrystalline nature of the semiconducting material 20 of graphene grown by CVD. It has been established that the frequency of the G band in graphene is sensitive to the doping state of graphene.

[0086] The position and the full width at half maximum (FWHM 2D) of the 2D peak in relation to the G peak position are shown on Figs. 10B and 10C. Average red shifts observed in the G peak and in the 2D peak wavelengths after the CPD treatment, from 1588.7 cm’1to 1586.0 cm’1and from 2682.8 cm’1to 2680.8 cm’1, respectively, are consistent with a lower doping level. The average red shifts can further be directly correlated to the Dirac point shift near zero gate voltage shown in Fig. 7.

[0087] An average decrease in the FWHM 2D from 43.5 to 39.7 implies a reduction in a strain inhomogeneity. The reduction in the strain inhomogeneity may have been caused by the polymeric residues 40 of PMMA that locally exert a mild tensile strain on the transferred semiconducting material 20 of graphene grown by CVD.

[0088] The Raman analyses of Fig. 10A provide further support for the observed enhancement of the charge transfer characteristics in the transport measurements of the semiconductor device 10.

[0089] An influence of ambient conditions on the transport characteristics of the semiconductor devices 10 of GFET after the treatment by CPD was investigated and the stability of the semiconductor devices 10 after the treatment by CPD was determined, as will be further explained.

[0090] Fig. 11 A shows that an increase of an ambient exposure time results in a shifting of a position of the Dirac point towards higher voltages. The term “ambient” means at room temperature and pressure together with usual air moisture. This phenomenon can be attributed to a p-type doping effect due to the adsorbed water as well as airborne hydrocarbon contamination.

[0091] An asymmetry in the electron conductivity and in the hole conductivity of the semiconductor device 10 becomes more pronounced with time, while the Dirac point shifts tohigher gate voltages. As can be seen, a new shoulder at the minimum conductivity begins further to emerge and becomes increasingly pronounced with prolonged ambient exposure time. This reflects the presence of heterogeneous doping in the semiconductor device 10 of GFET.

[0092] The phenomenological model considers a homogeneous doping in the semiconductor device 10 of GFETs and is presented below to describe the influence of the ambient conditions.

[0093] The model considers two doping areas Ai and A2 within the channel of the semiconducting material 20 of graphene with different doping levels.

[0094] This model enables to describe experimental data very well, see Figs. 11 A, 11B, 12A and 12B.

[0095] Fig. 12B shows the total fitting curve (grey) as a sum of two fitting peaks corre- spending to the level of the doping areas Ai and A2 shown with dashed curves.

[0096] The two doping areas A and d2correspond to conductivities given bywhere VCNPand VCNP 2are the charge neutrality point voltages of the doping areas A and / 12andwhere R is the resistance of the semiconductor device 10 and A = A1+ / 12. The schematic of heterogenous doping with two areas is shown in Fig. 12A.

[0097] It is possible to estimate a level of a growing heterogeneity by incorporating aratio as a fitting parameter by assuming a completely homogeneous doping, i.e. Ai equal to 100 % of the semiconductor device 10 treated by CPD.

[0098] Table 2 shows parameters used for calculating the transfer characteristics shown in Fig. 11 A.Table 2:

[0099] The difference in VCNPbetween the doping regions A and J42was found to be smaller than 15 V as shown in Table 2.

[0100] The resistance caused by a transition of charge carriers from the doping region A to the doping region J42can be disregarded, as the resistance is generally much less significant than the resistance resulting from charged impurity scattering in the semiconducting material 20 of graphene.

[0101] It appears that the model fits well with the experimental data based on the fitting results shown in Fig. 12B.

[0102] It is suggested that the semiconductor device 10 of graphene field-effect transistors may have primarily two distinct levels of doping after undergoing ambient condition exposure. The fitting parameters used for a representative device are shown the table 2.

[0103] As shown in Table 2, the doping area A2 was found in average to be 14 % after 1 day, 18 % after 3 days, and 23 % after 14 days of the ambient exposure.

[0104] The change of VCNPcan be described with an equation:which is in good agreement with an expected change of VCNPdue to the ambient adsorbates 50. The slight increase of n* after prolonged exposure to air, see table 2, can be similarly explained by increased doping from the ambient adsorbates 50. The increased doping provides charged impurities that increase the effective charge carrier density at the charge neutrality point.

[0105] Transfer curves were obtained on the semiconductor devices 10 that were exposed to the ambient conditions for 14 days and after repeating the CPD treatment. The transfercurves become again highly homogeneous in terms of doping and indistinguishable from the transfer curves obtained after the initial CPD treatment.

[0106] The mobility and the position of the Dirac point show a recovery of the electronic performance of the semiconductor devices 10, as shown by Fig. 1 IB and 11C.

[0107] The squares and circles in Figs, 11B and 11C correspond to the average values of the mobilities and the Dirac points for the ten semiconductor devices. The, the error bars indicate the standard deviation of the data. Lines solely serve as guides to the eye. Some error bars in Fig. 11C are of the same size as the dots representing the position of the Dirac point.

[0108] The recovery of the semiconductor devices 10 suggests that the CPD treatment can effectively remove the ambient adsorbates 50 impeding the unwanted doping of the semiconductor devices 10.

[0109] The superior permeability and the low viscosity of the supercritical CChcombined with a high driving force under high-pressure conditions enables strong exchange of trapped water molecules with the supercritical CO2.

[0110] Electrical characteristics of the semiconductor devices 10 of GFET are improved when the semiconductor devices 10 are measured immediately after the CPD treatment. A minimal unwanted p-type doping and an enhanced mobility due to a removal of the trapped ambient adsorbates 50 of water and the polymeric residues 40 are also measured on the semiconductor devices 10 after the CPD treatment.

[0111] However, the ambient adsorbates 50 of the water-borne contamination and the airborne contaminations lead to the p-type doping and to a decrease of the mobility over time at ambient conditions.

[0112] The charge transfer characteristics are recovered by repeating the CPD treatment. Fig. 13 A shows black-dashed lines that indicate respectively the position of the G band and the 2D band at 1589 cm'1and 2682 cm'1. Figs. 13A to 13C shows that the G band and the 2D bands of the semiconductor device 10 after CPD treatment shift to higher frequencies by exposing the semiconductor device 10 for 14 days to the ambient conditions.

[0113] This shift to higher frequencies is consistent with the increase of p-type doping observed in the charge transport measurements, as shown in Figs. 11B and 11C. The related 2D and G shifts imply that the hole doping is the main cause of the observed blueshifts rather than the strain variation. The FWHM of the 2D peak does not increase dramatically afterfourteen days exposure to ambient condition. This shift can result from the increase in the ambient adsorbates 50 of water and simultaneous negligible change in the polymeric residues 40 contamination.Experimental Methods

[0114] The semiconducting material 20 was a monolayer of graphene. The monolayer of graphene was grown on copper (Cu) foils by low-pressure chemical vapor deposition (CVD). The monolayer of graphene was then transferred on a dielectric layer 31 / substrate 32of SiC>2 / Si using PMMA-assisted wet transfer via electrochemical delamination as described.

[0115] The semiconducting material 20 of the monolayer of graphene sheets were patterned on the dielectric layer 31 / substrate 32 of SiC>2 / Si wafers using standard electron beam lithography (EBL, Vistec EBPG 5000+) with the PMMA resist (AR-P 671.04, Allresist). The pattern transfer was performed by dry etching in an argon / oxygen plasma. 80 / 20 nm Gold (Au) contacts were thermally evaporated in a two-step lithographic process while employing 2 nm of Titanium (Ti) as an adhesion promoter to directly contact the graphene channel and hence lower the contact resistance of the semiconductor device 10.

[0116] Transfer characteristics of the semiconductor device 10 of GFET were measured in a vacuum probe station of Lakeshore TTPX using a two-channel source measure unit (SMU) of Keithley 2634B. One SMU was used to apply the gate voltage with respect to the drain and another SMU was used to apply the drain voltage. The source terminal was grounded. The source-drain current, Ids, as a function of applied gate voltage, Vg, was measured.

[0117] The SMUs were operated via ethemet using Lab VIEW with the software MODU- LAB, from Franz Ahlers, PTB Braunschweig. Vacuum was generated with a turbomolecular pumping station Edwards nEXT 240.

[0118] The semiconductor devices 10 were kept for about 1 hour in the probe station to achieve the vacuum of <5* 10'6mbar as measured at the turbomolecular pumping station.

[0119] The CPD treatment was performed using an autosamdri-815 (Tousimis) automatic critical point dryer. Firstly, the semiconductor devices 10 of GFETs were kept in the solvent 15 of acetone inside a process chamber for 30 min. The solvent 15 of acetone was then replaced by liquid CO2 while sustaining the temperature at around -2 °C and a pressure at 5 MPa. Then the temperature was ramped up to 38 °C and the pressure was increased up to 10 MPa to achieve supercritical CO2 conditions. The temperature was ramped down to roomtemperature, i.e., approximately 25 °C, and the pressure went down to the atmospheric pressure after maintaining the semiconductor devices 10 in the supercritical CO2 state for 10 min. Gaseous CO2 was then vented.

[0120] Procedures of AFM, Raman and XPS measurements.

[0121] AFM measurements were performed with a NTEGRA system (NT-MDT) in semicontact mode at ambient conditions using n-doped silicon cantilevers (NSG01, NTMDT) with a typical tip radius of 6 nm and a resonance frequency of 87-230 kHz.

[0122] The obtained AFM images were binarized with an intensity threshold to quanti- tively determine the area covered by the polymeric residues 40 of PMMA.

[0123] Areas of individual particles of the polymeric residues 40 were determined using the particle detection feature of the ImageJ software based on the binary images from the AFM images.

[0124] Only particles of the polymeric residues 40 with a minimum diameter of 2 pixels were counted into the area determination to exclude artifacts introduced by the binarization or from low signal to noise ratio.

[0125] Raman measurements. Raman spectra were acquired using a Bruker Senterra spectrometer operated in the backscattering mode. Room temperature measurements were carried out with a frequency-doubled neodymium-doped yttrium aluminium garnet (Nd: YAG) Laser at 532 nm, a 50x objective and a thermoelectrically cooled Charge Coupled Device (CCD) detector.

[0126] The spectral resolution of the system was 2-3 cm'1. The Si peak at 520.7 cm'1was used for peak shift calibration of the spectrometer. The Raman spectra were collected by spotting 5 points on 8 devices (total statistics of 40 points) for different ambient condition exposure and analysed by fitting a single Lorentzian peak to the G peaks and 2D peaks of each spectrum after a linear background subtraction.

[0127] X-ray photoelectron spectroscopy (XPS) measurements. XPS was performed in an ultra-high vacuum (base pressure 2><1O'10mbar) Multiprobe system (Scienta Omicron) using a monochromatized X-ray source (Al Ka) and an electron analyser (Argus CU) with a spectral energy resolution of 0.6 eV.

[0128] Four additional peaks were used in the curve-fitting analysis besides the asymmetric function of sp2C-C bond. The four additional peaks correspond to the polymeric residues40 of PMMA resulting from the transfer process and resulting from hydrocarbon contaminants on the surface of the semiconducting material 30 of graphene. The XPS spectra were calibrated using the Si 2p peak (SiCh, 103.6 eV) and fitted using Voigt functions (30:70) after Shirley -background subtraction. The sp2C-C component assigned to the semiconducting material 20 of graphene was fitted using a Doniach-Sunjic line shape with an asymmetry factor a = 0.08, as typical for graphene.

[0129] AFM topography images. A quantitative analysis of the polymeric residues 40 of PMMA particles was performed using AFM images. The height histogram was determined by the Image Analysis software, while the lateral dimensions were measured using the Image! software. The images were transformed to binary using an intensity threshold to evaluate the diameters and areas of the particles through the particle detection feature of the software. To prevent the occurrence of artifacts such as graphene wrinkles, the analysis excluded particles with a lateral diameter of 1-2 pixels.

[0130] It is also possible to grow graphene monolayers for device fabrication on nickel substrates (foils) by CVD or by epitaxial growth on silicon carbide crystals by thermal decomposition of either Si-face or C-face of these crystals. Other techniques, like atomic layer deposition (ALD) on various substrates (e.g., sapphire) can be used, although experience has shown that these other techniques lead typically to a much lower structural and therefore electronics properties of the resulting graphene.

[0131] The monolayers and few layers of transition metal dichalcogenides (e.g., M0S2, WS2, MoSe2, WSe2, etc.) can be grown by CVD and MOCVD on a variety of substrates, including but not limited to Si-wafers, glass, metals, sapphire, glassy carbon, etc.Reference numerals10 semiconductor device11 exposed surface 12 supercritical fluid15 solvent20 semiconducting material31 dielectric layer32 substrate 40 polymeric residues50 ambient adsorbates

Claims

Claims1. A method for manufacturing a semiconductor device (10), wherein the semiconductor device (10) comprises a two-dimensional semiconducting material (20) disposed on a dielectric layer (31), the method comprising: cleaning (S300) an exposed surface (11) of the two-dimensional semiconducting material (20) by using a supercritical fluid (12).

2. The method of claim 1, wherein the supercritical fluid (12) is chosen from supercritical carbon dioxide, supercritical ethane, supercritical propane, supercritical ethylene, supercritical ammonia.

3. The method of claim 1 or 2, wherein the two-dimensional semiconducting material (20) is chosen from at least one of a chalcogenide, phosphorene, graphene, or a combination thereof.

4. The method of any of the above claims, wherein the chalcogenide is chosen from at least one of InSe, GaSe, GaS, SnS2, SnSe2, or a transition metal di chalcogenide of M0S2, WS2, MoSe2, WSe2, MoTe2, WTe2, PdSe2, PtSe2, PdTe2, PtTe2, PtS2, PdS2or a combination thereof.

5. The method of any of the above claims, wherein the two-dimensional semiconducting material (20) is a multilayer material, a bilayer material, or a monolayer material.

6. The method of any of the above claims, wherein the step of cleaning (S300) the exposed surface (11) of the semiconducting material (20) comprises: immersing (S310) the semiconductor device (10) in a solvent (15); and replacing (S320) the solvent (15) by the supercritical fluid (12).

7. The method of any of the above claims, wherein the step of cleaning (S300) the exposed surface (11) of the semi-conducting material (20) further comprises:removing (S340) one of polymeric residues (40), ambient adsorbates (50), or a combination thereof.

8. The method of any of the above claims, wherein the polymeric residues (40) are chosen from at least one of poly (methyl methacrylate) (PMMA), SU-8 polymer, copolymers comprising methyl methacrylate and methacrylic acid, styrene acrylates, diazonaphthoquinone, off-stoichiometry thiol-ene (OSTE) polymer, hydrogen silsesquioxane (HSQ) or a combination thereof.

9. The method of any of the above claims, wherein the solvent (15) is a ketone or an ester.

10. The method of any of the above claims, wherein the solvent (15) is one of acetone, isopropanol, dimethylformamide or ethanol.

11. The method of any of the above claims, wherein the ambient adsorbates (50) are water, alkanes, or a combination thereof.

12. A semiconductor device (10) comprising: a two-dimensional semiconducting material (20) disposed on a dielectric layer (31), wherein an exposed surface (11) of the two-dimensional semiconducting material (20) is a cleaned surface with supercritical carbon dioxide (12).

13. Use of the electronic device (10) of claim 12 as a field effect transistor, a photodetector, a diode, a logic element, a photovoltaic device, an electroluminescent light emitting device, a solution gated field effect transistor, a two-dimensional materialbased sensor, or a saturable absorber.

Images