Improvement of deposition of high-quality metal oxide and metal nitride layers

Abstract

Intermittent catalytic reaction-induced surface processes (ICRISP) produce metal oxide or metal nitride films or metal-acid-nitride films. In this process, the metal-containing precursor is intermittently added together with a nitrogen-containing compound or a sulfur-containing compound. Appropriate apparatus for this process is described below.

Description

[Technical Field]

[0001] Related applications This application claims priority from U.S. Provisional Patent Application No. 63 / 523,579, filed on 27 June 2023.

[0002] Introduction Dielectric layers and metal electrodes of oxides and some nitrides are crucial for the performance and function of microelectronic devices such as transistors and memory capacitors. For example, the gate dielectric layer and gate electrode layer are essential components required for the operation of metal-oxide-semiconductor field-effect transistor (MOSFET) devices. Similarly, dynamic random-access memory (DRAM) capacitors require dielectric layers and internal electrodes, which are used for charge storage within the capacitor and access to the stored charge. To accommodate the shrinking of devices, there is a growing demand for improvements in the deposition processes of high-dielectric-constant (high-k) materials, particularly high-k materials used as gate insulators and capacitor dielectrics as alternatives to SiO2. Zirconium dioxide (ZrO2) has already proven to be a suitable high-k material and is used industrially due to its wide bandgap (approximately 5.8 eV), high dielectric constant (17–47), and high thermal stability. Hafnium dioxide (HfO2) is a noteworthy new material for use as a dielectric barrier in high-power SiC and GaN electronics, both MOSFETs and HEMTs, due to its even higher bandgap, dielectric constant, and thermal stability. Silicon nitride (Si3N4, Si x N y , or abbreviated as SiN), zirconium nitride (Zr3N4, Zr x N yMetal nitrides (or ZrN for short), other metal nitrides, and a broader range of metal oxynitrides also provide sufficient dielectric properties to be applicable to certain types of microelectronic devices. The reduction in device size is accompanied by an increase in aspect ratio, which means that highly efficient thermal ALD systems and processes are increasingly advantageous over plasma-based processes that involve reactive species that may disappear before being exposed to high-aspect-ratio, hidden, or other hard-to-reach surfaces. As with many ALD processes, the success of a coating that can be produced for a particular device or feature often depends on the specific equipment used for deposition.

[0003] Existing thermal ALD systems grapple with the trade-off between the requirements of reducing reaction time and improving chemical utilization efficiency, while simultaneously minimizing purge gas residence time and chemical removal time. Certain prior art ALD systems include a chemical supply manifold that uses the synchronous operation of multiple valves. In such systems, it is virtually impossible to completely synchronize the operation of the valves, making it impossible to satisfactorily eliminate flow fluctuations. As a result, unavoidable flow fluctuations are known to cause gas backflow and lead to the mixing of harmful chemicals.

[0004] Conventional ALD (Advanced Liquid Processing) systems tend to experience a decrease in efficiency due to a "memory" effect. This memory effect is caused by the adsorption of chemicals onto the walls of the ALD reactor, resulting in their release from the reactor walls over a timescale determined by the adsorption energy and wall temperature. This phenomenon tends to increase the residence time of trace amounts of chemicals within the ALD reactor. Consequently, the memory effect tends to increase the purging time required to remove the chemicals. Therefore, an ALD system that minimizes the memory effect is needed.

[0005] Therefore, there is a need for an ALD system that uses a thermal ALD process with lower ownership costs, can shorten reaction times without sacrificing precursor utilization efficiency, minimizes purge gas residence time and chemical removal time, and can improve the deposition of high-k dielectric materials (oxides, nitrides, and oxynitrides).

[0006] Current techniques for depositing many high-k materials by thermal atomic layer deposition (ALD) tend to produce low-density, low-performance films, leading many microelectronics manufacturers to adopt plasma-enhanced ALD (PEALD) processes instead. While PEALD is employed to improve film performance, the high electric field and fast switching requirements of devices can result in insufficient dielectric barriers resulting from PEALD. Furthermore, not all applications can withstand plasma, damaging incoming substrates and degrading their performance. Most importantly, plasma processes have inherent limitations on the aspect ratio at which conformable coatings can be achieved. A short distance from the aperture, typically 10:1, i.e., about 10 times deeper than the aperture diameter or aperture distance, sufficient concentrations of plasma species cannot interact with the surface. Because PEALD relies on the amount of radical species to adjust the film saturation per cycle and composition, PEALD films lose thickness conformability below this point (less than 10 times the aperture depth). More importantly, the film composition also gradients depending on the concentration of plasma-active species as the feature moves downwards. This dynamic composition can cause device failure in critical areas such as the corners at the bottom of the feature. Therefore, there is a need to further improve and refine thermal ALD processes that can produce excellent dielectric barriers under low thermal loads.

[0007] For example, Sneh et al. (US2003 / 0180458, incorporated herein in whole by reference) teach about an apparatus that helps solve some of the above problems by using a flow-draw synchronous modulation ("SMFD") approach in a thermal ALD apparatus. In the SMFD approach, the inflow of process gas into the ALD chamber is called the "flow," and the discharge of gas from the ALD chamber is called the "draw." Under steady-state conditions, the draw generally coincides with the flow, while under transient conditions, the flow and draw can be "mismatched." A key aspect of Sneh's invention was addressing the trade-off in conventional ALD systems between the conflicting requirements of high flow rates during purging the deposit chamber and low flow rates during chemical injection. SMFD provides the ability to adjust pressure and gas flow rates in a rapid response by purging the process chamber with low pressure and high purge gas flow rates, followed by chemical injection into the process chamber with high pressure and low flow rate chemical reaction gases. Sneh's apparatus teaches a means for adjusting the pressure drop (ΔP) between the ALD chamber and the downstream subsystem while maintaining the ALD chamber pressure substantially constant. The draw control chamber is incorporated into the ALD apparatus and located downstream of the ALD chamber, and the draw control chamber controls conductance, C Draw It has an outlet. By flowing the control gas through it into the draw control chamber, P DrawThe draw control can be controlled independently of the pressure in the upstream ALD chamber. The gas flow into the draw control chamber is the sum of the draw from the process chamber and the flow of directly injected draw gas. Since the draw is a function of ΔP, controlling ΔP independently made the draw substantially match the flow, independently of the flow. In principle, the draw chamber can be made very small, so adjustment of ΔP in certain embodiments can be carried out at a sub-millisecond rate. Choosing a small-volume draw control chamber facilitates process conditions with short transient times, while employing a larger draw control chamber offers the advantage of providing secondary benefits to the system (e.g., trapping or removing particularly hazardous or expensive chemicals) at the expense of draw control speed. In Sneh's apparatus, the draw control response time was in the range of 10 to 20 milliseconds in a typical 200 mm wafer deposition apparatus. Since 300mm wafer deposition equipment is well-established in semiconductor manufacturing, it was determined that Sneh's equipment, which can achieve a processing time range of 10-20 milliseconds, would not provide the necessary inter-plane uniformity required for manufacturing 300mm wafers.

[0008] However, Sneh's SMFD approach facilitated a catalytic reaction for an induced surface process ("CRISP"), a then-novel type of ALD method, taught in US2003 / 016800 and incorporated herein by reference as a whole. Surface catalysis (CRISP) can function as a source of hydrogen atoms and other reactants necessary for nonmetallic ALD reactions. In this regard, the chemical process is enhanced by the addition of one or more sequential chemical reactions that can generate intermediate reactive molecular fragments. These reactive intermediates further react in the chemical process. The sole purpose of these additional side reactions is to generate reactants for the original process. CRISP is designed to provide a volatile pathway for adsorbed reactants in the event that the surface reaction sites are depleted. Therefore, CRISP is robust and can be over-excessive beyond saturation to ensure reproducible and satisfactory results, but it was limited by the availability of suitable precursors, the speed and flexibility of the SMFD process in the equipment available at the time, and the ability to produce more desirable reactive intermediates that can achieve the more stringent technical requirements for the production of 300 mm wafers, such as inter-plane uniformity, film density, refractive index, and dielectric constant.

[0009] Sneh's initial continuous CRISP process relied on unsaturated surface chemical reactions, referred herein as “catalytic reactions.” Sneh’s catalytic reactions did not deposit solid material onto the surface by themselves. A catalytic reaction requires two or more reactive chemicals, referred herein as “catalytic reactants,” which react with each other, preferably violently, to produce stable volatile byproduct molecules and unstable surface-adsorbed intermediate reactive molecular fragments. The catalytic reaction is thermodynamically driven and irreversible based on the volatilization of the stable byproducts. The reactive molecular fragments produced in the catalytic reaction generally form as adsorbed radical species. These adsorbed reactive molecular fragments are atomic or molecular parts, or both. Nevertheless, if the surface contains reaction sites, the reactive fragments react with the reaction sites in fragment-surface reactions, usually producing volatile surface byproducts, or being incorporated into the grown film, or both. Thus, the catalytic reaction generally proceeds (stepwise) from the catalytic reactants, from both the catalytic reaction and the surface reaction sites, to the final volatile species. Sneh's catalytic reaction was distinctly continuous, not intermittent. Nevertheless, in most methods, including a continuous catalytic reaction is part of a process on a substrate with saturating properties, e.g., an ALD reaction step, a surface treatment step, or a surface cleaning step. Thus, Sneh's catalytic reaction was uniquely different from saturated ALD or prior art substrate treatment processes. When a saturated process driven by a continuous catalytic reaction reaches saturation, the catalytic reaction has further unique properties; namely, they have an undamaging pathway for the complete volatilization of all parts of the catalytic reactant. A saturated process only approaches saturation in practice (since ideal saturation requires infinite time), but they are clearly converging. In contrast, a catalytic reaction continues as long as the corresponding catalytic reactant is present. While this approach had clear technical advantages at that point, it proved to introduce inherent defects in inter-plane uniformity as the diameter of the substrate increased, which increased the distance between the introduction of the catalytic reactant into the reactor and the draw control of the catalytic reactant to remove any reactive precursors or catalytic reactants. [Overview of the project]

[0010] A novel conversion process called the "ICRISP" process (Intermittent CRISP) was developed to improve the deposition of metal oxides, metal nitrides, and metal oxynitrides with a controllable O:N ratio via thermal ALD. Leveraging the system's unique ability to intermittently introduce small amounts of non-metallic catalyst during conversion half-cycles, the novel ICRISP process further utilizes surface catalysis to increase growth per cycle, improve stoichiometry, increase density, and modify crystalline morphology compared to prior art metal oxide films. Figure 1 and Table 1 below show the improvement of HfO2 films grown using the prior art metal oxide film deposited at the same temperature (250°C) as the prior art film, using the prior art CRISP method with continuous injection of O3 and catalyst molecules.

[0011] The present invention overcomes the shortcomings of prior art SMFDs in at least two ways: 1) by utilizing a pressure controller to improve the flow and flexibility of the inlet gas source in synchronization with the draw control of the prior art, and 2) by utilizing a substrate chuck having a controllable height to shorten the distance between the substrate and the inlet(s)(multiple) to increase the internal volume of the reaction chamber beneath the substrate.

[0012] This CRISP technology offers superior time constant control (less than 1 ms to 10 ms for a 200 mm wafer system compared to prior art) and the ability to periodically pulse different gas compositions into the reaction chamber. A non-limiting example of ICRISP is the periodic and / or intermittent ingestion of small amounts of monomethylhydrazine (MMH), for example, about 0.03 Langmuir (from a 5 ms MMH pulse for every 45 ms O3), which can produce film properties superior to those of prior art.

[0013] With ICRISP, it becomes possible to handle substrates of various heights / dimensions without physically adjusting the tool externally. The tool is configured by computer control to set the distance between the substrate and the precursor supply device (e.g., a showerhead) to 24 mm or less, or less than 10 mm, preferably less than 5 mm, more preferably 2 - 5 mm. In some embodiments, for example, in a thermal ICRISP process, it can be shortened to 1.5 - 2.5 mm. In embodiments including a plasma-enhanced process, it may be advantageous for the distance between the plasma supply device and the substrate to be greater to promote deposition uniformity. In all embodiments, the height-adjustable chuck of the present invention facilitates the use of alternative showerhead shapes, such as those having a concave shape and / or a slight dome that enhance the draw control features of the system and device. The concave shape, in combination with pressure control, facilitates a high pressure difference between the draw control zone and the ALD process zone that could not be achieved in prior art devices or systems.

[0014] In one aspect, the present invention provides an intermittent catalyst reaction-induced surface process (ICRISP) for forming an oxide or nitride film on a substrate, the process comprising providing a substrate in a chamber, adding a metal-containing ALD precursor to the substrate to create a first added substrate, purging or flashing the chamber, and either A) adding an oxidizing agent and a nitrogen-containing reactant to the first added substrate to form a metal oxide, or B) adding a sulfur-containing reactant and a nitrogen-containing reactant to the first added substrate to form a metal nitride, and purging or flashing the chamber.

[0015] In another aspect, the present invention provides an Intermittent Catalytic Reaction Induced Surface Process (ICRISP) for generating a film having a thickness T and a composition MxOyNz from a metal-containing precursor, an oxygen-containing precursor, and a nitrogen-containing precursor at a temperature of 300 degrees or less, where x is 1 or 2, y is 0 to 4, z is 0 to 2, and the process comprises: a) adding the metal-containing precursor for a pulse time of 5 to 500 ms; b) purging and / or flashing the metal-containing precursor for a time of 250 to 30,000 ms; c) when y is greater than 0, i. adding the oxygen-containing precursor for a time t3; ii. adding the nitrogen-containing precursor incorporated in the oxygen-containing precursor for a time t4, provided that t4 < t3; iii. adding the oxygen-containing precursor for a time t5, provided that t5 > t4; iv. optionally, repeating steps c.ii and c.iii the number of times necessary to convert at least 99% of the metal-containing precursor deposited in step a) into the MxOy layer; d) when z is greater than 0, i. adding the nitrogen-containing precursor for a time t6; ii. adding the oxygen-containing precursor incorporated in the nitrogen-containing precursor for a time t7, provided that t7 < t6; iii. adding the nitrogen-containing precursor for a time t8, provided that t8 > t7; iv. optionally, repeating steps d)ii and d)iii the number of times necessary to convert at least 99% of the metal-containing precursor deposited in step a) into the MxNz layer, and repeating steps a to d until the desired thickness T is obtained.

[0016] In any aspect of the present invention, it can be further characterized by one or any combination of the following: the process produces a metal oxide or metal nitride film or a metal-acid-nitride film, the oxide film may have 0.1% by mass or less or 0.01% by mass or less nitrogen, and the process according to claim 1 for producing an oxide film comprises adding an oxidizing agent and a nitrogen-containing reactant to the first added substrate to form a metal oxide, the adding step being a subpulse with the oxidizing agent without the nitrogen-containing reactant, and then the acid The process according to claim 1 for producing a nitride film comprises a second subpulse with a oxidizing agent and a nitrogen-containing reactant, followed by a third subpulse with an oxidizing agent without the nitrogen-containing reactant, wherein the oxidizing agent comprises ozone, and the nitrogen-containing reactant comprises hydrazine, and the step of adding the oxidizing agent to the first added substrate further comprises the addition of H2S, and the process according to claim 1 for producing a nitride film comprises adding a sulfur-containing reactant and a nitrogen-containing reactant to the first added substrate to form a metal nitride, wherein the adding step comprises a subpulse with the nitrogen-containing reactant, followed by the sulfur-containing reactant and the nitrogen-containing reactant The reaction comprises a second subpulse with a nitrogen reactant, followed by a third subpulse with the nitrogen-containing reactant, wherein the oxide film comprises Al2O3, SiO2, TiO2, Nb2O5, Ta2O5, La2O5, Y2O3, ZrO2, Ga2O3, or In2O3, the product comprises a mixed oxynitride and / or mixed metal, the nitride film comprises ZrN, GaN, YN, InN, or Si3N4, and the sulfur-containing reactant comprises H2S (hydrogen sulfide), H2S2 (dihydrogen sulfide), mercaptan (HS(CH3) or methanethiol), ethanethiol (ethyl mercaptan). The base material comprises tan, S(CH3)2(dimethyl sulfide), thionyl chloride (SOCl2), sulfuryl chloride (SO2Cl2), the base material comprises Si, SiO2, Al2O3 (preferably in the form of corundum), SiC, GaN, AlGaN, GaAs, or InP, the oxidizing agent comprises O2, O3, H2O, H2O2, or N2O (the categories described as "contains" may include additional components which can be mixed with the oxidizing agent), the nitrogen-containing reactant comprises hydrazine, monomethylhydrazine (MMH), 1,1-dimethylhydrazine, 1,The process comprises 2-dimethylhydrazine, tert-butylhydrazine (tBuNNH), ammonia (NH3), pyridine, 2,3-lutidine (2,3-dimethylpyridine), 2,4-lutidine (2,4-dimethylpyridine), 2,5-lutidine (2,5-dimethylpyridine), 2,6-lutidine (2,6-dimethylpyridine), 3,4-lutidine (3,4-dimethylpyridine), or 3,5-lutidine (3,5-dimethylpyridine), and the process is carried out for 10 to 5000 cycles, or 10 to 1000 cycles, or 20 to 200 cycles, and the ALD precursor comprises Ti-TiCl4, TTIP (titanium(IV) isopropoxide), Ti(NEtMe)4 (titanium(IV) tetrakisethylmethylamine), Ti(NMe)4 (titanium(IV) tetrakisdimethylamine), Hf-Hf(NEtMe)4 (Hafn This includes (Hafnium(IV)tetrakisethylmethylamine), HyALD (Hafnium(IV)cyclopentadienyl-trisdimethylamine), Zr-Zr(NEtMe)4 (Zirconium(IV)tetrakisethylmethylamine), ZyALD (Zirconium(IV)cyclopentadienyl-trisdimethylamine), Si-BEMAS (Bisethylmethylaminosilane), BDEAS (Bisdiethylaminosilane), 3DMAS (Trisdimethylaminosilane), Y-ArYA (Yttrium(III)bisethylcyclopentadienyl-isopropylamidinate), Al-TMA (Trimethylaluminum), AlCl3, AlMeCl2, and AlMe2Cl (ALD precursors are usually metals having ligands for organic groups, alkoxy groups, and / or heteroatom groups, and the organic groups are usually typically C1-C, 12or C1-C6 alkyl, aryl, arylalkyl, and heteroatoms suitable for the ALD precursor include H, amide, alkylamide, and halogens such as Cl. The step of adding the metal-containing ALD precursor to the substrate includes introducing the precursor into the chamber, and then simultaneously performing a pulse step of introducing the precursor and a draw control. Here, the draw control includes the flow of an inert gas and the application of a vacuum. The pulse step is performed for 5-50 ms, and then a step of 10-500 ms where there is no flow of the precursor in the system. Here, the adding step is performed for 150 ms or less, and the inert gas is flowed through the chamber for at least 200 ms. In this case, preferably, the precursor is not introduced into the chamber, the vacuum is not applied to the chamber, and the sub-pulse by the oxidant without the nitrogen-containing reactant is performed for a longer time than the sub-pulse by the oxidant and the nitrogen-containing reactant (these sub-pulses are summed for this calculation). The sub-pulse by the oxidant without the nitrogen-containing reactant is performed for at least twice the time of the sub-pulse by the oxidant and the nitrogen-containing reactant. The sub-pulse by the oxidant without the sulfur-containing reactant is performed for a longer time than the sub-pulse by the oxidant and the sulfur-containing reactant (these sub-pulses are summed for this calculation). The sub-pulse by the oxidant without the sulfur-containing reactant is performed for at least twice the time of the sub-pulse by the oxidant and the sulfur-containing reactant.

[0017] The present invention also includes any of the apparatus, systems (which may include apparatus and compositions and conditions within the apparatus) described herein, as well as any final or intermediate compositions. For example, the present invention includes apparatus for an ICRISP process having a precursor showerhead, an ALD reaction zone, a dynamically positionable chuck whose height (distance between the showerhead and the substrate) of the ALD reaction zone can be varied between 10 mm and 1.0 mm using computer control, a first series of high-speed pneumatic valves for pressure control upstream of the precursor showerhead, and a second series of high-speed pneumatic valves for draw control downstream of the ALD reaction zone. The dynamically positionable chuck (see Figure 6) floats on a piston-type linear actuator driven by a single stepper motor and belt assembly. The gear ratio of the assembly allows for real-time fine adjustment with a resolution of up to 0.1 mm. The present invention includes a system capable of performing a series of steps for an ICRISP process, including delay, pulse, dosing, purging, and pump purging. The present invention also includes a film composition MxOyNz deposited using an ICRISP process, wherein the deposition time of the ICRISP process is less than 6 seconds per cycle, the film uniformity exceeds 92%, and the x:y and x:z ratios and density are within + / - 10% of the bulk values ​​of the MxOy metal oxide and MxNz metal nitride.

[0018] The present invention also includes systems comprising any of the compositions and conditions described herein. The present invention may become even more apparent in the following examples. The present invention may be further characterized by any of the features in such examples. For example, any aspect of the present invention may be further characterized by values ​​within ±30%, ±20% (or ±10%) of any of the values ​​in such examples, tables, or figures, but the scope of the present invention is not intended to be limited by these examples in its broader embodiments. All ranges include boundary values ​​and are combinable. For example, if the range “1 to 5” is described, the described range should be interpreted to include the ranges “1 to 4”, “1 to 3”, “1 to 2”, “1 to 2 and 4 to 5”, “1 to 3 and 5”, “2 to 5”, 1, 2, 3, 4, or 5 individually, etc.

[0019] The present invention is often characterized by the term "comprising," meaning "including," without excluding additional components. In a narrower embodiment, the term "comprising" may be replaced by the more restrictive terms "consisting essentially of" or "consisting of." [Brief explanation of the drawing]

[0020] [Figure 1] X-ray diffraction patterns of HfO2 from prior art (O3 only - dark) and a novel ICRISP process (orange) over 20–60°²Θ. Note that equivalent peaks are of higher intensity, and since both films have equivalent thickness, this indicates a higher degree of crystallinity. Furthermore, in the ICRISP film, the (002) peak at approximately 35° is sharper than in the prior art, indicating a larger particle size. Finally, compared to the prior art, two new peaks appear in the ICRISP HfO2 film at 28.7° and 57°²Θ. This indicates that ICRISP modified the crystalline structure or orientation of the grown film. [Figure 2]Shows the flow of the precursor to the deposition zone. [Figure 3] Shows a central pulse step with two inert gas flows on both sides. [Figure 4] Shows two cross-sectional views of one high-speed pneumatic valve (FPV) used in the present invention, namely, in the closed (100a) state during operation and the open (100b) state during operation. [Figure 5] Shows cross-sectional views of two FPVs used in the present invention for two different precursors. FPV-1 is in the closed state during operation, and FPV-2 is in the open state during operation. [Figure 6] Shows a dynamically configurable chuck.

[0021] Metal oxides by ICRISP The time and cadence of the ICRISP process of the present invention are described for metal oxides, namely, MO x for the metal-containing precursor MR x (where R represents an organic ligand), and an O-containing precursor without metal, such as O3, and a metal-free N-containing cocatalyst, such as MMH. Step 1: MR x The precursor pulse for a time of t1 Step 2: The purge and / or flash reaction stage for a time of t2 Step 3a: O3 sub-pulse 1 for a time of t3 Step 3b: O3 and MMH (incorporated) sub-pulse 1 for a time of t4 Step 3c: O3 sub-pulse 2 for a time of t5 Step 4: The purge and / or flash reaction stage for a time of t6 Here, t3 and t5 are greater than t4.

[0022] The inventors unexpectedly improved the deposition rate and quality of metal oxide films deposited by ALD using the ICRISP process at a given temperature by intermittently adding a small amount of nonmetallic catalyst to the conversion step in the thermal ALD process. Specifically, when O3 was periodically incorporated, i.e., every 30-60 ms, preferably every 35-55 ms, more preferably every 40-50 ms, and sometimes every 45 ms, in small amounts, i.e., about 0.01-0.05 Langmuirs, preferably 0.02-0.04 Langmuirs, and more preferably 0.025-0.035 Langmuirs, in short pulse times of 2-10 ms, preferably every 4-7 ms, more preferably every 5-6 ms, and sometimes every 5 ms, monomethylhydrazine (MMH) was observed in O3 in amounts of about 0.01-0.05 Langmuirs, preferably every 0.02-0.04 Langmuirs, and more preferably every 0.025-0.035 Langmuirs, in the deposition of HfO2 by cyclopentadienyltrisdimethylaminehafnium ("HyALD"), different growth rates and film characteristics were observed compared to O3 alone. In some embodiments, the ratio of pulse duration to MMH incorporation period ranges from 0.5:20, preferably 0.75:15, more preferably 1:10, 1:9, or 1:8. For films grown at 250°C, the growth per cycle (GPC) at saturation increased by 29% from 0.77 to 0.99 Å / cy (measured by spectroscopic ellipsometry). Furthermore, the refractive index at 633 nm increased from 1.977 to 1.993. X-ray reflectance (XRR) measurements and modeling showed values ​​from 8.3 to 8.9 g / cm³. 3 A 7% increase was observed. Bulk HfO2 has a density of 9.63 g / cm³, while the published density result for the HyALD / O3 process at 250°C is 7.7 g / cm³. 3 ( https: / / aip.scitation.org / doi / full / 10.1063 / 5.0055847 ). As described in other data, ALD × The films deposited on the toolset appear to have a higher density than those deposited using the same ALD process at the same temperature. Unexpected density increases were also observed in the ICRISP processes based on TiO2 and Y2O3. Compared to the cross-flow reactor, the density of the HfO2 CRISP process ranged from 7.7 to 8.9 g / cm³. 3This increases by 15.6%. Films approximately 100 nm thick deposited using both O3 alone and the ICRISP process were polycrystalline in their deposited state, as measured by X-ray diffraction (XRD). This is in contrast to several literature reports that showed amorphous films at 250°C for the HyALD / O3 process. x Both processes in the tool were crystalline, but the ICRISP process yielded a more crystalline film, as shown in Figure 1, as indicated by the larger peak intensity when normalized against thickness. A rough comparison of Scherrer's equations also shows that the ICRISP process has 10–18% larger grains. One interesting and unexpected observation was the difference in crystalline structure between the O3 and ICRISP processes. At least two peaks are present in the ICRISP diffractogram but not in the O3 alone diffractogram. These are located at 28.7° and 56.0°²Θ and are associated with (1 bar 11) and (3 bar 11) orientations, respectively. These data suggest that the intermittent addition of N-based molecules such as MMH, or potentially other Lewis bases, may promote specific crystal orientations, and that surface mobility may be affected by the addition of other species during the ALD process. X-ray photoelectron spectroscopy (XPS) was used to confirm carbon and nitrogen content and compare O:Hf ratios (stoichiometry). Films deposited with ICRISP had lower carbon content, i.e., 1.8 to 3.2 atomic percent, and a more ideal O:Hf ratio, i.e., 2.2 to 2.4 (2.0 is ideal). Despite all ICRISP ALD cycles involving the intermittent incorporation of nitrogen-containing molecules MMH, no nitrogen was detected in any of the samples. [Table 1]

[0023] Improvement of dielectric constant is due to grain size ( https: / / www.sciencedirect.com / science / article / abs / pii / S1359645403000521 ) and impurities (both oxygen and carbon) https: / / www.sciencedirect.com / science / article / abs / pii / S0167931711003583?via%3Dihub It is directly related to ).

[0024] While both the CRISP film of the prior art and the ICRISP film of the present invention were crystalline in their deposited state, several reports indicated that they became amorphous under a 250°C process. Despite the higher density and polycrystalline nature of the O3 process of the prior art, the carbon content was nearly identical to the published data. This suggests that the carbon removal mechanism is driven by temperature and has a rate barrier independent of other factors that increase the density and packing of the grown film (e.g., surface catalysis utilized to form Hf-O bonds). However, the density of the ICRISP film was higher than that of either of the prior art processes and closer to the bulk density than expected. This crystalline structure of ICRISP, distinct from O3 alone, is highly unexpected and suggests a potential mechanism for promoting crystal orientation using ICRISP ALD, independent of the substrate / template effect common in the prior art, which is not typically achievable in pure epitaxial growth processes.

[0025] Assuming that the present invention increases the grain size, decreases the carbon content, and improves the Hf:O ratio, the dielectric constant of films obtained using the ICRISP process is improved and becomes more controllable. Furthermore, leakage current, a key characteristic of high field dielectrics found in power electronics, is improved along with the decrease in impurity content. https: / / avs.scitation.org / doi / 10.1116 / 1.5134135 ), leakage current is also reduced using the ICRISP process. The inventors further discovered that by proceeding to a deposition temperature of 300°C using the HyALD ICRISP process, all impurities are removed, density is further increased, and crystallinity is improved without forming intermediate metal nitride species embedded throughout the ICRISP layer.

[0026] The inventors also used this novel ICRISP process to deposit layers containing other metal oxides (such as Al2O3, SiO2, TiO2, Nb2O5, Ta2O5, La2O5, and Y2O3), and confirmed similar improvements in dielectric constant, leakage current, and dielectric breakdown voltage compared to prior art processes. This ICRISP process has excellent dielectric properties and can be controlled by both pressure control and draw control. xGiven the speed of the toolset and the low precursor consumption compared to other prior art tools, this process can be used for multiple applications for dielectric barriers, either as a single-layer or multilayer structure. New applications for precision optics and mirrors also exist, given the added high refractive index (RI) and the ability to precisely tune the RI based on ICRISP parameters. During development, the prior art CRISP process was examined using MMH, confirming that the RI could shift between the O3-only state and the prior art CRISP state. Subsequently, during the development of the ICRISP process, it was possible to further increase the RI between the prior art CRISP state and the novel ICRISP state, for example, by developing MMH, pressure control, draw control, and chuck height duration and sequencing. The present invention provides a further set of parameters for tuning for optical and capacitive applications when deployed in the production of metal oxides using monolayers and multilayers of ICRISP. [Table 2]

[0027] Metal nitride by ICRISP More broadly, the time and cadence of the ICRISP process of the present invention are related to metal nitrides, i.e., MN x Regarding metal-containing precursors, MR * x (R * The formulation may be described using an inorganic ligand (where is an inorganic ligand, e.g., a halide), a metal-free N-containing precursor, e.g., MMH, and a metal-free S-containing co-catalyst, e.g., H2S. Step 1: MR * x Precursor pulse for time t1 Step 2: Purge and / or flush reaction stage at time t2 Step 3a: MMH subpulse 1 for time t3 Step 3b: Subpulse 1 of MMH and H2S (taken in) at time t4 Step 3c: MMH subpulse 2 for time t5 Step 4: Purge and / or flush reaction step for t6 time Here, t3 and t5 are greater than t4.

[0028] Thermal ALD of metal nitrides using halide precursors and metal-free N-containing precursors is widely common in the prior art, but high temperatures are usually required to achieve high-quality nitride deposition. Plasma-enhanced ALD can be used to produce higher-quality nitride layers, but as described above, in embodiments involving a plasma-enhanced process, the distance between the plasma supply and the substrate should be increased to improve deposition uniformity, thereby eliminating the significantly reduced process time and precursor consumption achieved by the ICRISP process.

[0029] Extensive verification of the ability of the ICRISP process to lower the operating temperature of a thermal ALD approach has been carried out using TiN as the metal nitride. TiCl4 and NH3 at 380°C is a conventional process taught by prior art. The ICRISP process can deposit high-quality TiN films by intermittently incorporating extremely short H2S pulses using TiCl4 and MMH below 300°C. A further advantage of developing a process for nitrides that can be operated below 300°C is that these temperature ranges allow for the deposition of many other oxides for multilayer coatings containing oxides, without having to lower the temperature in the ALD system from above 380°C to below 300°C and then raise it back above 380°C if a subsequent nitride layer is required. For example, SiO2 using ICRISP can be deposited using trisdimethylaminosilane and O3 and intermittently incorporated MMH, and Si using ICRISP x N y(Typically Si3N4) can be deposited at 300°C using trisdimethylaminosilane with MMH and incorporated H2S, Al2O3 using ICRISP can be deposited at 300°C using trimethylaluminum with O3 and intermittently incorporated H2S, and AlN using ICRISP can be deposited at 300°C with trimethylaluminum with MMH and incorporated H3S or trimethylaluminum with hydrazine and incorporated H3S. A similar ICRISP process can be used for ZrO2 vs Zr x N y This has been demonstrated for Ga2O3 vs. GaN, Y2O3 vs. YN, In2O3 vs. InN, etc., using Si, SiO2 (sapphire), and SiC substrates. Furthermore, the ICRISP process has been demonstrated for polymetallic materials of similar quality, such as Al x Ga (1-x) N and In x Ga (1-x) N can be easily produced using MMH as a nitrogen-containing precursor at 300°C, with the intermittent incorporation of a co-catalyst, such as O3. [Table 3]

[0030] A general description of prior art atomic or molecular layer deposition. Atomic layer controlled growth technology allows for the deposition of a coating with a thickness of approximately 0.1 to 5 angstroms per reaction cycle, providing a means to very precisely control the surface coverage or coating thickness. By repeating the reaction sequence and sequentially depositing additional layers of coating material until the desired coating thickness is achieved, thicker coatings can be prepared.

[0031] The coating is deposited using atomic layer deposition (ALD) or molecular layer deposition (MLD) processes. In the ALD / MLD process, the coating formation reaction proceeds as a series of (typically) two half-reactions. In each of these half-reactions, a single reagent (precursor) is introduced to come into contact with the substrate surface. The conditions are such that the reagent is in gaseous form. In most cases, the reagent reacts with functional groups on the surface of the substrate and binds to the substrate. Because the reagent is a gas, it penetrates into the pores within the substrate and deposits on the inner surface of the pores and the outer surface of the substrate. This precursor is designed to react with the surface at all available surface sites, but does not react with itself. Thus, the first reaction takes place, forming a single monolayer, or sub-monolayer, and creating new surface functionality. Next, any excess reagent is removed. This helps prevent the growth of undesirable, larger inclusions in the coating material. Then, each of the remaining half-reactions proceeds in sequence, with the first reagent being introduced each time, reacting with the substrate surface, and any excess reagent being removed before introducing the next reagent. Typically, reagents are introduced using an inert carrier gas, and the reaction chamber is usually swept with the carrier gas during the continuous addition of reagents to help remove excess reagents and gaseous reaction products. Vacuum may be applied during and between the continuous addition of reagents to further remove excess reagents and gaseous reaction products.

[0032] After being exposed to a first precursor, the surface is exposed to a second precursor, which is typically also dispersed in an inert carrier gas. This precursor is designed to react with the functional groups positioned in the first reaction step. This reaction continues until all available surface sites react. The second precursor does not react with itself. Any excess second precursor is removed in an optional inert gas purging step. If the gas is properly metered, the purging step may not be necessary. This can be a process of at least four steps (precursor 1, purge, precursor 2, purge) to deposit a single monolayer of the growing film. This does not mean only a single precursor, as some ALD and MLD processes use multiple reactants in a single step, such as APTES / H2O / O3 for depositing SiO2. This process is repeated as many times as necessary to form the desired film thickness. The ALD / MLD process may be initiated with a "linker" agent or pretreatment gas (such as ozone) that promotes covalent bonding to the surface, or terminated with a hydrophobic, hydrophilic, or purpose-specific termination agent.

[0033] The reaction conditions are selected to meet three main criteria. The first criterion is that the reagents are in gaseous form under the reaction conditions. Therefore, the temperature and pressure conditions are selected so that the reactants volatilize before the reaction. The second criterion is reactivity. The conditions, in particular the temperature, are selected so that the desired reaction between the film-forming reagent (or, at the start of the reaction, between the initially introduced reagent and the substrate surface) occurs at a commercially reasonable rate. The third criterion is that the substrate is thermally stable from a chemical and physical standpoint. The substrate must not decompose or react at the process temperature, except in cases where it may react with one of the ALD precursors on its surface functional groups in the early stages of the process. Similarly, the substrate must not melt or soften at the process temperature, thereby maintaining the physical shape of the substrate, particularly its pore structure. The reaction is generally carried out at a temperature of about 270–1000K, preferably 290–450K, where the specific temperature in each case is below the temperature at which the substrate melts, softens, or decomposes.

[0034] While reagents are continuously added, the substrate is exposed to conditions sufficient to remove reaction products and unreacted reagents. This is, for example, by exposing the substrate to approximately 10 minutes of water after each reaction step. -5 This can be achieved by exposing the material to a high vacuum of Torr or higher. Another method to achieve this, which is more readily applicable to industrial applications, is to sweep the substrate with an inert purge gas between reaction steps. This purge gas can function as a fluidizing medium for the substrate and as a carrier for the reagents.

[0035] Several techniques are useful for monitoring the progress of a reaction. For example, vibrational spectroscopy studies can be performed using transmission Fourier transform infrared techniques. Deposited coatings can be investigated using in situ spectroscopic ellipsometry. Atomic force microscopy studies can be used to characterize the roughness of the coating relative to the roughness of the substrate surface. X-ray photoelectron spectroscopy and X-ray diffraction can be used to profile the depth of the coating and to confirm its crystal structure.

[0036] As shown in reaction sequence A1 / B1, the aluminum oxide coating is easily deposited using trimethylaluminum and water as precursors. The reactions shown are not in equilibrium and are intended to demonstrate only reactions at the substrate surface (i.e., no interlayer or intralayer reactions). Substrate -XH* + Al(CH3)3 = Substrate -X - Al* - CH3 + CH4 (Precursor reaction) Substrate-X-Al*-CH3+H2O=Substrate-X--Al-OH*+CH4(A1) Substrate-X-Al-OH*+Al(CH3)3=Substrate-X-Al-OAl*-CH3+CH4(B1)

[0037] In reaction A1 / B1, X is typically oxygen, nitrogen, or sulfur, and an asterisk (*) indicates a surface species that may undergo the next half-reaction. The aluminum oxide film is formed by alternating between reactions A1 and B1 until the desired coating thickness is achieved. Using this reaction sequence, the aluminum oxide film tends to grow at a rate of approximately 0.1 nm / cycle.

[0038] Titanium oxide coatings are readily deposited using titanium tetrachloride and water and / or hydrogen peroxide as precursors, as illustrated in reaction sequence A2 / B2. As before, the illustrated reactions are unbalanced and are intended solely to demonstrate reactions at the substrate surface (i.e., not interlayer or intralayer reactions). Substrate-XH* + TiCl4 = Substrate-X--Ti*-Cl3 + HCl (precursor reaction) Substrate-X--Ti*-Cl3+H2O2=Substrate-XT- i*-OH+HCl+Cl2(A2) Substrate-X-Ti*-OH+TiCl4 Substrate-X-Ti-O-Ti*-Cl3+-HCl (B2)

[0039] In reaction A2 / B2, X is typically oxygen, nitrogen, or sulfur, and the asterisk (*) indicates a surface species that may undergo the next half-reaction. The titanium oxide film is formed by alternately repeating reactions A2 and B2 until the desired coating thickness is achieved. Using this reaction sequence, the titanium oxide film tends to grow at a rate of approximately 0.05–0.1 nm / cycle.

[0040] As is known in ALD / MLD processes, the order may be AB, ABC, ABCD, ABCDABABCD, or any desired order if the chemical entities react with each other in the desired order. Each reactant has at least two reactive moieties (this includes the possibility that the reactant can be modified to have two reactive moieties, such as having a first reactive moiety and a second reactive moiety that are temporarily blocked by protecting groups or require subsequent activation of the reaction, such as UV activation). In some preferred embodiments, the reactants have exactly two reactive moieties, as a greater number of reactive groups can result in a lower packing density. In some preferred embodiments, the film has at least three repeating units (e.g., ABABAB), or at least five, or at least ten, or at least 50, sometimes in the range of 2 to 1000, or 5 to 100. “Reactive” means under normal MLD conditions and on a commercially appropriate timescale (e.g., at least 50% reacted within 10 hours under suitable reaction conditions). To control the quality of the film, the reactants must be single-reactive during each step of the MLD process, avoiding two reactions with the surface, and the reactants must not self-react and condense on the surface.

[0041] In some preferred embodiments, the reactive moiety of reactant A may include isocyanates (R-NCO), acrylates, carboxylic acids, esters, epoxides, amides, and amines, as well as combinations thereof. In some preferred embodiments, reactant A includes diisocyanates, diacrylates, dicarboxylic acids, diesters, diamides, or diamines. In some preferred embodiments, the reactive moiety on reactant B includes alcohols or amines, and combinations thereof. In some preferred embodiments, reactant B includes diols, amine alcohols, or diamines.

[0042] In some cases, particularly in the case of MLD, the selected gas-phase reactant reacts monofunctionally with the substrate or growing polymer chain; that is, under the reaction conditions, only one group or portion on the gas-phase reactant can react with the substrate or growing polymer chain. This prevents undesirable crosslinking or chain termination that may occur if the gas-phase reactant reacts multifunctionally. If the reactant forms a bond with only one polymer chain during the reaction and does not self-polymerize under the reaction conditions used, the reactant is considered to react "monofunctionally". In certain embodiments of the present invention, as fully described below, it is possible to use a gas-phase reactant that can react bifunctionally with the substrate or growing polymer chain, provided that the gas-phase reactant contains at least one additional functional group. In this aspect of the present invention, reactants having exactly two functional groups with substantially equal reactivity are preferably avoided.

[0043] A preferred gas-phase reactant of the first class is a compound having two different reactants, one of which is reactive with a functional group on the substrate or polymer chain, and the other which does not readily react with a functional group on the polymer chain but reacts with a functional group introduced by the other gas-phase reactant. Examples of reactants of this class are as follows:

[0044] a) Hydroxyl compounds having vinyl or allyl unsaturation. These can react with carboxylic acids, carboxylic acid halides, or siloxane groups to form esters or silicon-oxygen bonds, introducing vinyl or allyl unsaturation into polymer chains. Alternatively, the unsaturated group can react with a primary amino group in a Michael reaction to extend the polymer chain and introduce a hydroxyl group into the chain.

[0045] b) Amino alcohol compounds. The amino group can react with a carboxyl group, carboxylic acid chloride, vinyl or allyl group, or isocyanate group to, for example, extend polymer chains and introduce hydroxyl groups into the chains. Alternatively, the hydroxyl group can react with a siloxane species to form a silicon-oxygen bond and introduce free primary or secondary amino groups.

[0046] The second class of suitable gas-phase reactants includes a variety of cyclic compounds that can participate in ring-opening reactions. Ring-opening reactions generate new functional groups that do not readily react with the cyclic compound. Examples of such cyclic compounds include, for example:

[0047] a) Cyclic azasilanes. These can react with hydroxyl groups to form silicon-oxygen bonds, generating free primary or secondary amino groups.

[0048] b) Cyclic carbonates, lactones, and lactams. Carbonates can react with primary or secondary amino groups to form urethane links and generate free hydroxyl groups. Lactones and lactams can react with primary or secondary amino groups to form amide links and generate free hydroxyl groups or amino groups, respectively.

[0049] The third class of gas-phase reactants includes compounds containing two different reactive groups. Both reactive groups react with the functional groups of the polymer chain, but one is far more reactive than the other. This allows the more reactive group to react with the functional groups of the polymer chain, while the less reactive group remains unreacted and can be used to react with other gas-phase reactants.

[0050] A fourth class of gas-phase reactants includes compounds containing two reactive groups, one of which is blocked, or otherwise masked or protected, so that it cannot be used in the reaction until a blocking, masking, or protecting group is removed. The blocking or protecting group may be removed chemically in some cases, or otherwise by thermal decomposition of the blocking group to produce the underlying reactive group, by irradiating the group with visible or ultraviolet light, or by a photochemical reaction. Unprotected groups may include, for example, amino groups, anhydride groups, hydroxyl groups, carboxylic acid groups, carboxylic acid anhydride groups, carboxylic acid ester groups, and isocyanate groups. Protecting groups may, after removal of the protecting group, result in one of the aforementioned functional groups.

[0051] The reactants of this fourth class may have hydroxyl groups protected by leaving groups such as benzyl, nitrobenzyl, tetrahydropyranyl, -CH2OCH3, or similar groups. In these cases, the hydroxyl groups can be deprotected by various methods, such as treatment with HCl, ethanol, or possibly irradiation. Carboxyl groups can be protected by leaving groups such as -CH2SCH3, t-butyl, benzyl, dimethylamino, and similar groups. These groups can be deprotected by treatment with species such as trifluoroacetic acid, formic acid, methanol, or water to produce carboxylic acid groups. Amino groups can be protected by groups such as R-OOC-, which can be removed by reaction with trifluoroacetic acid, hydrazine, or ammonia. Isocyanate groups can be protected by carboxyl compounds such as formic acid or acetic acid.

[0052] The fifth class of gas-phase reactants contains a first functional group and a precursor group that can undergo further reactions to produce a second functional group. In such cases, the first functional group reacts to bond to the polymer chain, and then a chemical reaction occurs at the precursor group to produce the second functional group. The first functional group can be any of the aforementioned types, including siloxane groups, amino groups, anhydride groups, hydroxyl groups, carboxylic acid groups, carboxylic acid anhydride groups, carboxylic acid ester groups, isocyanate groups, etc. A wide variety of precursor groups can exist on this type of reactant.

[0053] Precursor groups may not react with polymer chains themselves, but can be converted into functional groups that can grow chains by reacting with other gas-phase reactants. Two notable types of precursor groups are vinyl and / or allyl unsaturated, and halogen-substituted, particularly chlorine or bromine. The vinyl and allyl unsaturated can be converted into functional groups using a variety of chemical reactions. These can react with ozone or peroxides to form carboxylic acids or aldehydes. They can also react with ammonia or primary amino acids to produce amines or imines. Halogens can be substituted with a variety of functional groups. They can react with ammonia or primary amines to introduce amino groups, and, if desired, can react with phosgene to produce isocyanate groups.

[0054] Reactants used to convert precursor groups to functional groups, or to demask or deprotect functional groups, are introduced into the gas phase. Excess reactants of this type are typically removed before introducing the next reactant, usually by sweeping a high vacuum into the reaction area, purging the chamber with a purge gas, or both. Reaction byproducts are removed in the same manner before introducing the next reactant into the reaction area.

[0055] In some preferred embodiments, at least one or all of the reactants within the MLD repeating unit have chain lengths between reactive parts of 2 to 20 atoms, or 2 to 10 atoms, or 2 to 5 atoms (heterogroups such as oxygen may be present, but typically carbon atoms). In some preferred embodiments, the reactants have straight chains between reactive parts (i.e., no branching) to increase packing density. In some preferred embodiments, the chains between reactive parts are non-reactive. However, in some embodiments, these may be parts within a chain that can bridge to adjacent chains. In some embodiments, the capping layer and / or MLD layer on or very close to the surface (e.g., the capping layer or within 5 cycles or 2 cycles of the surface) are branched to increase hydrophobicity.

[0056] The inorganic layer applied to the substrate in the first step preferably forms covalent bonds with the substrate. Covalent bonds can occur when the precursor compound initially applied reacts with functional groups on the surface of the substrate under the conditions of the atomic layer deposition process. Examples of such functional groups include, for example, hydroxyl, carbonyl, carboxylic acids, carboxylic acid anhydrides, carboxylic acid halides, and primary or secondary amino acids.

[0057] Some ALD coatings are aluminum oxide and / or titanium oxide coatings. “Aluminum oxide” is used herein to refer to a coating composed substantially entirely of aluminum and oxygen atoms, regardless of specific stoichiometry. In many cases, aluminum oxide coatings are expected to correspond somewhat closely to the empirical structure of alumina, i.e., Al2O3, but deviations from this structure are common and can occur. “Titanium oxide” is used herein to refer to a coating composed substantially entirely of titanium and oxygen atoms, regardless of specific stoichiometry. In most cases, titanium oxide coatings are expected to correspond closely to the empirical structure of titania, i.e., TiO2, but deviations from this structure are common and can occur. Similarly, considerations apply to understanding other formulations described herein, but in some embodiments, the present invention can be more specifically defined by using terms such as “consisting.”

[0058] Except for the half-reactions included in a broader embodiment of the present invention, the atomic layer deposition process is characterized by requiring at least two different reactants to form a coating layer. The reactants are introduced into the reaction region individually, sequentially, and in the gas phase. Excess reactants are removed from the reaction region before introducing the next reactant. Reaction byproducts are also removed during the sequential introduction of reagents. This procedure ensures that the reaction occurs on the surface of the substrate and not in the gas phase.

[0059] To further remove excess reactants, a purge gas is typically introduced between alternating feeds of reactants. The carrier gas is usually the same as, but not necessarily the same as, the purge gas, and is generally (but not necessarily) introduced as each reactant is introduced. The carrier gas can perform several functions, including (1) facilitating the removal of excess reactants and reaction byproducts, and (2) distributing the reactants through the reaction zone, thereby helping to expose all surfaces to the reactants. The purge gas does not cause undesirable reactions with the ALD reactants or deposited coatings, nor does it interfere with their interaction on the substrate surface.

[0060] Temperature and pressure conditions vary depending on the specific reaction system, as gaseous reactants must be supplied. As is well known in ALD / MLD processes, the temperature needs to be high enough to allow the reactants to be in the gas phase, but not so high that the product degrades.

[0061] The coating may include any coating that can be applied by depositing molecular or atomic layers. Some well-known coatings that can be applied to a core substrate of metal or other material are oxides or mixed oxides (e.g., Al2O3, TiO2, ZnO, ZrO2, SiO2, HfO2, Ta2O5, LiNb). x O y ), nitrides (e.g., TiN, TaN, W2N, TiY2N), sulfides (e.g., ZnS, CdS, SnS, WS2, MoS2, ZnIn2S4), and phosphides (e.g., GaP, InP, Fe 0.5 Co 0.5P) may be included. Some lesser-known materials that can be applied to core substrates include transition metals (e.g., those of Al, Cu, Co, W, Cr, Fe, Zn, Zr, Pt, Pd), metal fluorides (e.g., AlF3, MgF2, ZnF2), oxyfluorides and oxynitrides of transition metals, lanthanides in any form of elements, oxides, fluorides, nitrides, borides, or sulfides (e.g., Y, YN, La2O3, LaF3, Nb, Dy2O3, Nd, LaB6, La2S3, etc.), borides (e.g., TiB2), carbides ( Examples of carbon-containing materials include, but are not limited to, B4C, WC), silanes, silicides and other silicon-containing materials, polymers (e.g., polyamides, polyethylene, polyamides, polyureas, polyurethanes), hydrocarbons, polymers or fragments of amino acids or other biologically related molecules and polymers, and other materials), and fluorinated polymers (e.g., fluoro or perfluoro-polyamides, polyethylene, polyamides, polyureas, urethanes, hydrocarbons). The coating exhibits high uniformity across the substrate, preferably with a variation in coating thickness across the entire surface of the substrate of 20% or less, more preferably 10% or less, or 5% or less. This high level of uniformity is characteristic of the ALD / MLD process. [Modes for carrying out the invention]

[0062] Each ICRISP process for oxides and nitrides was first validated on Si wafers, based on an array of available measurement tools and prior art reports. Thicknesses under development were investigated using a fixed refractive index (RI) single-wavelength (633 nm) Stokesmeter-ellipsometer. Each process was first qualitatively analyzed with thick films exceeding 75 nm, with precise thickness, RI, and extinction coefficient (k) all at 633 nm. Measurements were determined using spectroscopic ellipsometry (SE) at a fixed angle (75°). These data were modeled from cos(2Ψ) and sin(2Ψ)cos(Δ) and measured from data generated using a dual light source (a combined deuterium and halogen source) in the 225–1000 nm range and a Si-based array detector. Regression analysis using the Levenberg-Marquardt method was examined employing variance methods (Cauchy, Exponential, Sellmeier, and Tauc-Lorentz), and the model was fitted over appropriate wavelength ranges from prior art reports (e.g., 250-1000 nm for Al2O3 and SiO2, and 300-800 nm for AlN and TiN). For the ALD process deposited on all samples, the growth per cycle (GPC), RI, and k are shown in Table 3 below, based on data obtained from these thick films. The degree of fit (GOF) is a term that measures the accuracy of the model to the raw data. A value greater than 0.97 is considered an accurate fit. Both TiN and AlN films have slightly lower GOFs, which may result in errors in RI and k values, due to the absorption characteristics of TiN and AlN films in the measured wavelength range. The thicknesses of both TiN and AlN films were confirmed with thinner films, but RI and k require thicker films. [Table 4]

[0063] The measured values ​​for Al2O3 films are within the expected range based on observations made by the inventors using this tool and those published in the literature; therefore, measurements from prior art Al2O3 processes are not shown above. However, as with all other films, the ultrashort cycle times for Al2O3 using the ICRISP process cannot be achieved using prior art processes. In fact, this ICRISP process for SiO2 currently runs at approximately 0.85–1.5 s / cycle. The prior art conventional ALD process for SiO2 (3DMAS / O3) struggled to bring to the tool due to the low growth pressure and difficult reaction rate of the Si-O surface bond formation process. Its GPC was lower than expected and the cycle time was very long, although the film properties were within expectations. The prior art 3DMAS / O3 process prefers to run at higher pressures than the ICRISP process with pressure and draw control. Therefore, significant improvements in cycle time cannot be expected in prior art processes that do not employ the ICRISP approach. The ICRISP TiN process at 300°C exhibited film properties within expectations, but surface oxidation (formation of TiO2) artificially increases resistivity, so capping should be performed before removal from the ALD tool. For the ICRISP TiN process, cycle times could potentially be reduced by a further 25–50% if optimized within the production tool. The prior art TiN process at 380°C also presented challenges due to high-temperature interactions between reactor components (e.g., the chuck in this invention) and reaction byproducts from the process. This effectively resulted in heterogeneity issues at the wafer edges during development. The data in Table 3 were obtained from the center point of the high-temperature TiN wafer to minimize any effects of heterogeneity. Regarding HfO2, while discussed in detail in the previous section, in short, the inventors acknowledge significant improvements in GPC, precursor doping exposure, and composition with ICRISP. Prior art processes require the use of plasma to reach these density levels or post-deposition annealing to induce crystallization.The ICRISP process for Y2O3 shows significant improvements in speed, i.e., both GPC and cycle time, as well as film composition, with increased density and reduced impurities. Furthermore, the crystallinity of the ICRISP film is not observed in the prior art. Finally, the ICRISP processes for the two AlNs demonstrate the ability of this ICRISP process to activate the ALD process, in this case either because one is not achieved, as in the TMA / MMH-H2S ICRISP process, and true ALD without CVD of TMA / MMH does not exist, or because, as in the TMA / hydrazine-H2S ICRISP process, it is a conventional improvement compared to the prior art with improved speed and film properties. In the AlN ICRISP process, GPC is increased, cycle time is shortened, and RI is higher, which is used as a substitute for film density and crystallinity.

[0064] The inventors have discovered a process for depositing ALD ZrN at low temperatures, regardless of the CVD components. Using cyclopentadienyltris(dimethylamine)zirconium (or "ZyALD," equivalent to HyALD) and monomethylhydrazine, the inventors can perform ALD of ZrN. The growth rate saturates with longer precursor addition times, i.e., at approximately 0.25 Å / cy at 225°C, and the GPC is stable with longer purging times (approximately 10 seconds). The thickness increases with the number of cycles, as expected (almost linear, but measurements are insufficient in the inventors' current setup). The inventors believe a temperature range exists between 200 and 250°C (approximately 0.2 to 0.35 Å / cy, respectively). In precursor degradation tests where only ZyALD was added to the chamber, no degradation was observed at 225°C, but significant degradation was observed at 275°C (a small amount was observed at 250°C, but the test was not conclusive). The film was deposited on a Si wafer and exhibited good nucleation and adhesion to the surface.

[0065] The Zr precursor ZyALD is used in the literature with ZrO2 along with H2O and O3, but no co-reactants are publicly disclosed for ZrN. The ALD process for ZrN exists for similar precursors, tetrakis(dimethylamine)zirconium and tetrakis(ethylmethylamine)zirconium, and ammonia (NH3). However, the inventors tested these tetrakisamine precursors and found significant decomposition at temperatures above 200°C (tests were not conducted below 200°C), contrary to published records. Supplying both tetrakisamine precursors at temperatures below 100°C resulted in decomposition in the supply line. The inventors also tested ZyALD with NH3 and found insufficient reactivity at temperatures below 275°C. Unfortunately, at 275°C, the ZyALD / NH3 process has CVD components that affect the film. This is ALD x This appears to be the first true ALD process at 225°C for ZrN in the tool. Furthermore, there is no work by nitrogen-nitrogen bonding (hydrazine or alkylated hydrazine) to the grown ZrN, which the inventors have recognized with any Zr precursor. This is the first ZrN process deposited with a hydrazine-type co-reactant.

[0066] ZrN is used as a Cu diffusion barrier, nuclear radiation barrier, and wear-resistant coating. PVD films of ZrN (5% TiN) exhibit superior Cu barrier performance compared to TiN or TaN, and greater advantages can be found by controlling the composition and layer thickness with ALD. Furthermore, ALD provides a conformal coating with uniform thickness, whereas PVD may have conformability and compositional issues. ZrN is a widely used material as a next-generation nuclear material. If the density of ZrN is sufficient using the ICRISP process, the application of an out-of-line wear-resistant coating can have multiple applications.

[0067] Synchronously regulated flow and draw (SMFD) SMFD is performed by strategic inert gas routing. The flow from the ALD space to the outside (draw) is decoupled from the flow to the ALD space. The draw depends on the gas flow conductance from the ALD space to the exhaust space and the pressure difference (DP) between these spaces. SMFD allows the system to deposit the ALD cycle at an extremely high rate while maintaining highly efficient use of the precursor. Figure 2 shows the precursor gas (downward arrow in the center) and the outer arrows (Figure 3) show the purge gas. Both the precursor and purge gases flow during the purge step.

[0068] The delay step releases the desired precursor into the showerhead and ALD space. The draw control gas is off, making the precursor the only chemical entering the ALD space. This is the least efficient use of the precursor, but it promotes the initial saturation of the substrate surface. The typical processing time for this step is 0-10 milliseconds.

[0069] The pulse step follows the delay step. In this step, both the precursor and the draw control flow simultaneously. This creates a pressure difference between the draw control space and the ALD space. The precursor exits the ALD space at a different rate than the draw control space, creating a longer residence time for the precursor. As the pressures between the draw control space and the ALD space begin to equalize, the pulse efficiency decreases as a function of time. However, the pulse step is the most efficient use of the precursor due to the prolonged saturation of the ALD space. The pulse step is typically 5–50 milliseconds.

[0070] The addition step follows the pulse step. The addition step uses only draw control but utilizes the precursor trapped in the ALD space. Draw control increases the total residence time of any precursor still present in the ALD space. This step acts to increase overall precursor utilization, inducing turbulence to move the precursor and remove reaction byproducts. The addition step has a duration ranging from 10 to 500 milliseconds and can be gradually optimized as new chemicals / processes are developed. Longer addition times, exceeding 150 milliseconds, tend to have a reduced impact on precursor saturation. During the addition step, no precursor flows into the system; typically, a purge (inert) gas continues to flow throughout the addition step.

[0071] The purging step follows the addition step and is intended to remove all possible precursor vapors before the next half-cycle. An inert gas (N2) purge is performed with all draw controls and precursor valves closed. The N2 flow also purges the entire precursor valve stack for complete exhaust. This cycle typically takes 250–5000 milliseconds for simple processes and up to 30000 milliseconds for difficult chemicals.

[0072] A further purge step, called a pump purge step, can also be performed without inert flow. This is less efficient than an inert-driven purge, but it lowers the chamber's base pressure as needed.

[0073] High-speed pneumatic valve (FPV) Forge Nano manufactures fast-acting pneumatic valves for SMFD processes. This FPV assembly consists of a normally closed diaphragm valve paired with a normally open solenoid valve. The solenoid valve regulates the operation of the pneumatic valve. These FPVs are available in two-way and three-way configurations, depending on their role in the valve stack.

[0074] Figure 4 shows a flow control valve 100 (represented as 100a or 100b) having a diaphragm 101 positioned to operate to seal the flow path between the valve seat 103 and the fluid outlet port 105. An orifice 104 is formed between the diaphragm 101 and the valve seat 103. A sealed diaphragm control space 102 is formed above the diaphragm 101. The diaphragm control space includes a control fluid inlet 123 and a control fluid outlet 133. The flow control valve 100 also includes an inlet fitting 110, an outlet fitting 111, a control fluid inlet 126, a solenoid valve 120, a solenoid valve 130, a controller 108, and an enclosure 109. Fluid (e.g., compressed air) from the inlet 126 is supplied into the solenoid valve 120. In order for the flow control valve 100 to actuate for the first time, the inventors determined that the diaphragm control space must be at least 80 psi pressure threshold. By utilizing the small pressure fluctuations from this threshold pressure that keeps the diaphragm closed to the precursor flow, the fluid can be transported very rapidly into and out of this diaphragm-controlled space, enabling millisecond control over processing steps.

[0075] During operation, when the solenoid valve 120 is activated, the plunger 121 opens the passage through the inlet 123, increasing the pressure inside the diaphragm control space 102, causing the diaphragm 101 to flex toward the valve seat 103 and reducing the orifice 104. This can reduce or completely stop the flow rate. Conversely, when the solenoid valve 130 is activated, the normally closed plunger 131 opens the passage through the outlet 133, reducing the pressure inside the diaphragm control space 102 by allowing the control fluid to flow out through the vent port 135. In response to the pressure drop, the diaphragm 101 flexes toward the valve seat 103, widening the orifice 104 and resuming or further increasing the required flow rate. Each flow control valve 100 representing an individual FPV includes a shut-off valve 150 that provides a fail-safe shut-off of the diaphragm 101 via the stem 152 in the event, for example, that the pressure of the control fluid is lost due to a system failure. The stem 152 can slide into the control space 102 through the opening 157. At the same time, the seal 158 maintains the overall fluid airtightness of the control space 102. The stem 152 is held in place by the shut-off valve 150 by the force of a spring 153 adapted to bias the stem 152 toward the diaphragm 101. The shut-off valve 150 opens when pressurized control fluid (e.g., compressed air) is injected into the actuator space 159, causing the piston 151 to move relative to the spring 153 and away from the diaphragm 101. The piston 151 is fitted with a slide seal 154. When the shut-off valve 150 is actuated to open, the control fluid from the actuator space 159 is supplied into the solenoid valve 120 through the inlet 126, where it becomes available to adjust the position of the diaphragm 101. Preferably, during use, the retraction of the stem 152 is set by the adjustment screw 155 to the minimum value required by the plunger 156 to allow the diaphragm 101 to move through its full range of motion. This limited movement is desirable to minimize the acceleration of the stem 152 in the event of a loss of pressure in the control fluid, and the resulting effects when the stem 152 moves beyond the valve seat 103 into the diaphragm 101.This effect is further mitigated by restricting the discharge of air from the actuator space 159 when the air supply at inlet 112 is stopped. In actual implementation, both measures have proven to be very effective in preventing particle generation by the fail-safe shut-off valve 150.

[0076] Figure 5 shows cross-sectional views of two FPVs used in the present invention for two different precursors. FPV-1 is in the closed state during operation, and FPV-2 is in the open state during operation. Gas supply is achieved through a stack of FPVs. Each FPV is stacked in series with the others such that each outlet 111 of each FPV is in fluid communication with the main gas supply line that supplies to the ALD showerhead. Because the precursor steps required in the ICRISP process are rapid in nature, multiple gas exposure sequences must be incorporated into the gas supply line before the precursor reaches the substrate surface. Therefore, this control sequence, and in particular the communication between the solenoid valves 120 and 130 of FPV-1 and the solenoid valves 120 and 130 of FPV-2, must be fully fluid and robust.

[0077] Metal Nitride MN of the present invention x During operation of the FPV stack using the ICRISP process, FPV-1 uses a metal-containing precursor MR. * x FPV-2 is used to supply metal-free nitrogen-containing precursors, such as MMH; FPV-3 is used to supply metal-free sulfur-containing co-catalysts, such as H2S; and FPV-4 is used to supply inert gases, such as nitrogen or argon.

[0078] Step 1: Activate solenoid valve 120 of FPV-1, then activate solenoid valve 130 of FPV-1 after time t1, thereby MR at time t1. * x Provides a precursor pulse, Step 2: Perform a purge and / or flash reaction step for time t2 by activating the solenoid valve 120 of the FPV-4, and then activate the solenoid valve 130 of the FPV-4 after time t2. Step 3a: By activating the solenoid valve 120 of FPV-2, a metal-free N-containing precursor such as the MMH subpulse 1 is provided for a time t3. Step 3b: Simultaneously activate solenoid valve 130 of FPV-2 and solenoid valve 120 of FPV-3 to intermittently suspend the supply of MMH, and start supplying a metal-free sulfur-containing co-catalyst, such as H2S, at time t4. Step 3c: Simultaneously activate solenoid valve 130 of FPV-3 and solenoid valve 120 of FPV-2 to intermittently resume the supply of MMH, and temporarily suspend the supply of a metal-free sulfur-containing co-catalyst, such as H2S, for a time t5. Step 4: MR deposited in Step 1 * Steps 3b and 3c are repeated as many times as necessary to convert at least 99% of the surface area into an MxNy metal nitride layer. Step 5: Perform the purging and / or flush reaction step for a time of t6. Here, t3 and t5 are greater than t4.

Claims

1. An intermittent catalytic reaction-induced surface process (ICRISP) for forming an oxide or nitride film on a substrate, To provide a substrate inside the chamber, A metal-containing ALD precursor is added to the aforementioned substrate to prepare a first substrate with the additive. Purge or flush the chamber, and A) Adding an oxidizing agent and a nitrogen-containing reactant to the first added substrate to form a metal oxide, or B) Adding a sulfur-containing reactant and a nitrogen-containing reactant to the first added substrate to form a metal nitride, or the other of the above, The process includes purging or flushing the chamber.

2. A process according to claim 1 for producing an oxide film, comprising adding an oxidizing agent and a nitrogen-containing reactant to the first added substrate to form a metal oxide, The process wherein the adding step comprises a subpulse with the oxidizing agent without the nitrogen-containing reactant, a second subpulse with the oxidizing agent and the nitrogen-containing reactant, and a third subpulse with the oxidizing agent without the nitrogen-containing reactant.

3. The process according to claim 2, wherein the oxidizing agent comprises ozone and the nitrogen-containing reactant comprises hydrazine.

4. The step of adding an oxidizing agent to the first added substrate is further H ２ The process according to claim 2, comprising the addition of S.

5. A process according to claim 1 for producing a nitride film, comprising adding a sulfur-containing reactant and a nitrogen-containing reactant to the first added substrate to form a metal nitride, The process wherein the adding step comprises a subpulse of the nitrogen-containing reactant, a second subpulse of the sulfur-containing reactant and the nitrogen-containing reactant, and a third subpulse of the nitrogen-containing reactant.

6. The oxide film is Al ２ O ３ , SiO ２ , TiO ２ , Nb ２ O ５ , Ta ２ O ５ , La ２ O ５ , Y ２ O ３ , ZrO ２ , Ga ２ O ３ , or In ２ O ３ The process according to claim 2, comprising

7. The nitride film is made of ZrN, GaN, YN, InN, or Si ３ N ４ The process according to claim 5, including the process described in claim 5.

8. The sulfur-containing reactant is H ２ S (hydrogen sulfide), H ２ S ２ (Dihydrogen sulfide), mercaptan (HS(CH) ３ ) or methanethiol), ethanethiol (ethyl mercaptan), S (CH ３ ) ２ (Dimethyl sulfide), thionyl chloride (SOCl ２ ), sulfuryl chloride (SO ２ Cl ２ The process according to claim 5, including )

9. The aforementioned substrate is Si, SiO ２ Al ２ O ３ A process according to any of the prior claims, comprising SiC, GaN, AlGaN, GaAs, or InP (preferably in the form of corundum).

10. The oxidizing agent is O ２ , O ３ H ２ O, H ２ O ２ , or N ２ A process according to any of the prior claims, comprising O. The category described by comprising may include additional components that can be mixed with the oxidizing agent.

11. The process according to any of the prior claims, wherein the nitrogen-containing reactant comprises hydrazine, monomethylhydrazine (MMH), 1,1-dimethylhydrazine, 1,2-dimethylhydrazine, tert-butylhydrazine (tBuNNH), ammonia (NH3), pyridine, 2,3-lutidine (2,3-dimethylpyridine), 2,4-lutidine (2,4-dimethylpyridine), 2,5-lutidine (2,5-dimethylpyridine), 2,6-lutidine (2,6-dimethylpyridine), 3,4-lutidine (3,4-dimethylpyridine), or 3,5-lutidine (3,5-dimethylpyridine).

12. A process according to any of the prior claims, which is performed for 10 to 5,000 cycles, or 10 to 1,000 cycles, or 20 to 200 cycles.

13. The ALD precursor is o Ti-TiCl4, TTIP (Titanium (IV) Isopropoxide), Ti(NETMe)4 (Titanium (IV) Tetrakisethylmethylamine), Ti(NMe)4 (Titanium (IV) Tetrakisdimethylamine) o Hf-Hf(NETMe)4(hafnium(IV)tetrakisethylmethylamine), HyALD(hafnium(IV)cyclopentadienyl-trisdimethylamine) o Zr-Zr(NETMe)4 (Zirconium (IV) tetrakisethylmethylamine), ZyALD (Zirconium (IV) cyclopentadienyl-trisdimethylamine) o Si-BEMAS (bis-ethylmethylaminosilane), BDEAS (bis-diethylaminosilane), 3DMAS (trisdimethylaminosilane) o Y-ArYA (Yttrium (III) bisethylcyclopentadienyl-isopropylamidinate) o Al-TMA (trimethylaluminum), AlCl ３ , AlMeCl2, and AlMe ２ Cl A process according to any of the prior claims, including the process described in any of the prior claims.

14. The step of adding a metal-containing ALD precursor to the substrate includes a step of introducing the precursor into the chamber, and thereafter a pulse step of introducing the precursor and draw control are performed simultaneously, the draw control includes the flow of an inert gas and the application of vacuum, and the pulse step is performed for 5 to 50 ms. The process according to claim 1, wherein a step of 10 to 500 ms is then performed in which no precursor flow is present in the system.

15. The process according to claim 14, wherein the adding step is performed for 150 ms or less.

16. The process according to claim 13, wherein an inert gas is flowed into the chamber for at least 200 ms, in which case preferably no precursor is introduced into the chamber and no vacuum is applied to the chamber.

17. The process according to claim 2, wherein the subpulse by the oxidizer without the nitrogen-containing reactant is performed for a longer duration than the subpulse by the oxidizer and the nitrogen-containing reactant (the subpulses are totaled for this calculation).

18. The process according to claim 17, wherein the subpulse by the oxidizer without the nitrogen-containing reactant is performed for at least twice as long as the subpulse by the oxidizer and the nitrogen-containing reactant.

19. The process according to claim 5, wherein the subpulse by the oxidizer without the sulfur-containing reactant is performed for a longer duration than the subpulse by the oxidizer and the sulfur-containing reactant (the subpulses are totaled for this calculation).

20. The process according to claim 17, wherein the subpulse by the oxidizer without the sulfur-containing reactant is performed for at least twice as long as the subpulse by the oxidizer and the sulfur-containing reactant.

21. An intermittent catalytic reaction-induced surface process (ICRISP) for producing a film having a thickness T and composition MxOyNz from a metal-containing precursor, an oxygen-containing precursor, and a nitrogen-containing precursor at a temperature of 300°C or less, wherein x is 1 or 2, y is 0 to 4, and z is 0 to 2. a) Adding the metal-containing precursor with a pulse time of 5 to 500 ms, b) A step of purging and / or flushing the metal-containing precursor for a time of 250 to 30,000 ms, c) If y is greater than 0, i. Adding the oxygen-containing precursor for a time t3, ii. Adding the nitrogen-containing precursor incorporated into the oxygen-containing precursor for a time t4, provided that t4 < t3. iii. Adding the oxygen-containing precursor for a time t5, provided that t5 > t4. iv. Optionally, repeat steps c. ii and c. iii a number of times necessary to convert at least 99% of the metal-containing precursor deposited in step a) into the MxOy layer. d) If z is greater than 0, i. Step of adding the nitrogen-containing precursor for a time t6. ii. Adding the oxygen-containing precursor incorporated into the nitrogen-containing precursor for a time t7, provided that t7 < t6. iii. Adding the nitrogen-containing precursor for a time t8, provided that t8 > t7. iv. Optionally, repeat steps d) ii. and d) iii. a number of times necessary to convert at least 99% of the metal-containing precursor deposited in step a. into the MxNz layer, and The process comprising the step of repeating steps a to d until a desired thickness T is obtained.

Images