Atomic layer etching with single process gas and rapid adjustment of chamber pressure

Abstract

The invention discloses atomic layer etching adopting single process gas and rapidly adjusting chamber pressure. The process includes a step time designed for optimizing the surface modification and sputtering process, thereby minimizing undesirable effects. The customized waveform generator enables rapid and accurate control of the substrate bias, thereby shortening the ALE cycle time. Besides, by presetting set values for the MFC and the vacuum valve, rapid adjustment of the pressure of the cavity is achieved, traditional PID control is not needed, and therefore the efficiency and precision in semiconductor manufacturing are further improved.

Description

Cross-reference to related applications

[0001] This invention claims priority to U.S. Patent Application No. 18 / 885,566, filed on September 14, 2024. Technical Field

[0002] This invention relates to the field of semiconductor manufacturing, and more specifically, to atomic layer etching (ALE) technology for the fabrication of semiconductor devices. The key feature of this invention is the use of a single gas or a single gas mixture throughout the ALE process, combined with rapid chamber pressure regulation technology to optimize the ALE process. Background Technology

[0003] Reactive ion etching (RIE) is a key technology in semiconductor manufacturing. In RIE, various substances, including neutral particles, free radicals, and ions, collectively influence the etching process. A crucial characteristic of RIE is the synergistic effect between ion flux and neutral particle flux, which significantly enhances the etching rate. Coburn and Winters first described this synergistic effect in their 1979 article, "Ion-and-electron-assisted gas-surface chemistry—an important effect in plasma etching," published in the *J. Appl. Phys.*, Vol. 50, pp. 3189-3196. They reported that the etching rate of silicon increased when using argon ion beams, XeF2 neutral beams, and combinations of both. Furthermore, Gottscho et al. proposed a model in "Microscopic uniformity in plasma etching" (J. Vac. Sci. Technol. B10, pp. 2133-2147, 1992) to quantify the effect of this synergistic effect on the etching rate (ER):

[0004] Where v represents the volume (cm³) removed from the saturated surface by a unit bombardment energy. 3 / eV), E i For ion energy (eV), J i The ion flux reaching the surface (cm) 2 / s), v n Volume removed (cm³) for each neutral particle in the reaction 3 ), J n Let s be the flux of neutral particles reaching the surface, and s be the adhesion probability of neutral particles on the bare surface.

[0005] To achieve efficient RIE, the coexistence of ion flux and neutral particle flux is necessary to leverage the synergistic effect discovered by Coburn and Winters. However, achieving a balance between these two factors in modern etching equipment has become increasingly complex, especially as etching dimensions tend towards nanoscale high aspect ratio (HAR) structures. Achieving uniform etching on 300mm wafers and ensuring consistent repeatability in mass production also presents additional challenges.

[0006] Over the past few decades, etching equipment has undergone numerous improvements to enhance etching uniformity. For example, plasma sources have evolved from single-coil structures (see U.S. Patent No. 4,948,458, Ogle) to multi-coil structures (see U.S. Patent No. 6,164,241, Chen et al.). Plasma sources can employ inductively coupled plasma (ICP) or transformer-coupled plasma (TCP) structures. Furthermore, gas injection techniques have been improved, employing multiple injection ports to ensure plasma uniformity within the vacuum chamber, as described in U.S. Patent Nos. 8,231,799 (Bera et al.) and 10,825,659 (Treadwell). Further improvements include optimization of the electrostatic chuck (ESC) to provide multiple independently temperature-controlled zones (see U.S. Patent Nos. 9,713,200, Pease; 10,056,225, Gaff et al.).

[0007] A radio frequency (RF) power generator coupled to the ESC provides an additional bias voltage to the ions in addition to the plasma sheath. This coupling is typically achieved via a DC blocking capacitor to prevent electrons from flowing to ground, thereby helping to establish a stable plasma sheath (see U.S. Patent No. 5,302,240, Hori et al.). Furthermore, to improve ion energy and angular momentum distribution, various pulsed schemes of the RF power generator are implemented to maximize the synergistic effect between ions and neutral particles, as described in U.S. Patent Nos. 8,264,154 (Banner et al.) and 10,121,639 (Kanarik). Wang et al., in “Experimental demonstration of multifrequency impedance matching for tailored voltage waveform plasmas” (J.Vac.Sci.Technol.A37,021303, pp. 1-11, 2019), also proposed a controllable waveform RF power generator for precise control of ion energy.

[0008] Furthermore, to improve performance, gas can be supplied via a cyclic pulse method, as described in U.S. Patent No. 10,121,639 (Kanarik). This method divides the RIE process into multiple steps, each optimized using a different process gas.

[0009] Despite these improvements, achieving high consistency in critical dimensions (CD), loading effects, and profiles on 300mm wafers remains a significant challenge and often requires substantial costs.

[0010] To overcome the limitations of RIE, ALE technology was developed. ALE equipment evolved from RIE equipment, but it has lower requirements for achieving uniformity on 300mm wafers. However, due to the characteristics of its process steps, ALE has its own unique requirements.

[0011] Karanik et al. introduced ALE technology in "Overview of atomic layer etching in the semiconductor industry" (J.Vac.Sci.Technol.A33, pp. 020802 1-14, 2015), and Lill further explored this technology in her book "Atomic Layer Processing: Semiconductor Dry Etching Technology" (Wiley-VCH GmbH, 2021). Anisotropic ALE enables the removal of material layers with atomic precision and is an etching technique utilizing sequential self-limiting reactions. The basic ALE process consists of two steps: a surface modification step and a sputtering step for removing the modified layer. The surface modification step is used to form a thin reactive layer of defined thickness, which is easier to remove than the unmodified material. Subsequently, the reactive layer is removed by the sputtering step, while retaining the underlying substrate, thus resetting the surface for the next cycle. The amount of material removed is quantified over multiple cycles, which can be achieved by thermal energy obtained from heating the wafer or by kinetic energy provided by ions from an inert gas. The isotropic process uses thermal energy to remove the modified layer, and the relevant technology is described in U.S. Patent No. 10,208,383 (George et al.); while when high-energy ions are used, the removal process is carried out by sputtering.

[0012] The anisotropic ALE process, which this invention focuses on, has been used to etch a variety of materials, demonstrating its broad applicability, for example:

[0013] Si and Ge: as described in U.S. Patent No. 10,727,073 (Tan et al.);

[0014] SiO2: as described in U.S. Patent No. 9,620,382 (Oehrlein et al.);

[0015] C: As described in U.S. Patent Publications No. 2017 / 0316935 and No. 2022 / 0216050 (Tan et al.);

[0016] Tungsten (W): as described in U.S. Patent Publication No. 2020 / 0286743 (Lai et al.) and No. 10,096,487 (Yang et al.);

[0017] Cobalt (Co): as described in U.S. Patent No. 10,096,487 (Yang et al.);

[0018] Ruthenium (Ru): as described in U.S. Patent Publication No. 2022 / 0199422 (Yang et al.);

[0019] Other refractory metals and materials with high surface bonding energy, such as those described in U.S. Patent No. 11,450,513 (Yang et al.);

[0020] Copper (Cu): As described in WIPO International Publication No. 2022 / 046429 (Yang et al.);

[0021] GaN and other group III-V materials: as described in U.S. Patent No. 10,056,264 (Yang et al.);

[0022] MRAM: as described in U.S. Patent No. 10,749,103 (Tan et al.);

[0023] EUV graphical representation: as described in U.S. Patent No. 9,922,839 (Wise et al.);

[0024] Surface smoothing of various materials: as described in U.S. Patent No. 10,304,659 (Karanik et al.).

[0025] Furthermore, deposition steps are often introduced during ALE, either between ALE steps or as independent steps, to improve performance, particularly in controlling the contours of the structure to be etched. For example, selective deposition on carbon-containing materials can improve ALE performance (see U.S. Patent Publication No. 2017 / 0316935, Tan et al.). As another example, in the method disclosed in U.S. Patent No. 9,805,941 (Karanik et al.), atomic layer deposition (ALD) is performed alternately with ALE in the same plasma chamber to prevent performance degradation during etching, improve selectivity, and coat sensitive layers in the wafer.

[0026] The different chemical reactions, particle morphologies, and plasma energy compositions involved in surface modification and sputtering steps facilitate more precise control of the flux of ions, electrons, and neutral particles, thereby expanding the process window. This step separation mechanism contributes to the realization of self-limiting reactions, which is crucial for maintaining an ideal etching process, namely ensuring uniformity, smoothness, and selectivity. Karanik et al., in "Predicting synergyin atomiclayer etching" (J.Vac.Sci.Technol.A35, pp. 05C302 1-7, 2017), defined the ALE synergy factor as follows:

[0027] Here, EPC (Etch Per Cycle) represents the total thickness of material removed in one cycle (usually the average of multiple cycles); the values ​​of "α" and "β" are the (non-ideal) contributions from the surface modification step and the sputtering step, respectively. Ideally, these two steps would not etch independently, but when combined, the synergy factor would approach 100%. However, in reality, due to the presence of ions in the plasma, which generate neutral particles to modify the surface, the RIE effect in the surface modification step is unavoidable; similarly, physical sputtering of the underlying unmodified layer in the sputtering step is also unavoidable.

[0028] In the surface modification step of ALE (Alternating Electrode Process), it is desirable to avoid ion bombardment by the plasma. However, the unintended introduction of RIE (Residual Injection Effusion) in this step presents a challenge. Since ion bombardment of the substrate surface is difficult to completely avoid, conventional ALE struggles to maintain ideal etching characteristics. Traditional ALE methods are ineffective at eliminating these RIEs, resulting in suboptimal etching performance, especially as device structures become increasingly complex and their dimensions shrink. RIEs in the ALE process cause non-uniform layer removal and undesirable etching profiles, which is particularly severe in advanced device manufacturing; even small deviations can significantly impact device performance and yield.

[0029] To address this problem, U.S. Patent No. 9,362,131 (Agarwal et al.) proposed a solution: using an electron beam source, in the passivation step (i.e., the surface modification step), a remote plasma source is used to supply passivating material to the main chamber, and the ion energy is controlled below the etching threshold; while in the etching step, the material flow from the remote plasma source is stopped and the ion energy is increased above the etching threshold. This solution requires the introduction of an additional remote source, increasing system complexity and cost.

[0030] In addition to enhancing the ALE synergistic effect, it is also necessary to increase the ALE rate. For example, US Patent No. 10,763,083 (Yang et al.) uses high-energy ions (greater than 150 eV) to remove the modified layer. Another approach is to provide a bias voltage for the ions in the plasma by applying pulses to an RF power generator (which is coupled to the ESC). WIPO International Publication No. 2023 / 183129 discloses a method to reduce the etching time of the sputtering step to between 10 ms and 600 ms.

[0031] Another approach to improve ALE speed is to relax the self-limiting reaction conditions (see Taiwan Patent No. 1757334, Cottle et al.). Although this speeds up the process, it may sacrifice synergy, thereby reducing ALE performance.

[0032] Standard ALE processes typically require dynamic regulation of multiple gas flow rates using mass flow controllers (MFCs). This necessitates frequent adjustments to gas flow rates to maintain optimal etching conditions, increasing process complexity and the likelihood of inconsistencies. Switching between etching stages, such as from a surface modification step to a sputtering step, often requires a complete gas replacement and adjustment of different chamber pressures, further extending cycle time. For example, sputtering steps are typically performed at lower pressures.

[0033] Furthermore, continuously adjusting gas flow and valve positions during production reduces efficiency and leads to fluctuations in process results. These challenges limit the throughput and scalability of traditional ALE processes, becoming a bottleneck in large-scale semiconductor manufacturing.

[0034] This invention addresses the aforementioned limitations by introducing an innovative ALE method that uses a single gas or pre-mixed gas composition throughout the process, thereby simplifying gas management and significantly reducing cycle time. Furthermore, this invention provides a custom waveform generator for precisely controlling the substrate bias voltage, thereby optimizing the sputtering step in the ALE process and minimizing undesirable surface modifications caused by neutral particles. This invention also introduces a novel method for rapidly adjusting chamber pressure during step switching, enabling efficient switching between surface modification and sputtering steps. This approach improves the overall efficiency, accuracy, and consistency of ALE, representing a significant advancement in semiconductor manufacturing. Summary of the Invention

[0035] This invention proposes a system and method for application in the field of ALE (Alternating Electrochemical Processes), utilizing a single gas or a mixture of gases throughout the process. In some embodiments, the method simplifies gas management, thereby reducing cycle time and improving overall process efficiency.

[0036] In some implementations, the present invention further provides a novel method for rapid chamber pressure regulation between surface modification and sputtering steps. During setup or training, the system controller identifies setpoints for the MFC and vacuum valves to achieve steady-state chamber pressure for each step. In some embodiments, the system controller activates the MFC and PID control of the process system during this training phase. Once the setpoints are determined, the PID control is deactivated, and the process system subsequently retrieves and applies these setpoints, thereby achieving steady-state chamber pressure more quickly and significantly improving the speed at which stable pressure is achieved during step transitions.

[0037] Furthermore, in some embodiments, the present invention provides a custom waveform generator for precise control of the substrate bias voltage during the sputtering step. This allows for rapid ion acceleration, thereby effectively shortening the sputtering step and minimizing undesirable surface modifications caused by neutral particles. In some implementations, this custom waveform generator allows for rapid adjustment of the substrate bias voltage, thereby improving etching accuracy and process efficiency.

[0038] In several embodiments, the present invention also optimizes the operating parameters of the chamber, such as the RF power of the plasma source and the bias voltage setting, to ensure its coordinated operation with the single gas composition and rapid pressure regulation. These adjustments ensure that the ALE process is performed under optimal conditions, further improving overall process efficiency.

[0039] Although this invention illustrates its innovative concept through ALE processes using a single process gas or a mixed gas, the method for rapidly establishing steady-state chamber pressure is universally applicable and can be applied to a variety of other semiconductor manufacturing processes. These processes include, but are not limited to, RIE, plasma-enhanced chemical vapor deposition (PECVD), ALD, hot-based etching, and radical-based etching or deposition. By utilizing stored setpoints to rapidly and accurately regulate chamber pressure, this method is widely applicable across various process systems, providing higher efficiency and reliability.

[0040] In summary, this invention provides a method for improving ALE or other process systems by using a single process gas or a mixture of gases, a novel method for rapidly regulating chamber pressure, and a custom waveform generator for shortening sputtering steps. In various embodiments and implementations, these technical solutions significantly improve the accuracy, consistency, and overall efficiency of the ALE process. Attached Figure Description

[0041] To clearly describe the technical solution, the following explanation will refer to the accompanying drawings:

[0042] Figure 1A An exemplary process system for ALE process is shown, highlighting the key components in the system and their layout.

[0043] Figure 1B This is a schematic diagram of a valve that has a movable part for adjusting the gas conductivity through the pump to achieve precise control of the chamber pressure.

[0044] Figure 1C The following is an exemplary functional block diagram of MFC, which illustrates in detail how it regulates gas flow in the ALE process.

[0045] Figure 2 Various waveforms related to the ALE process are shown, including plasma source waveforms, substrate bias voltage waveforms, output waveforms of a custom waveform generator, MFC electromagnetic coil current waveforms, and chamber pressure waveforms.

[0046] Figure 3A A flowchart is shown illustrating the process of determining MFC and vacuum valve setpoints during the setup or training phase.

[0047] Figure 3B A flowchart illustrating the execution of ALE process steps according to an embodiment of the present invention is shown, including the switching between surface modification steps and sputtering steps.

[0048] Figure 4 A flowchart illustrating the steps for rapidly establishing chamber pressure is shown. Detailed Implementation

[0049] To ensure a full understanding, detailed embodiments of the invention are described herein. While certain specific details are provided for clarity, any modifications and variations consistent with the technical principles of the invention are considered appropriate. Conventional methods and components are also described to highlight the unique features of the invention.

[0050] The terms used in this invention are defined as follows:

[0051] Anisotropic etching (ALE) is an etching process used in semiconductor manufacturing that removes material layer by layer at the atomic scale, allowing for precise control over etching depth and morphology. ALE operates in a cyclic manner and typically includes a surface modification step (chemically modifying the surface) and a sputtering step (removing the modified surface layer through physical ion bombardment). This method offers high precision and selectivity.

[0052] Aspect Ratio: The ratio of the height to the width of a structure on a semiconductor wafer, which is crucial for defining the geometry and performance of microstructures.

[0053] Bias Unit: In some embodiments, this unit generates a controlled voltage to accelerate ion bombardment of the wafer carried by the electrostatic chuck (ESC). The electric field generated by the bias voltage enhances the ion bombardment, ensuring precise control over the energy and directionality of the ions during etching.

[0054] Chamber: A closed environment used to perform semiconductor manufacturing processes such as etching or deposition.

[0055] Chuck: A component used to carry and hold the position of a wafer in semiconductor manufacturing processes.

[0056] Electrostatic Chuck (ESC): A type of chuck that uses electrostatic force to fix the position of a wafer in a semiconductor manufacturing process, providing uniform clamping force and stability.

[0057] Gas Distribution Unit: In some embodiments, this unit is used to introduce and distribute process gases onto the substrate surface. For example, the injectors can be placed in a central position or arranged at a specific angle, or a shower head structure can be used to achieve uniform gas dispersion. Furthermore, a lateral injection mechanism can be used to promote lateral gas flow to improve distribution uniformity.

[0058] Gas Source: Refers to the origin or supply point of process gas in a vacuum process chamber. In some embodiments, the gas box regulates and controls the gas flow rate and delivers gas under controlled pressure and conditions.

[0059] High aspect ratio: refers to a structure on a semiconductor wafer whose height is much greater than its width. It is difficult to manufacture because it is hard to maintain uniformity.

[0060] PID control: A proportional-integral-derivative (PID) control loop mechanism used to regulate variables such as temperature, pressure, and gas flow rate. In some embodiments, PID control ensures the stability and accuracy of process conditions during semiconductor manufacturing.

[0061] Plasma Process Chamber: A vacuum chamber designed specifically for processes involving plasma, such as etching or deposition, where plasma provides the energy needed to activate chemical reactions or remove materials.

[0062] Plasma Source: In some embodiments, a plasma source is used to generate plasma for etching, deposition, or surface modification. As examples, plasmas include inductively coupled plasma (ICP), transformer-coupled plasma (TCP), and capacitively coupled plasma (CCP).

[0063] Process System: refers to the integrated equipment used in semiconductor manufacturing to perform various processes (such as deposition and etching).

[0064] Pulsing: A technique that modulates RF power in the form of pulses (discontinuous wave form) to better control plasma energy and improve process effects such as accuracy and uniformity.

[0065] Pulse Train: A series of precisely controlled pulses generated by a custom waveform generator that form a specific waveform profile.

[0066] Reactive ion etching (RIE) is a plasma-based etching technique that combines physical ion bombardment and chemical reactions to synergistically remove substrate material. RIE provides highly selective and anisotropic etching, which is crucial for complex microfabrication.

[0067] A resonator is a device or circuit component that generates resonance under specific RF conditions, used in applications such as RF impedance matching.

[0068] RF Power Generator: A device that generates RF power to excite plasma in processes such as etching or deposition.

[0069] Sheath: The boundary layer between the plasma and the substrate surface, where an electric field is formed to control the flow of ions and electrons.

[0070] Substrate: The basic material upon which semiconductor devices are manufactured, usually a silicon wafer.

[0071] System Controller: The central unit that manages and controls the operation of the process system, ensuring the coordinated and efficient operation of the semiconductor manufacturing process.

[0072] Tailored Waveform Generator: In some embodiments, this generator produces a custom electrical waveform to optimize the plasma process. Custom waveforms improve control over plasma characteristics and enhance etching and deposition results.

[0073] Transmission line: In RF technology, a conductor used to transmit RF signals with minimal loss and distortion. Transmission lines ensure efficient power transfer in semiconductor manufacturing processes such as etching or deposition.

[0074] Vacuum Chamber: A closed space that removes air and other gases to create a low-pressure environment, which is crucial for precision semiconductor manufacturing processes.

[0075] Window: In a vacuum chamber, this window isolates the plasma generation area from external components. It is typically made of a non-conductive material, allowing electromagnetic waves such as RF or microwave energy to pass through.

[0076] Figure 1A An ALE process system 100 is illustrated as an example. The system includes a plasma process chamber 102, the operation of which is coordinated by a system controller 140. The chamber 102 is enclosed by a cavity structure 104, thereby creating a vacuum environment suitable for plasma processing. The cavity structure 104 may be made of materials such as aluminum or quartz, and its aluminum inner surface may be treated by processes such as anodizing, or coated with yttrium oxide to enhance its resistance to the plasma environment.

[0077] Located above the cavity structure 104 is the plasma source 106. Below the plasma source 106 (not shown in the figure) is a window that seals the cavity 102. This window can be made of materials such as quartz or ceramic, and its inner surface can be coated with a plasma-resistant material such as yttrium oxide. The plasma source 106 can take various forms, such as an inductively coupled plasma (ICP) or transformer-coupled plasma (TCP) source, and can include multi-turn coils or coil groups, which can be cylindrical or conical.

[0078] Plasma source 106 is functionally connected to RF power generator 108 via resonator 110. RF power generator 108 can generate single-frequency or multi-frequency RF power, with frequencies including but not limited to 100kHz, 200kHz, 400kHz, 2MHz, 13.56MHz, 27MHz, 40MHz, and 60MHz. Resonator 110 plays a crucial role in matching the output impedance of RF power generator 108 to the plasma load of chamber 102, while taking into account transmission line effects.

[0079] Gas distribution unit 112 is connected to gas source 114 via MFC 116. Gas source 114 supplies gas 118 to gas distribution unit 112. A valve 120 is provided between MFC 116 and gas distribution unit 112. In some embodiments, gas 118 may be a single substance, such as chlorine for silicon etching, or a mixture, such as a mixture of chlorine and argon, or a mixture of oxygen and argon. A significant feature of the present invention is the use of a single gas or a single gas mixture throughout the ALE process, thereby significantly reducing cycle time by avoiding gas switching. However, the flow rate of the gas or gas mixture can be adjusted using rapid modulation techniques to obtain appropriate chamber pressures in the surface modification and sputtering steps of the ALE process, respectively.

[0080] The flow rate of gas 118 is controlled by MFC 116 along the gas passage. In some implementations, a manifold (not shown) can be used to mix the gas before it enters MFC 116, and then the mixed gas is delivered to gas distribution unit 112. Gas source 114 can include various gas supply mechanisms, such as gas tanks. Depending on the specific embodiment, gas distribution unit 112 can be an injector or a shower head. In some configurations, a window can be integrated with gas distribution unit 112, serving both as a shower head and as a seal for chamber 102. Pressure gauge 124 is used to measure the pressure within chamber 102.

[0081] In some implementations, the gas distribution unit 112 may also include a gas injection port from the inner wall of the process chamber 102.

[0082] Furthermore, chamber 102 includes pump 128 and vacuum valve 126. Pump 128, in some implementations, may be a turbomolecular pump (TMP), tasked with extracting unused gases and reaction byproducts from chamber 102 and discharging them into exhaust port 132 via exhaust line 130. The opening and closing position of movable parts (e.g., cover plates or disks) of vacuum valve 126 plays a crucial role in coordinating with pump 128 to determine the gas extraction rate.

[0083] The position of the movable part of vacuum valve 126 plays a crucial role in determining gas conduction capacity and works in conjunction with the capacity of pump 128. It should be noted that the movable cover used herein is for illustrative purposes only. Valves can take many forms known in the art, such as, but not limited to, gate valves, butterfly valves, ball valves, and diaphragm valves. Each type of valve has a differently designed movable part. This movable part can be moved by an actuator controlled by a drive current, which serves as the setpoint for vacuum valve 126.

[0084] To adjust the valve position, an actuator controlled by a valve controller is used. The drive current for the actuator is used as the setpoint for the vacuum valve 126. This controller utilizes a proportional-integral-derivative (PID) control mechanism to determine the desired position of the movable part to maintain a steady chamber pressure and works in conjunction with the system controller 140. Typically, PID control can take hundreds of milliseconds to correctly position the movable part, which can limit the cycle time of the ALE process, especially during gas switching. Therefore, minimizing or eliminating the need for gas switching is beneficial for improving process efficiency.

[0085] System controller 140 manages MFC 116, pump 128 and vacuum valve 126 to maintain steady-state chamber pressure monitored by pressure gauge 124.

[0086] Three types of PID control are used in process system 100: MFC 116 includes MFC PID control 170 for establishing a specified flow rate; Vacuum valve 126 includes valve PID control 145 for establishing the position of the movable part to coordinate with pump 128 to adjust conduction capability; and • The system PID control 141 operated by the system controller 140 is used to establish a stable chamber pressure via the pressure gauge 124, MFC 116 and vacuum valve 126.

[0087] Chamber 102 also includes a chuck 134 for supporting substrate 136. The chuck 134 can be designed in various forms, such as an electrostatic chuck (ESC) or a vacuum chuck. The chuck 134 is connected to a bias unit 138 for applying a bias voltage to substrate 136. The bias unit 138 is crucial for controlling ion energy during the process and can be an RF power generator operating in the frequency range of approximately 100 kHz to 60 MHz. In such a configuration, a resonator can be used to match the output impedance of the RF power generator to the load impedance of chamber 102 connected via chuck 134. Simultaneously, a DC blocking capacitor can be provided to establish a steady-state bias for chuck 134. Although such techniques for establishing a steady-state bias are known, their limitation is that a steady-state bias cannot be achieved instantaneously and typically requires tens of milliseconds to establish.

[0088] In the context of this invention, a significantly shorter sputtering step is required after the surface of substrate 136 undergoes a neutral particle-induced surface modification step. This shortened duration is crucial to prevent further surface modification, which could otherwise unintentionally trigger a re-intervention (RIE) during the sputtering step.

[0089] Therefore, in a preferred embodiment of the invention, a custom waveform generator is used to provide bias to the substrate 136. This custom waveform generator is capable of establishing substrate bias in microseconds or less. It maintains the bias by applying a negative ramp voltage, such as... Figure 2 As shown. In this preferred embodiment, the sputtering step includes a series of custom waveform pulses, each pulse providing a stable bias to the substrate for a duration of a few microseconds or less. Within such a short duration, the energetic ions remove only a small fraction of the modified layer by physical sputtering. Furthermore, the duration of the sputtering step is designed so that the modified layer can be removed without triggering further surface modification due to neutral particles.

[0090] Returning to the chamber pressure control mechanism, such as Figure 1BAs shown, an exemplary vacuum valve 126 includes a valve body 148 and a movable part 146. The valve body 148 has an opening, and the position 150 of the movable part 146 plays a crucial role in determining gas conduction capability, which is achieved by adjusting the opening size and works in conjunction with the pumping capacity of the pump 128. To adjust the position 150 of the part 146, an actuator 142 controlled by a valve controller 144 is provided. The valve controller 144 utilizes valve PID control 145 to establish the desired position so as to maintain a steady-state chamber pressure according to a set value and in coordination with the controller 140. The set value may be the current driving the actuator 142. Typically, the valve PID control 145 requires hundreds of milliseconds to bring the movable part 146 of the vacuum valve 126 to the appropriate position, which constitutes a limiting factor in the cycle time of the ALE process, especially during gas switching. Therefore, addressing this bottleneck is considered beneficial to process efficiency.

[0091] The system controller 140 operates the MFC 116, pump 128, and vacuum valve 126 via system PID control 141 to maintain the steady-state chamber pressure measured by pressure gauge 124.

[0092] Figure 1C A schematic diagram of an MFC 116 is shown as an example. The MFC 116 includes an inlet 152 and an outlet 154, which are interconnected via a gas conduction channel 156. A proportional valve (not shown) directs a portion of the gas into channel 158. A flow sensor 160 (typically a thermal sensor) connected to channel 158 measures the flow rate by detecting the temperature difference between two specific points along the flow path. The measured flow rate serves as an indication of the overall flow rate in the gas conduction channel 156.

[0093] The MFC also includes a solenoid valve with a spring 162 for holding the plunger 164 in a corresponding position. The ability of gas to pass through the orifice 163 is determined by the position of the plunger 164. When the plunger blocks the passage within the orifice 163, the gas flow is cut off. The solenoid coil 165 controls the position of the plunger 164, and the plunger position is determined by the magnetic force generated by the coil current and the elastic force of the spring 162.

[0094] Flow sensor 160 transmits its detection result to MFC controller 168, which compares the result with a preset value representing the desired gas flow rate. When a difference exists between the actual flow rate and the desired gas flow rate, MFC controller 168 instructs valve actuator 166 to adjust the current in coil 165, thereby changing the plunger position. This adjustment process continues until the flow rate matches the desired target value. To speed up this process, MFC controller 168 employs MFC PID control 170. However, this adjustment process can still take hundreds of milliseconds, which is not ideal for ALE processes.

[0095] In some embodiments of the invention, the setpoints for the MFC and vacuum valve 126 are determined during a training or setup phase. In one embodiment, the MFC setpoint includes an electromagnetic coil current for determining the position of the plunger 164. The vacuum valve 126 setpoint includes a drive current for the actuator 142 in the vacuum valve 126 for moving the movable part 146.

[0096] These settings are stored in the storage medium of the system controller 140. The settings for MFC 116 may be related to the type of gas and the design of the MFC. The settings for vacuum valve 126 may depend on the specific valve used.

[0097] When the process formulation is used in production, the MFC PID control 170 is disabled, significantly reducing MFC operating time. Applying current to the solenoid coil 165 quickly positions the plunger 164 in the correct position. Although some gas is still diverted to channel 158 for monitoring by the flow sensor 160, this monitoring primarily serves a confirmatory function to ensure consistency with the desired gas flow rate.

[0098] Similar to the MFC setup, the position 150 of the movable part 146 of the vacuum valve 126 is also determined during the training or setup phase. Once the ALE process is used for production operation, the position of the vacuum valve remains fixed, thereby significantly reducing the time required to establish steady-state chamber pressure when the ALE process formulation is repeatedly used for the processing of multiple substrates.

[0099] Once the training or setup phase is complete, the system PID control 141 is also disabled.

[0100] Figure 2 A waveform diagram of ALE operation in process system 100 is shown as an example. Figure 2 The upper part shows the waveforms of the plasma source voltage Vsource, the substrate bias voltage Vsubstrate, and the output voltage Voutput of the custom waveform generator. Figure 2 The lower part shows the electromagnetic coil current of the MFC 116, and the corresponding adjustment relationship of the chamber pressure with the coil current.

[0101] The custom waveform generator first outputs a brief positive voltage pulse to neutralize the positive charge remaining on the substrate from the previous pulse with electrons. Subsequently, the custom waveform generator applies a negative bias voltage V. B Then it enters a phase that gradually decreases to a more negative voltage. This negative ramp is used to compensate for the retained positive charge by ions, thereby maintaining a constant substrate bias Vsubstrate for ion acceleration, such as... Figure 2 As shown. For the sake of simplicity, Figure 2 This is just one example of a custom waveform shown during the sputtering step. However, a large number, even thousands, of such waveforms may be used in the sputtering step of an ALE process, depending on the application and requirements. Furthermore, an ALE process may also include several or more cycles, again depending on the specific application.

[0102] The duration of the surface modification step is T. A The duration of the sputtering step is T. B In the ALE process chamber, the gap between the gas distribution unit 112 and the chuck 134 is typically 3 to 30 centimeters. Neutral particles diffuse from inside the chamber 102 near the plasma source to the substrate 136 held by the chuck 134, taking approximately 20 to 200 milliseconds. Theoretically, surface reactions can be completed in milliseconds, as described by Karanik et al. in their paper "Atomic Layer Etching: Rethinking the Art of Etch" (J. Phys. Chem. Lett., Vol. 9, 2018, pp. 4814–4821). Considering practical factors, neutral particles may require an additional 30 to 300 milliseconds to diffuse and penetrate to the bottom of the etched structure to react with surface atoms. Therefore, T A Ideally, the time should be set between 50 and 500 milliseconds, depending on the chamber volume and the aspect ratio of the structure to be etched on the substrate. B The duration was designed to be 10 to 50 milliseconds to allow sufficient time for ion removal of the modified layer while minimizing further surface modification during the sputtering step. Chamber pressure, plasma source RF power, and bias voltage were all selected during the sputtering step to meet these requirements.

[0103] exist Figure 2In this process, the RF power of plasma source 106 remains constant throughout the ALE process. The power during the surface modification step can be applied in pulses at predetermined frequencies from 100 Hz to 100 kHz, with a duty cycle of 1% to 50%. The RF power of the sputtering step may differ from that of the surface modification step. In some embodiments, the RF power of the sputtering step is higher than that of the surface modification step, which increases the ion density in the plasma and helps to shorten the sputtering time. In other embodiments, the RF power of the sputtering step is lower than that of the surface modification step, which reduces the neutral particle density and the likelihood of additional surface modification, but may require a longer time to remove the modified layer due to the reduced ion density. The precise process window for RF power and time can be determined through simulation (considering all relevant factors) or by implementing Design of Experiment (DOE) tests.

[0104] In the ALE process, the sputtering step is typically performed at significantly lower chamber pressures to improve the directionality of ions in the plasma. Since a single gas or a mixture of gases is used throughout the ALE process, rapid changes in chamber pressure are required between steps. In another implementation, system controller 140 can regulate the chamber pressure without changing the gas composition. This can be achieved by adjusting the gas flow rate of MFC 116 or changing the position of movable part 146 of vacuum valve 126. However, regulating the chamber pressure is relatively slow due to the involvement of multiple PID controls, typically requiring tens to hundreds of milliseconds. MFC controller 168 uses MFC PID control 170 to move plunger 164 to the position required to achieve the desired flow rate. Subsequently, system controller 140, via system PID control 141, utilizes valve controller 144 to adjust the position 150 of movable part 146 to achieve the desired chamber pressure. For MFC 116, the first electromagnetic coil current corresponds to the first steady-state chamber pressure of the surface modification step, and the second electromagnetic coil current corresponds to the second steady-state chamber pressure of the sputtering step. The setting value of vacuum valve 126 remains unchanged from the surface modification step to the sputtering step.

[0105] A significant feature of this invention is its ability to "learn" these setpoints through training or a setup program. This program can be dedicated or as part of the execution of the ALE process formulation. The determined setpoints are stored in the storage medium of the system controller 140.

[0106] During production, once the system controller 140 receives the process recipe, it retrieves the setpoints of MFC 116 and vacuum valve 126 and disables PID control. During the surface modification step, the first setpoint of MFC 116 and the setpoint of vacuum valve 126 move plunger 164 and movable part 146 to a designated first position, thereby rapidly establishing a first chamber pressure within milliseconds. During the sputtering step, a second setpoint of MFC 116 is retrieved, causing plunger 164 to move to a designated second position while movable part 146 remains stationary, thereby rapidly achieving a second steady-state chamber pressure. In some embodiments, the setpoints can be retrieved once and stored in a high-speed buffer (such as SRAM) for use throughout the ALE process.

[0107] Figure 3A An exemplary flowchart for determining setpoints in a training or setup procedure is shown. Process 300 begins at step 302, where a single gas or a mixture of gases is supplied from gas source 114 to gas distribution unit 112 according to the flow rate specified in the process recipe. In step 304, a first steady-state chamber pressure for the surface modification step is established using PID control. In step 306, the first solenoid current of MFC 116, as a first setpoint, and the setpoint of vacuum valve 126 are recorded in the storage medium of system controller 140. In step 308, the setpoint of MFC 116 is adjusted to achieve a second steady-state chamber pressure for the sputtering step, while keeping the position of the movable part of the vacuum valve unchanged. In step 310, the second solenoid current corresponding to the second steady-state chamber pressure is recorded as a second setpoint for MFC 116.

[0108] Figure 3BAn exemplary ALE process 301 is illustrated. Process 301 begins at step 312, retrieving first settings for MFC 116 and vacuum valve 126 from a storage medium. System controller 140 deploys these settings, moving the plunger 164 of MFC 116 and the movable part 146 of vacuum valve 126 to the positions required for the surface modification step. In step 314, a single gas or a mixture of gases from gas source 114 is introduced into gas distribution unit 112 via MFC 116. In step 316, a first steady-state chamber pressure is established after a predetermined duration. This duration can be obtained by monitoring the output of pressure gauge 124 during training or setup. In step 318, RF power required for the surface modification step is supplied to plasma source 106 by RF power generator 108. The RF power may include at least one frequency in the range of 100 kHz to 60 MHz, with a power range of 50 watts to 5000 watts. In one embodiment, the RF power remains constant during the surface modification step. In another embodiment, RF power can be pulsed at a predetermined frequency (100 Hz to 100 kHz) during the surface modification step, with a duty cycle of 1% to 50%. Plasma is then generated within chamber 102.

[0109] In step 320, the surface of substrate 136 is exposed to plasma for 50 to 500 milliseconds to complete the surface modification step. In step 322, the second set value of MFC 116 is retrieved. In step 324, a second steady-state chamber pressure is established by adjusting the position of plunger 164 while keeping the position of movable part 146 of vacuum valve 126 unchanged.

[0110] Upon proceeding to step 326, the sputtering step, a custom waveform generator is activated to apply a bias voltage to the substrate via a pulse sequence consisting of a custom waveform, lasting from 10 to 50 milliseconds. Within this timeframe, the plasma ions accelerated by the substrate bias remove the modified surface layer while preventing neutral particles in the plasma from causing additional surface modification. The substrate bias voltage can be between 100 and 10,000 volts.

[0111] In step 328, system controller 140 checks whether all ALE loops have been completed. If not, the ALE loop is repeated.

[0112] Although ALE with a single process gas or a mixture of gases is used as an example of the concept of this invention, the method for rapidly establishing steady-state chamber pressure is not limited to ALE. This method can be widely applied to other processes requiring precise chamber pressure control, such as RIE, PECVD, ALD, hot-based etching, radical etching, and radical deposition. In these processes, rapidly and accurately establishing chamber pressure using predetermined setpoints can improve overall process efficiency, accuracy, and yield, making this method applicable to a variety of semiconductor manufacturing technologies.

[0113] Figure 4 A flowchart of this general rapid chamber pressure establishment method is shown. In this general scenario, multiple MFCs can be used. Process 400 begins at step 402, where the system controller executes a training or setup procedure to determine the required setpoints for the MFCs and vacuum valve. This procedure includes adjusting the flow rate of the gas or gas mixture via the MFCs and controlling the position of a movable component in the vacuum valve to regulate gas conduction. At this stage, the system relies on PID control and pressure gauge feedback to establish and maintain a steady-state chamber pressure. Once a steady-state flow rate is established, the system continuously adjusts the position of the movable component in the vacuum valve until a stable chamber pressure is reached.

[0114] In step 404, once the optimal setpoints are determined, the system controller stores these setpoints in a storage medium. These setpoints may include the specific solenoid current of the MFC and the drive current of the vacuum valve actuator. After completing the training or setup procedure, in step 406, the system controller retrieves the stored setpoints whenever the process recipe requires establishing vacuum chamber pressure. In step 408, the system does not rely on continuous operation of PID control but directly applies the setpoints to the MFC and vacuum valve actuator. In this embodiment, PID control is turned off after the setpoints are determined, thereby significantly reducing the need for real-time adjustments during production. By applying predetermined solenoid current and actuator setpoints, the system can quickly and stably achieve the required chamber pressure.

[0115] In step 410, the flow rate of the gas or gas mixture introduced into the chamber is adjusted according to the retrieved set value, enabling the system to efficiently reach the steady-state chamber pressure. Optionally, a pressure gauge can be used to verify the steady-state chamber pressure.

[0116] This method ensures rapid chamber pressure establishment with minimal fluctuations, thereby improving the accuracy and repeatability of semiconductor manufacturing, ALE, and other processes. By identifying and storing optimal setpoints during the training or setup phase, the system can reliably reproduce these conditions across multiple cycles, simplifying operation and increasing throughput.

Claims

1. A process system for performing ALE processes, characterized in that, include: Plasma process chamber, which is the internal space used to maintain a vacuum environment; A plasma source, connected to an RF power generator, is used to generate plasma in the chamber; The bias unit is connected to the chuck. A gas distribution unit, which is used to receive a single gas or a mixture of gases from a gas source via an MFC; Vacuum valves and pumps, used to extract unconsumed gases and reaction byproducts; and The system controller is used for: The process system is operated to perform an ALE process, which includes a surface modification step and a sputtering step using the same process gas. The bias unit is operated, which includes a custom waveform generator for providing a bias voltage to accelerate ions generated in the plasma, wherein the duration of the sputtering step is designed to be below a threshold for generating additional surface modification during the step. as well as The chamber is switched between two steady-state chamber pressures by using predetermined MFC and vacuum valve setpoints, wherein the setpoints are determined by training or setting a program, and the steady-state chamber pressures are achieved without involving MFC and PID control of the process system.

2. The process system according to claim 1, wherein the set value of the MFC includes the electromagnetic coil current.

3. The process system of claim 1, wherein the MFC setpoint includes a first setpoint corresponding to a first steady-state chamber pressure for performing a surface modification step, and a second setpoint corresponding to a second steady-state chamber pressure for performing a sputtering step.

4. The process system according to claim 1, wherein the setpoint of the vacuum valve includes a current for controlling the actuator.

5. The process system of claim 1, wherein the training or setup procedure further includes executing a process recipe or a portion thereof to record the set value, wherein the PID control of the MFC and the vacuum valve is activated.

6. The process system of claim 1, wherein the system controller sets the surface modification step to last for 50 to 500 milliseconds and the sputtering step to last for 10 to 50 milliseconds.

7. The process system of claim 1, wherein the plasma source is drawn from RF during the surface modification step. The power generator receives RF power, which is applied in pulses at a predetermined frequency of 100Hz to 100kHz and with a duty cycle of 1% to 50%.

8. The process system according to claim 1, wherein the system controller adjusts the pressure value of the chamber to between 1 m Torr and 500 m Torr.

9. A method for performing an ALE process on a substrate in a process system, characterized in that, This includes the following steps performed by the system controller: a) Retrieve the first setpoint of the MFC and the setpoint of the vacuum valve from the storage medium of the system controller, and move the movable parts of the MFC plunger and the valve to the designated positions corresponding to the setpoints; b) Introduce a single gas or a mixture of gases into the chamber using a gas distribution unit connected to the gas source via an MFC; c) Establish the first steady-state chamber pressure; d) Receive RF power from an RF power generator to generate plasma in the chamber; e) Exposing the substrate surface to the plasma for a predetermined duration; f) Retrieve the second setting value of MFC; g) While keeping the vacuum valve setpoint unchanged, establish a second steady-state chamber pressure using the second setpoint of the MFC; h) Activate the custom waveform generator to provide a bias voltage to the substrate; and i) Repeat steps a) through h) to complete the ALE process.

10. The method of claim 9, wherein the method further comprises performing a training or setup procedure by executing a process recipe or a portion thereof to record the setpoint, wherein PID control of the MFC and the vacuum valve is activated.

11. The method of claim 9, wherein the duration of step e) is 50 to 500 milliseconds.

12. The method of claim 9, wherein the duration of step h) is 10 to 50 milliseconds.

13. The method of claim 9, wherein in step h), the bias voltage range established by the custom waveform generator is 100 to 10000 volts.

14. The method of claim 9, wherein the set value of the MFC includes the electromagnetic coil current.

15. The method of claim 9, wherein the setpoint of the vacuum valve includes a current for controlling the actuator.

16. A method for establishing vacuum chamber pressure by a system controller, characterized in that, Includes the following steps: a) Perform a training or setup procedure to determine the setpoints for the MFC and vacuum valve; b) Store the set value in the storage medium of the system controller; c) Retrieve the set value from the storage medium; d) Apply the setpoints to the actuators of the MFC and vacuum valve; and e) Introduce a single gas or a mixture of gases into the chamber to establish a steady-state chamber pressure corresponding to the set value.

17. The method of claim 16, wherein in step a), the system controller establishes the steady-state chamber pressure by utilizing system PID control comprising a pressure gauge, a vacuum valve, and an MFC, and PID control of the MFC and the vacuum valve.

18. The method of claim 17, wherein the PID control is disabled once the setpoint is determined.

19. The method of claim 16, wherein the set value of the MFC includes the electromagnetic coil current.

20. The method of claim 16, wherein the setpoint of the vacuum valve includes a current for the vacuum valve actuator.

Images