System and Method For Chamber Cleaning Endpoint Detection

Abstract

The present invention relates to a chamber cleaning endpoint detection system and method capable of accurately detecting an endpoint by analyzing a signal measured during a cleaning process. According to one embodiment, an endpoint detection system may be provided for a cleaning process of a semiconductor chamber, comprising: a sensor that detects the amount of deposited material remaining in the chamber and generates a product signal that calculates the amount of product generated in the cleaning process; and a computing unit that classifies the cleaning process into a plurality of regions over time based on the amount of change in the product signal and detects the endpoint.

Description

Technology Field

[0001] The present invention relates to a chamber cleaning endpoint detection system and method capable of accurately detecting an endpoint by analyzing a signal measured during a cleaning process. Background Technology

[0002] The present invention belongs to the field of chamber cleaning control and process diagnosis during semiconductor manufacturing processes. In particular, it is a technology that detects cleaning endpoints by quantitatively analyzing process diagnosis signals (e.g., Time-of-Flight Mass Spectrometer, TOF-MS) based on the thickness distribution of deposits within the chamber, and can be utilized to diagnose the state of a semiconductor process in real time.

[0003] In addition to TOF-MS, various measurement instruments such as Optical Emission Spectroscopy (OES) and Residual Gas Analyzers (RGA) can be used for process diagnostics. These instruments play a crucial role in enhancing the precision of process control by enabling the simultaneous monitoring of plasma gas flow, temperature distribution, gas composition, and product formation status.

[0004] TOF sensors measure the flight time of ions to provide the intensity of the product signal in the cleaning process, and OES analyzes the emission spectrum within the plasma to identify the type and concentration of reactants. By integrating and analyzing this diverse sensor data, even minute variations in the process can be precisely identified, and appropriate cleaning timing can be calculated.

[0005] Conventional chamber cleaning endpoint detection technologies have often remained simple methods that rely primarily on the absolute amount or rate of change of signals measured by sensors. This approach has the following limitations.

[0006] (1) Unclear signal criteria

[0007] If the endpoint is determined based solely on a single reference value or simple rate of change analysis, misjudgment may occur due to minute changes in the process environment or disturbances. For example, it may fail to reflect signal changes caused by pressure differences within the chamber during the cleaning process or the inflow status of cleaning gas into the sampling line, thereby risking a situation where the yield decreases due to impurities resulting from insufficient cleaning, or productivity decreases due to excessive cleaning time.

[0008] (2) Deposited thickness distribution not reflected

[0009] Deposits within the chamber are not spatially uniform and are divided into thick and thin regions depending on process conditions such as temperature distribution and plasma gas flow. However, existing technologies do not consider this spatial heterogeneity and determine endpoints based on a single signal standard, which can lead to reduced precision.

[0010] Accordingly, a method for accurate endpoint detection of the cleaning process is required. The problem to be solved

[0011] The present invention provides a chamber cleaning endpoint detection system and method for accurate endpoint detection of a cleaning process. means of solving the problem

[0012] According to one embodiment, an endpoint detection system may be provided for a cleaning process of a semiconductor chamber, comprising: a sensor that detects the amount of deposited material remaining in the chamber and generates a product signal that calculates the amount of product generated in the cleaning process; and a computing unit that classifies the cleaning process into a plurality of regions over time based on the amount of change in the product signal and detects the endpoint.

[0013] Additionally, the above-mentioned operation unit can be classified according to the amount of change of the product signal into a first region, which is a region where the intensity of the product signal is maintained relatively constant for the cleaning process; a second region, where the speed of the cleaning process begins to decrease as the product signal decreases; and a third region, where the slope of the second region begins to become gentle.

[0014] In addition, the above operation unit can be defined with respect to the first region by the following Equation 1.

[0015] [Formula 1]

[0016] S(t)=S0for (t≤t1)

[0017] (Here, t is the cleaning process time, S(t) is the intensity of the product signal of the cleaning process over time measured during the cleaning process, and S0 is the initial signal amount)

[0018] In addition, the above operation unit can be defined as an exponential function according to the following Equation 2 for the above second region.

[0019] [Equation 2]

[0020] S(t) = A·e^(-t / τ)

[0021] In addition, the above operation unit may differentiate Equation 2 to produce the following Equations 3 and 4, and define the point in time when the result value according to Equation 4 becomes constant as the third region.

[0022] [Equation 3]

[0023] S'(t) = -(A / τ·e^(-t / τ= -S(t) / τ

[0024] [Equation 4]

[0025] S(t) / S'(t) = -τ or S'(t) / S(t) = -1 / τ

[0026] In addition, the above computation unit can perform modeling for the above third region through the following Equation 5.

[0027] [Formula 5]

[0028] S(t) = S A ·exp(-(t-(t₂+Δτ A ) + S B ·exp(-(t-(t₂+Δτ B ), (t ≥ t₂ + Δ

[0029] (Here, S A , S B is the initial signal strength of each exponential component, τ A , τ B ε is the decay time constant of each exponential function, t₂ is the start time of the third region (exponential decay region), and Δ is the time difference to reflect the period after the inflection point.

[0030] In addition, the above-mentioned computational unit determines the damping time constant (τ) through regression analysis with respect to the above-mentioned Equation 5. A , τ B ) and initial signal strength (S A , S B ) can be derived.

[0031] In addition, the above operation unit can detect the endpoint through the following Equation 6.

[0032] [Equation 6]

[0033] t endpoint = t₂+Δ·(max{τ A , τ B})

[0034] (here, max{τ A , τ B} is the damping time constant τ A , τ B The larger of the two, m is the damping time constant (τ A , τ B Multiples for

[0035] In addition, according to one embodiment, an endpoint detection method for a cleaning process of a semiconductor chamber may be provided, comprising: a sensor detecting the amount of deposited material remaining in the chamber to generate a product signal that calculates the amount of product generated in the cleaning process; a computing unit classifying the cleaning process into a plurality of regions over time based on the amount of change of the product signal; and detecting the endpoint.

[0036] Additionally, the above-mentioned operation unit can be classified according to the amount of change of the product signal into a first region, which is a region where the intensity of the product signal is maintained relatively constant for the cleaning process; a second region, where the speed of the cleaning process begins to decrease as the product signal decreases; and a third region, where the slope of the second region begins to become gentle.

[0037] In addition, the above operation unit can be defined with respect to the first region by the following Equation 1.

[0038] [Formula 1]

[0039] S(t)=S0for (t≤t1)

[0040] (Here, t is the cleaning process time, S(t) is the intensity of the product signal of the cleaning process over time measured during the cleaning process, and S0 is the initial signal amount)

[0041] In addition, the above operation unit can be defined as an exponential function according to the following Equation 2 for the above second region.

[0042] [Equation 2]

[0043] S(t) = A·e^(-t / τ

[0044] In addition, the above operation unit may differentiate Equation 2 to produce the following Equations 3 and 4, and define the point in time when the result value according to Equation 4 becomes constant as the third region.

[0045] [Equation 3]

[0046] S'(t) = -(A / τ·e^(-t / τ= -S(t) / τ

[0047] [Equation 4]

[0048] S(t) / S'(t) = -τ or S'(t) / S(t) = -1 / τ

[0049] In addition, the above computation unit can perform modeling for the above third region through the following Equation 5.

[0050] [Formula 5]

[0051] S(t) = S A ·exp(-(t-(t₂+Δτ A ) + S B ·exp(-(t-(t₂+Δτ B ), (t ≥ t₂ + Δ

[0052] (Here, S A , S B is the initial signal strength of each exponential component, τ A , τ B ε is the decay time constant of each exponential function, t₂ is the start time of the third region (exponential decay region), and Δ is the time difference to reflect the period after the inflection point.

[0053] In addition, the above-mentioned computational unit determines the damping time constant (τ) through regression analysis with respect to the above-mentioned Equation 5. A , τ B ) and initial signal strength (S A , S B ) can be derived.

[0054] In addition, the above operation unit can detect the endpoint through the following Equation 6.

[0055] [Equation 6]

[0056] t endpoint = t₂+Δ·(max{τ A , τ B})

[0057] (here, max{τ A , τB} is the damping time constant τ A , τ B The larger of the two, m is the damping time constant (τ A , τ B Multiples for ) Effects of the invention

[0058] A chamber cleaning endpoint detection system and method according to an embodiment of the present invention can improve process stability by precisely reflecting the spatial distribution of a deposit and the cleaning response over time to minimize endpoint detection errors, and by minimizing the reduction in productivity caused by unnecessary cleaning processes and the occurrence of defects caused by insufficient cleaning. Brief explanation of the drawing

[0059] Figure 1 illustrates an exemplary pattern of a product signal generated during a cleaning process. Figure 2 illustrates the rate of change over time of the product signal generated during the cleaning process. Figure 3 illustrates the ratio of the product signal according to the rate of change of the product signal generated in the cleaning process. Figure 4 illustrates the process of modeling the third region, which is exponentially damped, with two exponential functions. Specific details for implementing the invention

[0060] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings.

[0061] The embodiments of the present invention are provided to more fully explain the invention to those skilled in the art, and the following embodiments may be modified in various different forms, and the scope of the invention is not limited to the following embodiments. Rather, these embodiments are provided to make the disclosure more faithful and complete and to fully convey the spirit of the invention to those skilled in the art.

[0062] Additionally, in the drawings below, the thickness or size of each layer is exaggerated for convenience and clarity of explanation, and like reference numerals in the drawings refer to like elements. As used herein, the term "and / or" includes any one of the listed items and all combinations of one or more thereof. Furthermore, in this specification, the meaning of "connected" refers not only to cases where Member A and Member B are directly connected, but also to cases where Member C is interposed between Member A and Member B so that Member A and Member B are indirectly connected.

[0063] The terms used herein are for describing specific embodiments and are not intended to limit the invention. As used herein, the singular form may include the plural form unless the context clearly indicates otherwise. Additionally, as used herein, "comprise" and / or "comprising" specify the presence of the mentioned features, numbers, steps, actions, parts, elements, and / or groups thereof, and do not exclude the presence or addition of one or more other features, numbers, actions, parts, elements, and / or groups.

[0064] Although terms such as "first," "second," etc. are used in this specification to describe various components, parts, regions, layers, and / or parts, it is obvious that these components, parts, regions, layers, and / or parts should not be limited by these terms. These terms are used only to distinguish one component, part, region, layer, or part from another region, layer, or part. Accordingly, the first component, part, region, layer, or part described below may refer to the second component, part, region, layer, or part without departing from the teachings of the present invention.

[0065] Spatial terms such as "beneath," "below," "lower," "above," and "upper" are used to facilitate understanding of one element or feature depicted in the drawings and another element or feature. These spatial terms are intended to facilitate understanding of the invention according to various process or usage states of the invention and are not intended to limit the invention. For example, if an element or feature in the drawings is inverted, an element described as "beneath" or "below" becomes "upper" or "on top." Therefore, "below" is a concept that encompasses "upper" or "below."

[0066] Below, a detailed process of the chamber cleaning endpoint detection system and method according to an embodiment of the present invention will be described.

[0067] Figure 1 illustrates an exemplary pattern of a product signal generated during a cleaning process.

[0068] First, the cleaning process refers to a process of cleaning impurities and deposits within the process chamber by means such as etching, and the product signal can be defined as the deposition thickness or amount of these impurities and deposits.

[0069] A chamber cleaning endpoint detection system according to an embodiment of the present invention may be equipped with, for example, a Time-of-Flight Mass Spectrometer (TOF) located within a process chamber, and the product signal may be detected through the TOF sensor. However, in addition to the TOF sensor, the chamber cleaning endpoint detection system according to an embodiment of the present invention may also generate the product signal through sensors such as Optical Emission Spectroscopy (OES) or Residual Gas Analyzer (RGA). Furthermore, the chamber cleaning endpoint detection system according to an embodiment of the present invention may further include a computational unit, and as described below, the computational unit may calculate the amount of change in the product signal and, based on the amount of change, divide the cleaning process into, for example, three regions to detect (calculate) the final endpoint.

[0070] In addition, the chamber cleaning endpoint detection method according to an embodiment of the present invention can be performed based on the configuration of such an endpoint detection system, and each step of the chamber cleaning endpoint detection method is described below.

[0071] In the chamber cleaning endpoint detection method according to an embodiment of the present invention, the deposition area within the chamber can first be classified. Referring to FIG. 1, in the chamber cleaning endpoint detection method according to an embodiment of the present invention, the cleaning process within the process chamber can first be divided into a first area (Section 1), which is an area where the thickness of the remaining impurities or deposit is relatively thick, a second area (Section 2), which is an intermediate area, and a third area (Section 3), which is a relatively thin area.

[0072] (1) First area

[0073] The first region can be defined as a region where the thickness of impurities or deposits is relatively thick, or as a bulk deposition region. The first region is a wide area where key process conditions, such as temperature distribution and plasma gas flow within the process chamber, are optimized, and the cleaning gas can access it relatively easily, resulting in a relatively fast cleaning speed. Additionally, it may have the characteristic of having minimal surface area change during cleaning, as it is a state where thick impurities or deposits are formed.

[0074] (2) Second area

[0075] The second region can be defined as a transition region corresponding to the first region described above and the third region described below. The second region is a region of intermediate thickness existing between a region where the impurity or deposit is relatively thick and a region where it is thin, and it exhibits the characteristic that the slope of the product signal changes (inflection point occurs) as the surface area of ​​the deposit gradually decreases during the cleaning process.

[0076] (3) Third area

[0077] The third region can be defined as a region where the thickness of impurities or deposits is relatively thin (Diffusion-Limited Region). The third region is primarily an outer part of the deposition region where deposition conditions, such as gas flow, are not suitable for deposition, and where very thin deposition occurs due to the diffusion of precursors. Since the surface area decreases during cleaning, the cleaning speed decreases exponentially. In other words, the cleaning speed may be relatively slow because access or flow of the cleaning gas to the third region may not be sufficient.

[0078] That is, the chamber cleaning endpoint detection method according to an embodiment of the present invention classifies the cleaning process into three regions and can detect the endpoint through the classified regions. Below, an example of classifying a product signal into the corresponding region and mathematical modeling will be described.

[0079] In semiconductor manufacturing, various types of deposition processes are performed, and cleaning processes are applied differently depending on the characteristics of the deposited material remaining in the chamber after each process. Examples of representative deposition processes and subsequent cleaning reactions are as follows.

[0080] (1) Silicon-based material deposition

[0081] 1) SiO₂ deposition (CVD / PECVD)

[0082] Process: Formation of an oxide film by reacting silane (SiH₄) with oxygen (O₂) or an oxidizing agent.

[0083] Cleaning: Oxide film residues are removed using an HF solution or O₂ plasma, and the product is SiCl x Signals such as (e.g., SiCl₄) are observed.

[0084] 2) Si₃N₄ deposition (LPCVD / CVD)

[0085] Process: Deposition of silicon nitride thin films using silane, ammonia (NH₃), etc.

[0086] Cleaning: Plasma-based cleaning or chemical cleaning methods are used, and SiCl is the reaction product. x and NH x Series signals may occur.

[0087] (2) Metal oxide deposition (ALD / PVD)

[0088] Metal oxides are frequently used in high-precision processes due to their unique electrical and physical properties.

[0089] 1) Al₂O₃ deposition (ALD)

[0090] Process: Deposition of a uniform Al₂O₃ thin film by alternating reaction of trimethylalumin (TMA) with H₂O or O₃.

[0091] Cleaning: Residual aluminum halide (AlCl) signals may be detected using HCl plasma, etc.

[0092] 2) TiO₂ deposition (ALD / PVD)

[0093] Process: Deposition using TiCl₄ or TTIP (titanium tetraisopropoxide) as a precursor.

[0094] Cleaning: When using Cl₂ or HCl plasma, TiCl x A product signal such as was observed.

[0095] 3) HfO₂ deposition (ALD)

[0096] Process: Use HfCl₄ or Hf[N(CH₃)₂]₄ as a precursor.

[0097] Cleaning: HfCl x Product signals may appear, and a special process is required for residue removal.

[0098] (3) Metal deposition and other deposition processes (PVD / Sputtering)

[0099] 1) Cu deposition (PVD / Sputtering)

[0100] Process: CuCl during cleaning after copper deposition x Metal halide generation signals may appear.

[0101] 2) Ti deposition and TiN deposition

[0102] Process: After depositing metallic titanium or titanium nitride, TiCl in the cleaning process x The product can be observed.

[0103] In this way, in the semiconductor manufacturing process, not only silicon-based materials such as Si, SiO₂, and Si₃N₄, but also metal oxide (Al₂O₃, TiO₂, HfO₂) and metal (e.g., Cu, Ti) deposition processes are all carried out, and after each process, residues are removed by a cleaning process.

[0104] Figure 1 shows the general pattern of product signals generated in these various cleaning processes. In the initial first region, a stable signal (S) state indicated by 100% appears. Then, in the second region, which is after the first time (t₁), the signal begins to decrease as the surface area to be cleaned decreases. After the second time (t₂), a rapid signal decrease and an inflection point occur. Additionally, in the third region (t₂+Δ), which is after a certain time (Δ) has passed from the second time (t₂), an exponential decay pattern can be observed.

[0105] More specifically, the product signal obtained from actual sensors (TOF, OES, RGA, etc.) may contain various types of noise. To compensate for this, noise can be removed by applying signal processing techniques such as lowess, Kalman filter, moving average filter, or low-pass filter to the product signal. After noise removal, the point at which the signal remains constant until it begins to decrease due to the cleaning response can be defined as the first time (t₁), and the point at which the signal slope increases in the negative direction until it decreases can be defined as the second time (t₂). These boundary points serve as important criteria for clearly distinguishing the TOF signal into a stable first region, an accelerated decay second region, and an exponential decay third region, and can subsequently be utilized as basic data for regression analysis and endpoint prediction based on the third region data.

[0106] Below, the classified first to third regions will be described in more detail.

[0107] (1) Phase 1: Steady-State / Bulk Reaction Phase

[0108] [Formula 1]

[0109] S(t)=S0for (t≤t1)

[0110] Here, t represents the cleaning process time, S(t) represents the intensity of the product signal of the cleaning process over time measured during the cleaning process, and S0 represents the initial signal amount.

[0111] In the first region, a constant amount of cleaning is performed, so it can be confirmed that dS(t) / dt is close to 0. In the first region, process products are mainly generated in the bulk deposition region and the transition region, and because the thickness of the deposit is thick, cleaning proceeds at a uniform rate without a change in surface area, and the amount of product measured by process diagnostic equipment may also not change.

[0112] At this time, deposition also occurs in the diffusion-limited region, but the amount is small enough to be negligible. Therefore, since the change in the exposed area during cleaning is small, the product signal can be maintained constant during the cleaning process.

[0113] (2) Second Phase: Accelerated Etch / Transition Reaction Phase

[0114] In the second region, the cleaning rate may decrease as the surface area of ​​the material decreases in the transition region. Additionally, as cleaning is completed in some regions and the area exposed to cleaning is reduced, the product signal in the cleaning process may begin to decrease. The inflection point (t₂), obtained by flattening the product signal of the cleaning process and differentiating it twice, is calculated as the point where the slope begins to flatten in the second region. The inflection point (t₂) can be viewed as the point where the influence of the reaction in the diffusion-limited region becomes more dominant than the reaction in the bulk region.

[0115] (3) Third Phase: Exponential Tail / Diffusion-Limited Phase

[0116] In the third region, which is after the third time (t₂ + Δ) delayed by a certain amount of time (Δ, Δ > 0) from the inflection point (t₂), the deposit to be cleaned becomes very thin, and the reduction rate of the surface area is proportional to the amount of remaining deposit, and accordingly, the product signal decays exponentially.

[0117] Below, the rate of change for an exemplary product signal in each region of FIG. 1 will be explained.

[0118] Figure 2 illustrates the rate of change of a product signal generated in a cleaning process over time. Figure 3 illustrates the ratio of the product signal according to the rate of change of the product signal generated in a cleaning process.

[0119] Referring to FIGS. 2 and 3, since the point at which cleaning completely transitions to the exponential decay region may actually be later than immediately after the inflection point, the starting point of the third region can be set to the third time (t₂ + Δ), which is delayed by a certain amount of time (Δ, Δ > 0) from the inflection point. At this time, the point at which exponential decay begins can be derived by utilizing the characteristic of the exponential function that the ratio of the exponential function to its derivative is constant, thereby determining the third time (t₂ + Δ).

[0120] For any exponential function represented by Equation 2

[0121] [Equation 2]

[0122] S(t) = A·e^(-t / τ)

[0123] When differentiation is performed, it is as follows in Equation 3.

[0124] [Equation 3]

[0125] S'(t) = -(A / τ)·e^(-t / τ)= -S(t) / τ

[0126] In addition, S(t) / S'(t) can be simplified as shown in Equation 4 below.

[0127] [Equation 4]

[0128] S(t) / S'(t) = -τ or S'(t) / S(t) = -1 / τ

[0129] That is, S(t) / S'(t) can be expressed as a constant. Therefore, a third region (t₂+Δ) can be defined as the point in time when S(t) / S'(t) starts to become constant in the interval where signal reduction begins during cleaning.

[0130] In addition, in the third region (t₂+Δ), as the surface area of ​​the deposit decreases in the bulk deposition region and the transition region, the amount of product generated in these two regions may decrease. Therefore, the amount of product generated in the diffusion-limited region cannot be ignored, and additional consideration must be given to the parts that are cleaned slowly due to issues such as the access of the cleaning gas. Accordingly, in the present invention, the third region (t₂+Δ) is modeled as the sum of two or more exponential functions with different decay constants.

[0131] Figure 4 illustrates the process of modeling the third region, which is exponentially damped, with two exponential functions.

[0132] Modeling with two exponential functions is as shown in the following Equation 5.

[0133] [Formula 5]

[0134] S(t) = S A ·exp(-(t-(t₂+Δτ A ) + S B ·exp(-(t-(t₂+Δτ B ), (t ≥ t₂ + Δ)

[0135] Here, S A , S B is the initial signal strength of each exponential component, τ A , τ B θ is the decay time constant of each exponential function, t₂ is the starting point of the third region (exponential decay region), and Δ can represent the time difference to reflect the period after the inflection point.

[0136] Subsequently, in the chamber cleaning endpoint detection method according to an embodiment of the present invention, the derivation of a time constant and endpoint detection through regression analysis can be performed.

[0137] Using data collected in the third interval (t ≥ t₂+Δ) of the product signal of the cleaning process, regression analysis such as the nonlinear least squares method is performed on the model of Equation 5 above, and two or more exponential damping time constants (τ) A , τ B ) and the initial signal strength of each component (S A , S B derives ).

[0138] In this case, t-(t₂+Δ) is the damping time constant (τ A , τ B The signal exhibits a characteristic where it decreases to approximately 0.2% and 0.1% or less, respectively, after passing 6 and 7 times ) of ). In particular, the larger of the two decay time constants (max{τ A , τ BThis dominates the overall damping characteristics, and based on this, the endpoint can be calculated according to the process objective as shown in the following Equation 6.

[0139] [Equation 6]

[0140] t endpoint = t₂+Δ·(max{τ A , τ B})

[0141] Here, m is the damping time constant (τ A , τ B It can mean a multiple of ). For example, the damping time constant (τ) explained earlier A , τ B To set it to 6 or 7 times ) it can be set as m=6 or m=7.

[0142] Therefore, the chamber cleaning endpoint detection method according to an embodiment of the present invention can quantify the cleaning endpoint through the above process, and thus can be utilized as an indicator for periodic monitoring of the chamber. In particular, the time constant (τ A , τ B By continuously monitoring the trend of change in the time constant, changes in the deposition pattern and cleaning efficiency within the chamber can be detected. For example, if the time constant measured over multiple process cycles gradually increases, this can be interpreted as an indication of the accumulation of process byproducts within the chamber, clogging of the shower head orifice, or a decrease in the efficiency of the RF matching network. In such cases, a system can be established to prevent process instability and yield degradation and to detect abnormal signs early by adjusting the chamber's PM (Preventive Maintenance) cycle or performing chamber conditioning through additional cleaning processes such as wet cleaning.

[0143] The actual cleaning response fundamentally follows exponential decay, and the decay constant can be determined from the initial signal in the third region. Therefore, since the remaining cleaning process time (or the time to reach the endpoint) can be predicted, the cleaning endpoint can be determined in real time and appropriate control can be performed.

[0144] Although a chamber cleaning endpoint detection system and method according to the embodiments of the present invention have been described above as specific embodiments, this is merely illustrative and the present invention is not limited thereto, and should be interpreted as having the broadest scope in accordance with the basic concept disclosed in this specification. Those skilled in the art may implement unspecified embodiments by combining or substituting the disclosed embodiments, and this also does not deviate from the scope of the present invention. In addition, those skilled in the art may easily change or modify the embodiments disclosed based on this specification, and it is evident that such changes or modifications also fall within the scope of the present invention.

Claims

Claim 1 An endpoint detection system for a cleaning process of a semiconductor chamber, comprising: a sensor that detects the amount of deposited material remaining in the chamber and generates a product signal that calculates the amount of product generated in the cleaning process; and a computing unit that classifies the cleaning process into a plurality of regions over time based on the amount of change of the product signal and detects the endpoint. Claim 2 An endpoint detection system according to claim 1, wherein the operation unit classifies the cleaning process according to the amount of change of the product signal into a first region, which is a region where the intensity of the product signal is maintained relatively constant; a second region, where the speed of the cleaning process begins to decrease as the product signal decreases; and a third region, where the slope of the second region begins to become gentle. Claim 3 In claim 2, the operation unit is an endpoint detection system defined by the following Equation 1 for the first region. [Equation 1] S(t) = S0 for (t ≤ t1) (where t is the cleaning process time, S(t) is the intensity of the product signal of the process cleaning process over time measured during the cleaning process, and S0 is the initial signal amount) Claim 4 In claim 2, the endpoint detection system wherein the operation unit is defined by an exponential function according to the following Equation 2 for the second region. [Equation 2] S(t) = A·e^(-t / τ) Claim 5 An endpoint detection system according to claim 4, wherein the operation unit differentiates Equation 2 to produce the following Equations 3 and 4, and defines the point in time at which the result value according to Equation 4 becomes constant as a third region. [Equation 3] S'(t) = -(A / τ)·e^(-t / τ)= -S(t) / τ [Equation 4] S(t) / S'(t) = -τ or S'(t) / S(t) = -1 / τ Claim 6 In claim 2, the endpoint detection system wherein the operation unit performs modeling with respect to the third region through the following Equation 5. [Equation 5] S(t) = S A ·exp(-(t-(t₂+Δτ A ) + S B ·exp(-(t-(t₂+Δτ B ), (t ≥ t₂ + Δ)(where, S A , S B is the initial signal strength of each exponential component, τ A , τ B ε is the decay time constant of each exponential function, t₂ is the start time of the third region (exponential decay region), and Δ is the time difference to reflect the period after the inflection point. Claim 7 In claim 6, the above-mentioned operation unit determines the damping time constant (τ) through regression analysis with respect to Equation 5. A , τ B ) and initial signal strength (S A , S B An endpoint detection system that derives ). Claim 8 In claim 6, the above-mentioned operation unit is an endpoint detection system that detects an endpoint through the following Equation 6. [Equation 6]t endpoint = t₂+Δ·(max{τ A , τ B }) (here, max{τ A , τ B } is the damping time constant τ A , τ B The larger of the two, m is the damping time constant (τ A , τ B Multiples for ) Claim 9 An endpoint detection method for a cleaning process of a semiconductor chamber, comprising: a sensor detecting the amount of deposited material remaining in the chamber and generating a product signal that calculates the amount of product generated in the cleaning process; a computing unit classifying the cleaning process into a plurality of regions over time based on the amount of change of the product signal and detecting the endpoint. Claim 10 An endpoint detection method according to claim 9, wherein the operation unit classifies the cleaning process according to the amount of change of the product signal into a first region, which is a region where the intensity of the product signal is maintained relatively constant; a second region, where the speed of the cleaning process begins to decrease as the product signal decreases; and a third region, where the slope of the second region begins to become gentle. Claim 11 In claim 10, the endpoint detection method wherein the operation unit is defined by the following Equation 1 for the first region. [Equation 1] S(t)=S0 for (t≤t1)(where t is the cleaning process time, S(t) is the intensity of the product signal of the process cleaning process according to time measured during the cleaning process, and S0 is the initial signal amount) Claim 12 An endpoint detection method according to claim 10, wherein the operation unit is defined by an exponential function according to the following Equation 2 for the second region. [Equation 2] S(t) = A·e^(-t / τ) Claim 13 An endpoint detection method according to claim 12, wherein the operation unit differentiates Equation 2 to produce the following Equations 3 and 4, and defines the point in time at which the result value according to Equation 4 becomes constant as a third region. [Equation 3] S'(t) = -(A / τ)·e^(-t / τ)= -S(t) / τ [Equation 4] S(t) / S'(t) = -τ or S'(t) / S(t) = -1 / τ Claim 14 An endpoint detection method according to claim 10, wherein the operation unit performs modeling with respect to the third region through the following Equation 5. [Equation 5] S(t) = S A ·exp(-(t-(t₂+Δτ A ) + S B ·exp(-(t-(t₂+Δτ B ), (t ≥ t₂ + Δ)(where, S A , S B is the initial signal strength of each exponential component, τ A , τ B ε is the decay time constant of each exponential function, t₂ is the start time of the third region (exponential decay region), and Δ is the time difference to reflect the period after the inflection point. Claim 15 In claim 14, the above-mentioned operation unit determines the damping time constant (τ) through regression analysis with respect to Equation 5. A , τ B ) and initial signal strength (S A , S B An endpoint detection method that derives ). Claim 16 In claim 14, the endpoint detection method wherein the operation unit detects an endpoint through the following Equation 6. [Equation 6]t endpoint = t₂+Δ·(max{τ A , τ B }) (here, max{τ A , τ B } is the damping time constant τ A , τ B The larger of the two, m is the damping time constant (τ A , τ B Multiples for )

Images