Systems and methods for detecting contamination of a thin film

Abstract

A thin film deposition system deposits a thin film on a wafer. A radiation source irradiates the wafer with excitation light. An emission sensor detects an emission spectrum from the wafer in response to the excitation light. A machine learning-based analysis model analyzes the spectrum and detects contamination of the thin film based on the spectrum. Embodiments of the present invention also relate to systems and methods for detecting contamination of a thin film.

Description

Technical Field

[0001] Embodiments of the present invention relate to systems and methods for detecting contamination of thin films. Background Art

[0002] There has been a continuing need to increase computing power in electronic devices, which include smart phones, tablet computers, desktop computers, laptop computers, and many other types of electronic devices. Integrated circuits provide computing power for these electronic devices. One way to increase computing power in an integrated circuit is to increase the number of transistors and other integrated circuit components that a given area of a semiconductor substrate can include.

[0003] To continue to reduce the size of components in integrated circuits, various thin film deposition techniques have been implemented. These techniques can form very thin films. However, thin film deposition techniques also face severe difficulties in ensuring proper formation of the thin films. Summary of the Invention

[0004] Embodiments of the present invention provide a method for detecting contamination of a thin film, including: depositing a thin film on a wafer in a thin film deposition chamber; irradiating the thin film with an excitation light; detecting an emission spectrum from the thin film in response to the excitation light; and detecting contamination of the thin film by analyzing the emission spectrum using an analysis model of a control system.

[0005] Another embodiment of the present invention provides a system for detecting contamination of a thin film, including: a thin film deposition chamber configured to deposit a thin film on a wafer; a radiation source configured to irradiate the thin film with an excitation light; an emitter sensor configured to detect an emission spectrum from the wafer in response to the excitation light; and a control system coupled to the radiation source and the emitter sensor and configured to detect contamination of the thin film by analyzing the emission spectrum and to stop a thin film deposition process in the thin film deposition chamber in response to detecting contamination of the thin film.

[0006] Yet another embodiment of the present invention provides a method for detecting contamination of a thin film, including: training an analysis model to detect contamination of a thin film using a machine learning process that uses data from a plurality of spectra detected under a plurality of contaminated and non-contaminated conditions; depositing a thin film on a wafer in a thin film deposition chamber; irradiating the wafer with an excitation light; detecting an emission spectrum from the wafer in response to the excitation light; and detecting whether the thin film is contaminated by analyzing the emission spectrum using the analysis model. Brief Description of the Drawings

[0007] Figure 1 is a block diagram of a thin film deposition system according to one embodiment.

[0008] Figure 2 is an illustration of a thin film deposition system according to one embodiment.

[0009] Figure 3 is an illustration of a thin film deposition system according to one embodiment.

[0010] Figure 4 is a cross-sectional view of a semiconductor wafer according to one embodiment.

[0011] Figure 5 is a cross-sectional view of a semiconductor wafer according to one embodiment.

[0012] Figure 6 is a block diagram of a control system according to one embodiment.

[0013] Figures 7 to 13 is a flowchart of a method for detecting defects in a thin film according to various embodiments. DETAILED DESCRIPTION

[0014] In the following description, numerous thicknesses and materials are described for various layers and structures within an integrated circuit die. For each embodiment, specific dimensions and materials are given by way of example. Those skilled in the art will recognize, in accordance with the present invention, that other dimensions and materials may be used in many instances without departing from the scope of the present invention.

[0015] The following disclosure provides many different embodiments or examples for implementing different features of the provided subject matter. Specific examples of components and arrangements are described below to simplify the present invention. Of course, these are merely examples and are not intended to be limiting. For example, in the following description, forming a first component above or on a second component may include embodiments where the first and second components are in direct contact, and may also include embodiments where additional components may be formed between the first and second components such that the first and second components may not be in direct contact. In addition, the present invention may repeat reference numerals and / or letters in the various examples. This repetition is for the purpose of simplicity and clarity and does not in itself indicate a relationship between the various embodiments and / or configurations discussed.

[0016] In addition, for ease of description, spatially relative terms such as "below", "beneath", "lower", "above", "upper", etc. may be used herein to describe the relationship of one element or component to another as shown in the figures. In addition to the orientation shown in the figures, spatially relative terms are intended to include different orientations of the device in use or operation. The device may be otherwise oriented (rotated 90 degrees or at other orientations), and the spatially relative descriptors used herein may be interpreted accordingly.

[0017] In the following description, certain specific details are set forth in order to provide a thorough understanding of various embodiments of the present invention. However, those skilled in the art will understand that the present invention may be practiced without these specific details. In other instances, well-known structures associated with electronic components and manufacturing techniques have not been described in detail to avoid unnecessarily obscuring the description of the embodiments of the present invention.

[0018] Unless the context otherwise requires, throughout the specification and the following claims, the word "comprising" and its variations (such as "comprises" and "comprising") shall be construed in an open, inclusive sense, i.e., "including but not limited to".

[0019] The use of ordinal numbers such as first, second, and third does not necessarily imply a sense of order of precedence, but may only distinguish multiple instances of an action or structure.

[0020] Throughout the specification, references to "one embodiment" or "an embodiment" mean that a particular component, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Thus, the phrases "in one embodiment" or "in an embodiment" appearing throughout the specification do not necessarily all refer to the same embodiment. Moreover, in one or more embodiments, the particular components, structures, or characteristics may be combined in any suitable manner.

[0021] As used in this specification and the appended claims, the singular forms "a", "an", and "the" include plural referents unless the context clearly dictates otherwise. It should also be noted that the term "or" is generally used in its inclusive sense, including "and / or", unless the context clearly indicates otherwise.

[0022] Embodiments of the present invention provide thin films having reliable thickness and composition. Embodiments of the present invention utilize machine learning techniques to detect contamination or other defects in the thin films. Contamination can be detected in-situ, enabling the thin film deposition process to be stopped immediately after a defective deposition process. Thus, instead of having a large number of wafers receive defective thin films before testing can detect the problem, the problem is detected immediately and other wafers are not affected. Integrated circuits including the thin films will not have performance issues (which may result if the thin films are not properly formed). Additionally, batches of semiconductor wafers will have improved yield and fewer discarded wafers.

[0023] Figure 1FIG. 0 is a block diagram of a thin film deposition system 100 according to an embodiment. The thin film deposition system 100 includes a thin film deposition chamber 102 that defines an internal volume 103. The thin film deposition system 100 includes a deposition apparatus 104 configured to perform a thin film deposition process on a wafer 106 located within the internal volume 103. The thin film deposition system 100 includes a radiation source 108, an emission sensor 110, and a control system 112. The radiation source 108, the emission sensor 110, and the control system 112 cooperate to detect contamination of the thin film formed on the wafer 106.

[0024] The thin film deposition apparatus 104 may include deposition apparatuses located outside the thin film deposition chamber 102, inside the thin film deposition chamber 102, or both inside and outside the thin film deposition chamber 102. The deposition apparatus 104 may include apparatuses for performing one or more of physical vapor deposition (PVD) processes, chemical vapor deposition (CVD) processes, atomic layer deposition (ALD) processes, or other types of thin film deposition processes for forming a thin film on the wafer 106.

[0025] In some cases, the thin film may become contaminated during or after the thin film deposition process. Contamination may occur due to contaminants or impurities in the target material, precursor material, or other materials introduced into the thin film deposition chamber 102 during the thin film deposition process. In many cases, the thin film deposition process is performed under vacuum conditions. During a thin film deposition process that is expected to be under vacuum conditions, air leakage or other defects in the equipment may allow external air to enter the thin film deposition chamber 102. In such a case, oxygen in the air may oxidize the thin film. If oxidation of the thin film occurs, the thin film may not have the expected structure, composition, and properties for performing the selected function in the wafer 106.

[0026] The thin film deposition system 100 utilizes the radiation source 108, the emission sensor 110, and the control system 112 to detect contamination of the thin film. The radiation source 108 outputs excitation light 109. The excitation light 109 irradiates the thin film formed on the wafer 106. Irradiation of the thin film formed on the wafer 106 may excite atoms or molecules of the thin film. The excited atoms or molecules of the thin film may output emissions 111 in response to being excited by the excitation light 109. The emissions 111 may include light or particles (such as electrons) or a combination of light and particles. The emissions 111 collectively have an energy spectrum. The energy spectrum corresponds to the spectrum of the emitted light or the energy spectrum of the emitted particles. If the emissions 111 include photons, the energy spectrum corresponds to a photon spectrum including various wavelengths associated with the emitted photons. If the emissions 111 include particles such as electrons, the energy spectrum corresponds to the energy of the emitted particles. The energy spectrum may provide an indication of the types of atoms, molecules, or compounds included in the thin film.

[0027] As used herein, the terms "radiation" and "excitation light" refer to electromagnetic radiation either within or outside the visible spectrum. Thus, the excitation light 109 can include electromagnetic radiation outside the visible spectrum.

[0028] Although Figure 1 it is shown that the radiation source 108 is located within the thin film deposition chamber 102, the radiation source 108 can be located outside the thin film deposition chamber 102. The radiation source 108 can be partially located within the thin film deposition chamber 102 and partially located outside the thin film deposition chamber 102. If the radiation source 108 is located outside the thin film deposition chamber 102, the radiation source 108 can irradiate the wafer 106 through one or more windows or openings in the wall of the thin film deposition chamber 102.

[0029] The emission sensor 110 is configured to sense the emissions 111. The thin film can emit emissions 111 in all directions. The emission sensor 110 is positioned such that the emission sensor 110 will receive some of the emissions 111. The emission sensor 110 senses the wavelengths of the various emissions 111 received by the emission sensor 110. The emission sensor 110 outputs a sensor signal indicative of the spectrum of the emissions 111.

[0030] Although Figure 1 it is shown that the emission sensor 110 is located within the thin film deposition chamber 102, the emission sensor 110 can be located outside the thin film deposition chamber 102. The emission sensor 110 can be partially located within the thin film deposition chamber 102 and partially located outside the thin film deposition chamber 102. If the emission sensor 110 is located outside the thin film deposition chamber 102, the emission sensor 110 can receive the emissions 111 through one or more windows or openings in the wall of the thin film deposition chamber 102.

[0031] The control system 112 is coupled to the radiation source 108, the emission sensor 110, and the deposition device 104. The control system 112 receives the sensor signal from the emission sensor 110. The control system 112 analyzes the sensor signal and determines the spectrum of the emissions 111. The control system 112 can detect contamination of the thin film based on the spectrum of the emissions 111.

[0032] The control system 112 includes an analysis model 114. The analysis model 114 is trained through a machine learning process to detect contamination of the thin film based on the emission spectrum sensed by the emission sensor 110. The analysis model 114 can include a neural network or other types of machine learning models. As regarding Figure 6 and Figure 7More specifically, the analysis model 114 is trained using a training set that includes a plurality of spectra, each spectrum being associated with a contaminated or uncontaminated thin film. The machine learning process uses the training set to train the analysis model 114 to reliably detect contamination of the thin film based on the emission spectrum sensed by the emission sensor 110.

[0033] If the control system 112 detects contamination of the thin film, the control system 112 can take various response actions. For example, the control system 112 can cause the thin film deposition system 100 to stop operating based on the detected contamination in the thin film. The control system 112 can output information indicating the type of contamination. For example, if the analysis model 114 detects that the spectrum indicates an undesired oxidation of the thin film, the control system 112 can indicate oxygen contamination. If the analysis model 114 detects other types of contamination, the control system 112 can output information indicating the other types of contamination.

[0034] The control system 112 can detect contamination of the thin film in-situ. In other words, when the wafer 106 is still in the thin film deposition chamber, the control system 112 can detect contamination of the thin film during or shortly after the thin film deposition process. This can provide significant benefits because each wafer 106 can be monitored and contamination can be detected immediately, rather than detecting contaminants when a large number of wafers 106 may have already been processed and contaminated. This improves the wafer yield and reduces the number of discarded wafers.

[0035] The control system 112 can include processing, memory, and information transmission resources. The processing, memory, and information transmission resources can be located at the facility of the thin film deposition system 100. Optionally, the processing, memory, and information transmission resources can be located at a facility remote from the thin film deposition system 100. The control system 112 can be a distributed control system that includes resources in multiple locations. The control system 112 can include cloud-based resources and local physical resources.

[0036] Figure 2 is a block diagram of a thin film deposition system 200 according to one embodiment. The thin film deposition system 200 is similar to Figure 1 the thin film deposition system 100, except that the thin film deposition system 200 is a PVD system that uses an excitation laser and an optical sensor. The system 100 can utilize the components, systems, and processes described with respect to Figure 2 The thin film deposition system 200 includes a thin film deposition chamber 202 that defines an internal volume 203. A wafer support 222 supports a wafer 206 within the internal volume 203 of the thin film deposition chamber 202.

[0037] In Figure 2In an example, the thin film deposition system 200 is a PVD sputtering deposition system, but other types of deposition systems can be utilized without departing from the scope of the present invention. The thin film deposition system 200 includes a sputtering magnetron cathode 216 and a sputtering target 218. During the deposition process, a voltage source 220 applies a voltage signal to the sputtering magnetron cathode 216. As a result, atoms are ejected from the sputtering target 218. The wafer support 222 serves as the grounded anode of the sputtering system. Atoms from the sputtering target 218 accumulate on the surface of the wafer 206. As a result, a thin film is deposited on the wafer 206.

[0038] The thin film deposition system 200 further includes an exhaust passage 224 that is communicatively coupled to the interior volume 203 of the thin film deposition chamber 202. A valve 226 couples the exhaust passage 224 to a pump 228. Before the thin film deposition process, the valve 226 is opened and the pump 228 is activated. The pump 228 generates a vacuum within the interior volume 203 by pumping fluid from the interior volume 203. When the interior volume is below a threshold pressure, i.e., substantially under vacuum, the PVD process can begin.

[0039] In one example, the thin film deposition process is a titanium nitride deposition process. In this case, the sputtering target 218 is a titanium sputtering target. After evacuating the interior volume by the pump 228, nitrogen gas flows into the interior volume 203. The sputtering process is initiated in the presence of nitrogen gas. As a result, a titanium nitride thin film is deposited on the wafer 206.

[0040] In some cases, due to a leak in the thin film deposition chamber 202, air may enter the interior volume 203 during the deposition process. If this occurs, the titanium nitride thin film may oxidize, resulting in the presence of titanium oxide in the thin film. Depending on the circumstances, the presence of titanium oxide in the thin film will cause the thin film to fail to perform its function as a barrier layer or an adhesion layer.

[0041] To detect oxidation or other contamination of the thin film, the thin film deposition system 200 includes an excitation laser 208 and a light sensor 210. The excitation laser 208 and the light sensor 210 can be located in a tube 213 that extends from outside the thin film deposition chamber 202 into the thin film deposition chamber 202. Over most of the extent of the tube 213, the inner surface of the tube 213 can be substantially reflective. The end of the tube 213 can be transparent such that the excitation light and the emitted photons can pass through. In Figure 2 the figure, the transparent portion of the tube is indicated by a dashed line. In one example, the transparent portion of the tube differs from the reflective portion of the tube in that the transparent portion of the tube 213 does not include a reflective coating, while the reflective portion of the tube 213 includes a reflective coating.

[0042] The excitation laser 208 outputs excitation light 209. The excitation light is reflected within the reflective portion of the tube. At the non-reflective portion of the tube, the excitation light 209 enters the internal volume 203 of the thin film deposition chamber 202 from the tube 213. Then, the excitation light 209 irradiates the wafer 206. In particular, the excitation light 209 irradiates the thin film deposited on the wafer 206.

[0043] Some of the excitation light 209 is absorbed by the thin film. As a result, the valence electrons in the atoms or compounds of the thin film transition from a lower energy level to a higher energy level. Subsequently, the electrons return from the higher energy level to the lower energy level. When the electrons return from the higher energy level to the lower energy level, the electrons emit emission photons 211 having an energy corresponding to the energy level difference.

[0044] The spectrum of the emission photons 211 emitted by the thin film indicates the composition of the thin film. Different materials will emit photons of different wavelengths based on the atoms or molecules that make up the material. Therefore, in response to the irradiation of the excitation light, the photon spectrum of the emission photons 211 from the thin film indicates the composition of the thin film.

[0045] The light sensor 210 is configured to receive and sense the emission photons 211. In particular, the light sensor 210 is located at the transparent end of the tube 213 such that the emission photons 211 can pass through the tube 213 and be received by the light sensor 210. The light sensor 210 can be configured to sense light within a wavelength range corresponding to the desired wavelength range of the emission photons 211 from contaminated and non-contaminated thin films.

[0046] Returning to the example of depositing a titanium nitride thin film in the thin film deposition system 200, the excitation laser 208 emits excitation light between 300 nm and 330 nm. This range is selected because one possible contamination of titanium nitride is the oxidation of titanium nitride. The oxidation of titanium nitride produces titanium oxide, which is a semiconductor material having a bandgap of approximately 3.2 eV. Photons in the range of 300 nm and 330 nm have sufficient energy to excite the transition of electrons in the valence band to the conduction band. Therefore, the excitation laser 208 is selected such that the excitation photons have an energy greater than the bandgap of titanium oxide. The electrons in the valence band can absorb the photons of the excitation light and transition to the conduction band through the bandgap. When the electrons transition from the conduction band back to the valence band, the electrons will emit emission photons 211 having an energy corresponding to the difference between the energy levels of the valence band and the energy level to which the electrons return to the conduction band.

[0047] In one example, TiO2 has a photoluminescence spectrum, and the intensity peak of the photoluminescence spectrum is concentrated at about 360 nm. Thus, analysis of the spectrum of the emitted photons 211 can indicate whether the spectrum corresponds to TiO2, indicating that the thin film is oxidized and contaminated. In this example, the optical sensor 210 can be an ultraviolet radiation sensor that has a specific sensitivity to ultraviolet light in the range between 200 nm and 400 nm. Accordingly, the excitation laser 208 can be an ultraviolet laser.

[0048] Although specific examples of titanium nitride thin films that are oxidized and contaminated have been given, the thin film deposition system 200 can include the specific examples described herein without departing from the scope of the present invention: many other types of deposition processes, thin films, contamination detection, radiation sources, and radiation sensors. For example, the thin film can include Ti, TiAl, TiON, TiAlO, TiAl, TiAlC, middle-of-line (MEOL) contact metal, front-of-line (FEOL) high-k capping or metal gate, or other types of thin films.

[0049] The control system 212 is coupled to the excitation laser 208, the optical sensor 210, and the voltage source 220. The control system 212 receives a sensor signal from the optical sensor 210. The sensor signal from the optical sensor 210 indicates the spectrum of the emitted photons 211 received by the optical sensor 210 from the thin film 211. The control system 212 analyzes the sensor signal to determine whether the spectrum of the emitted photons corresponds to a contaminated thin film or a properly formed thin film. If the control system 212 determines that the photon spectrum corresponds to a contaminated thin film, the control system 212 can shut down the thin film deposition system 200 and can output an alarm. In the example of the oxidation of the titanium nitride thin film, the alarm can indicate a leak in the thin film deposition chamber 202.

[0050] The control system 212 includes an analysis model 214. The analysis model 214 is trained through a machine learning process to reliably detect photon spectra indicating various types of contamination or properly formed thin films. Regarding Figure 6 and Figure 7 , more details about training the analysis model are provided.

[0051] Although Figure 2 the excitation laser 208 and the optical sensor 210 that senses the emitted photons are shown, the thin film deposition system 200 can include other types of excitation sources and radiation sensors. In one embodiment, the excitation source is an X-ray source that emits X-ray radiation. The X-ray source irradiates the thin film with X-rays. X-rays are high-energy photons. The high-energy X-ray photons excite the thin film and cause the thin film to emit electrons via the photoelectric effect.

[0052] In an example of an X-ray source, the light sensor 210 is alternatively a photoelectron sensor. The emissions from the thin film in response to the X-rays are photoelectrons. The photoelectron sensor can determine the energy of the electrons received from the thin film via the photoelectric effect. Since the energy of the X-rays is known and since the energy of the emitted electrons is measured by the photoelectron sensor, the electron binding energy of the emitted electrons can be determined. The electron binding energy of the electrons emitted from the thin film indicates the material of the thin film. Thus, by analyzing the spectrum of the emitted electrons, the composition of the thin film can be determined. In this case, the analysis model can determine whether the thin film is contaminated based on the energy spectrum of the photoelectrons.

[0053] Figure 3 is a diagram of a thin film deposition system 300 according to one embodiment. The thin film deposition system 300 is similar to the thin film deposition systems 100, 200 of Figure 1 and Figure 2 in many respects, except that the thin film deposition system 300 includes separate chambers for thin film deposition and thin film measurement. The thin film deposition system 300 includes a thin film deposition chamber 302 and a thin film analysis chamber 334. The thin film deposition system 300 includes a deposition device 304 configured to perform a thin film deposition process on a wafer 306 supported by a support 322 in the thin film deposition chamber 302.

[0054] The thin film analysis chamber 334 and the thin film deposition chamber 302 are communicatively coupled via a transfer channel 332. After performing the thin film deposition process on the wafer 306, the wafer 306 is transferred to the thin film analysis chamber 334 via the transfer channel 332. The thin film analysis chamber 334 includes a support 330. The wafer 306 is located on the support 330 after being transferred from the thin film deposition chamber 302. The wafer 306 can be transferred by a robotic arm ( Figure 3 not shown in).

[0055] In one embodiment, since the thin film deposition chamber 302 and the thin film analysis chamber 334 are coupled together via the transfer channel 332, the vacuum or pressure conditions in the thin film deposition chamber 302 are communicated with the thin film analysis chamber 334. This means that when the wafer 306 is transferred from the thin film deposition chamber 302 to the thin film analysis chamber 334, the wafer 306 does not pass through an additional contamination environment.

[0056] A radiation source 308 and an emissions sensor 310 are located in or adjacent to the thin film analysis chamber 334. The radiation source 308 is configured to irradiate the wafer 306 with excitation light. The emissions sensor 310 is configured to receive and sense emissions from the wafer 306 in response to the excitation light. The radiation source 308 and the emissions sensor 310 can include the same types of radiation sources and emissions sensors as those described with respect to Figure 1 and Figure 2 described.

[0057] The control system 312 can receive an emission spectrum signal from the emitter sensor 310. The analysis model 314 can analyze the spectrum and determine whether the thin film is contaminated. If the thin film is contaminated, the control system 312 can output an alarm and stop the operation of the thin film deposition system 300.

[0058] Figure 4 FIG. 4 is a cross-sectional view of a wafer 406 according to an embodiment. The wafer 406 includes a nanosheet structure 440. An interface layer 442 is located above the nanosheet structure. A high-k gate dielectric layer 444 is located on the interface layer 442 and on the sidewalls of trenches 448 formed in a dielectric material layer 445. The trenches 448 are formed to be filled with the gate electrodes of the nanosheet transistors.

[0059] Before depositing the gate electrodes, a titanium nitride layer 446 is formed on the high-k gate dielectric layer 444. The titanium nitride layer serves as a work function layer for the gate electrodes that will be formed in the trenches 448. The titanium nitride increases the work function and improves the overall functionality of the nanosheet transistors.

[0060] The titanium nitride layer 446 is an example of a thin film layer formed in a Figures 1 to 3 thin film deposition system. After forming the titanium nitride layer 446, a radiation source irradiates the titanium nitride layer 446 with excitation light 409. The excitation light 409 can correspond to various examples of the excitation light given with respect to Figures 1 to 3 Thus, the excitation light can include ultraviolet radiation, X-ray radiation, or other types of radiation emitted from the radiation source.

[0061] The titanium nitride layer 446 absorbs some of the excitation light 409 and outputs an emitter 411. The emitter 411 can include photoluminescence photons emitted from the titanium nitride layer 446 in response to absorption of the excitation light 409. Optionally, the emitter 411 can include photoelectrons emitted from the titanium nitride layer 446 in response to absorption of the excitation light 409.

[0062] A radiation sensor (not shown) can receive and sense the emitter 411. The radiation sensor can include the type of emitter sensor described with respect to Figures 1 to 3 or other types of radiation sensors. The radiation sensor can provide a sensor signal to a control system that includes the analysis model described with respect to Figures 1 to 3 The control system and the analysis model can determine whether the titanium nitride is contaminated based on the spectrum of the emitter 411, as previously described with respect to Figures 1 to 3 and as will be described in more detail with respect to Figure 6 and Figure 7

[0063] Figure 5FIG. 506 is a cross-sectional view of a wafer 506 according to one embodiment. The wafer 506 includes a nanosheet structure 540. An interface layer 542 is located above the nanosheet structure 540. A high-K dielectric layer 544 is located on the interface layer 542. Sidewall spacers 551 are positioned adjacent to the high-K dielectric layer 544. A titanium nitride work function layer 546 is located on the high-K dielectric layer 544. A gate electrode 550 is formed to contact the titanium nitride work function layer 546. Trenches 552 are formed in a dielectric material layer 555. The trenches 552 are for source and drain electrodes. Source regions 553 and drain regions 557 are adjacent to the nanosheet structure 540. Silicide layers 554 and 556 are located at the source regions 553 and drain regions 557. A titanium nitride adhesion layer 556 is formed on the dielectric material layer 555, on the sidewalls of the trenches 552, and on the gate electrode 550.

[0064] The titanium nitride layer 556 is an example of a thin film layer formed in a Figures 1 to 3 thin film deposition system. After the titanium nitride layer 556 is formed, a radiation source irradiates the titanium nitride layer 556 with excitation light 509. The excitation light 509 can correspond to various examples of excitation light given with respect to Figures 1 to 3 . Thus, the excitation light can include ultraviolet radiation, X-ray radiation, or other types of radiation emitted from the radiation source.

[0065] The titanium nitride layer 556 absorbs some of the excitation light 509 and outputs an emission 511. The emission 511 can include photoluminescence photons emitted from the titanium nitride layer 556 in response to absorption of the excitation light 509. Optionally, the emission 511 can include photoelectrons emitted from the titanium nitride layer 556 in response to absorption of the excitation light 509.

[0066] A radiation sensor (not shown) can receive and sense the emission 511. The radiation sensor can include the types of radiation sensors described with respect to Figures 1 to 3 , or other types of radiation sensors. The radiation sensor can provide a sensor signal to a control system that includes an analysis model described with respect to Figures 1 to 3 . The control system and the analysis model can determine whether the titanium nitride is contaminated based on the spectrum of the emission 511, as previously described with respect to Figures 1 to 3 and as will be described in more detail with respect to Figure 6 .

[0067] Although a specific titanium nitride layer is described with respect to Figure 4 and Figure 5 , the principles of the present invention extend to other types of deposited thin films and other types of structures and other types of deposition processes. Detecting contamination using an analysis model, a radiation source, and an emission sensor can be used for a variety of thin films and many different types of contamination.

[0068] Figure 6is a block diagram of a control system 612 according to one embodiment. Figures 1 to 3 The control systems 112, 212, and 312 can include the components and functions of the control system 612. Thus, a control system 612 having the systems, processes, and components described with respect to Figures 1 to 5 can be employed. According to one embodiment, Figure 6 the control system 612 is configured to control the operation of a thin film deposition system. The control system 612 utilizes machine learning to determine whether a thin film is contaminated or whether the thin film is properly formed. The control system 612 can detect contamination of the thin film and stop further thin film deposition processes before the improperly formed thin film can affect additional wafers. The control system 612 can issue an alarm indicating a problem with the thin film deposition process.

[0069] In one embodiment, the control system 612 includes an analysis model 614 and a training module 660. The training module trains the analysis model 614 using a machine learning process. The machine learning process trains the analysis model 614 to detect whether a thin film is contaminated based on the emission spectrum from the thin film. Although the training module 660 is shown as separate from the analysis model 614, in practice, the training module 660 can be part of the analysis model 614.

[0070] The control system 612 includes or stores training set data 662. The training set data 662 includes historical thin film spectral data 664 and contamination label data 666. The historical thin film spectral data 664 includes the emission spectra of a large number of thin films. For each emission spectrum in the historical thin film spectral data 664, the contamination label data 666 includes data indicating whether the emission spectrum corresponds to a contaminated thin film. As will be elaborated in more detail below, the training module 660 utilizes the historical thin film spectral data 664 and the contamination label data 666 to train the analysis model 614 using a machine learning process.

[0071] In one embodiment, the historical thin film spectral data 664 includes data related to the emission spectra of a large number of thin films. The emission spectra include the spectra of photons or photoelectrons emitted by each of the large number of thin films. In the case of a photon spectrum, each spectrum includes the distribution of photons of various energies, wavelengths, or frequencies for a particular previously analyzed thin film. In the case of an electron spectrum, each spectrum includes the distribution of the energies of electrons for a particular previously analyzed thin film.

[0072] In one embodiment, for each historical thin film spectrum in the historical thin film spectral data 664, the contamination label data 666 includes a corresponding label. Each label indicates contamination or non - contamination. Because there can be multiple types of contamination, there can be multiple types of contamination labels. In other words, depending on the various ways in which a thin film can be contaminated during the thin film deposition process, the labels indicating contamination can fall into one of multiple contamination categories.

[0073] In one embodiment, the analysis model 614 includes a neural network. The training of the analysis model 614 will be described with respect to the neural network. However, other types of analysis models or algorithms may be used without departing from the scope of the present invention. The training module 660 utilizes the training set data 662 to train the neural network using a machine learning process. During the training process, the neural network receives historical thin-film spectral data 664 as input from the training set data 662. During the training process, the neural network outputs predicted class data. For each thin-film spectrum provided to the analysis model 614, the predicted class data predicts the class to which the spectrum belongs. The classes may include contamination, non-contamination, or various individual classes of contamination and non-contamination. The training process trains the neural network to generate predicted class data that matches the contamination label data 666 for each thin-film spectrum.

[0074] In one embodiment, the neural network includes a plurality of neural layers. The various neural layers include neurons that define one or more internal functions. The internal functions are based on weighted values associated with the neurons of each neural layer of the neural network. During training, the control system 612 compares the predicted class data with the actual labels from the contamination label data 666 for each set of historical thin-film spectral data. The control system generates an error function that indicates how well the predicted class data matches the contamination label data 666. The control system 612 then adjusts the neural network internal functions. Since the neural network generates the predicted class data based on the internal functions, adjusting the internal functions will result in different predicted class data being generated for the same combination of historical thin-film spectral data. Adjusting the internal functions can produce predicted class data that results in a larger error function (poorer match to the contamination label data 666) or a smaller error function (better match to the contamination label data 666).

[0075] After adjusting the neural network internal functions, the historical thin-film spectral data 664 is passed to the neural network again, and the analysis model 614 generates predicted class data again. The training module 660 compares the predicted class data with the contamination label data 666 again. The training module 660 adjusts the neural network internal functions again. This process is repeated for a large number of iterations of monitoring the error function and adjusting the neural network internal functions until a set of internal functions is found that produces predicted class data that matches the contamination label data 666 across the entire training set.

[0076] At the start of the training process, the predicted class data may not match the contaminated label data 666 very well. However, as the training process proceeds through many iterations that adjust the internal functions of the neural network, the error function will tend to become smaller and smaller until a set of internal functions is found that produces predicted class data that matches the contaminated label data 666. The identification of the set of internal functions that produces predicted class data that matches the contaminated label data 666 corresponds to the completion of the training process. Once the training process is complete, the neural network can be used to adjust the thin film deposition process parameters.

[0077] In one embodiment, after the analytical model 614 has been trained, the analytical model 614 can be used to analyze the emission spectrum of a thin film. In particular, as described with respect to Figures 1 to 5 an excitation light from a radiation source can be used to irradiate the thin film. The thin film will then output emissions, such as photoluminescence photons or photoelectrons, depending on the situation. A radiation sensor can then sense the energy spectrum of the emissions. The radiation sensor can transmit a sensor signal to the analytical model 614. The analytical model 614 analyzes the spectral data and labels the spectrum as a class. The class can include various subclasses of contamination, non - contamination, or both contamination and non - contamination. If the class indicates contamination, the control system 612 can output an alert and stop the further thin film deposition process.

[0078] In one embodiment, the control system 612 includes processing resources 668, memory resources 670, and communication resources 672. The processing resources 668 can include one or more controllers or processors. The processing resources 668 are configured to execute software instructions, process data, make thin film deposition control decisions, perform signal processing, read data from memory, write data to memory, and perform other processing operations. The processing resources 668 can include physical processing resources 668 located at the site or facility of the thin film deposition system. The processing resources can include virtual processing resources 668 that are remote from the site of the thin film deposition system or the facility in which the thin film deposition system is located. The processing resources 668 can include cloud - based processing resources, which include processors and servers accessed via one or more cloud computing platforms.

[0079] In one embodiment, the memory resource 670 may include one or more computer-readable memories. The memory resource 670 is configured to store software instructions associated with the functions of the control system and its components, including but not limited to the analysis model 614. The memory resource 670 may store data associated with the functions of the control system 612 and its components. The data may include training set data 662, current process condition data, and any other data associated with the operation of the control system 612 or any of its components. The memory resource 670 may include physical memory resources located at the site or facility of the thin film deposition system 100. The memory resource may include virtual memory resources remote from the site or facility of the thin film deposition system 100. The memory resource 670 may include cloud-based memory resources accessed via one or more cloud computing platforms.

[0080] In one embodiment, the communication resource may include resources that enable the control system 612 to communicate with devices associated with the thin film deposition system 100. For example, the communication resource 672 may include wired and wireless communication resources that enable the control system 612 to receive sensor data associated with the thin film deposition system and control the devices of the thin film deposition system. The communication resource 672 may enable the control system 612 to control the various components of the thin film deposition system. The communication resource 672 may enable the control system 612 to communicate with remote systems. The communication resource 672 may include or facilitate communication via one or more networks, such as a wide area network, a wireless network, the Internet, or an intranet. The communication resource 672 may enable the components of the control system 612 to communicate with each other.

[0081] In one embodiment, the analysis model 614 is implemented via the processing resource 668, the memory resource 670, and the communication resource 672. The control system 612 may be a distributed control system, and the components, resources, and locations of the distributed control system are remote from each other and from the thin film deposition system.

[0082] The components, functions, and processes described with respect to the control system 612 and the analysis model 614 may be extended to the Figures 1 to 5 described control system and analysis model.

[0083] Figure 7 is a flowchart of a process 700 for training an analysis model to determine whether a thin film is contaminated according to one embodiment. The process 700 may be implemented using the Figures 1 to 6 described systems, components, and processes. The process 700 may also be implemented using other systems, components, and processes. An example of the analysis model is the Figure 6 analysis model 614, but Figure 7 the process 700 ofFigures 1 to 6 The described components, processes, and technologies. Thus, reference Figures 1 to 6 description Figure 7 .

[0084] At 702, process 700 collects training set data that includes historical thin film spectral data and contamination label data. An example of the training set data is Figure 6 the training set data 662 of. This can be done by using a data mining system or process. The data mining system or process can collect the training set data by accessing one or more databases associated with the thin film deposition system and collecting and organizing various types of data contained in the one or more databases. The data mining system or process or another system or process can process and format the collected data to generate the training set data. The training set data can include historical thin film spectral data and contamination label data as described with respect to Figure 6 above.

[0085] At 704, process 700 inputs the historical thin film spectral data into the analysis model. In one example, this can include inputting the contamination label data into the analysis model using a training module as described with respect to Figure 6 above. An example of the contamination label data is Figure 6 the contamination label data 666 of. The historical thin film spectral data can be provided to the analysis model in the form of consecutive discrete sets. An example of the historical thin film data is Figure 6 the historical thin film data 664 of. Each discrete set can correspond to a single thin film or group of thin films. The historical thin film spectral data can be provided to the analysis model as a vector. Each set can include one or more vectors that are formatted for the analysis model to receive and process. The historical thin film spectral data can be provided to the analysis model in other formats without departing from the scope of the present invention.

[0086] At 706, process 700 generates predicted class data based on the historical thin film spectral data. In particular, the analysis model generates predicted class data for each set of the historical thin film spectral data. The predicted class data corresponds to a predicted class of contaminated or non - contaminated. This can be performed by Figure 6 the training module 660 or the analysis model 614 of.

[0087] At 708, the predicted class data is compared with the historical thin film spectral data 664. In particular, the predicted class data for each set of historical thin film spectral data is compared with the contamination label data associated with that set of historical thin film spectral data. This comparison can produce an error function that indicates the degree of match between the predicted class data and the contamination label data. This comparison is performed for each set of predicted class data. In one embodiment, the process can include generating an overall error function or indication that indicates the overall comparison of the predicted class data with the contamination label data. These comparisons can be performed by a training module or an analysis model. Without departing from the scope of the present invention, the comparison can include other types of functions or data in addition to those described above. This can be performed by Figure 6 the training module 660 or the analysis model 614 of

[0088] At 710, process 700 determines whether the predicted class data matches the contamination label data based on the comparison generated at step 708. In one example, if the overall error function is less than the error tolerance, process 700 determines that the predicted class data does not match the error tolerance. In one example, if the overall error function is greater than the error tolerance, process 700 determines that the predicted class data does indeed match the contamination label data. In one example, the error tolerance can include a tolerance between 0.1 and 0. In other words, if the total percentage error is less than 0.1 or 10%, process 700 considers the predicted class data to match the contamination label data. If the total percentage error is greater than 0.1 or 10%, process 700 considers the predicted class data not to match the contamination label data. Without departing from the scope of the present invention, other tolerance ranges can be utilized. The error score can be calculated in a variety of ways without departing from the scope of the present invention. The training module or the analysis model can make the determination associated with process step 710. This can be performed by Figure 6 the training module 660 or the analysis model 614 of

[0089] In one embodiment, if the predicted class data at step 710 does not match the historical thin film spectral data, the process proceeds to step 712. At step 712, process 700 adjusts the internal function associated with the analysis model. In one example, the training module adjusts the internal function associated with the analysis model. From step 712, the process returns to step 704. At step 704, the historical thin film spectral data is provided to the analysis model again. Since the internal function of the analysis model has been adjusted, the analysis model will generate predicted class data different from the previous cycle. The process proceeds to steps 706, 708, and 710, and the total error is calculated. If the predicted class data does not match the contamination label data, the process returns to step 712, and the internal function of the analysis model is adjusted again. This process iterates until the analysis model generates predicted class data that matches the contamination label data. This can be performed by Figure 6 the training module 660 or the analysis model 614.

[0090] In one embodiment, if the predicted class data matches the historical contamination label data and process step 710, process 700 proceeds to 714. At step 714, the training is completed. Now the analysis model is ready to be utilized to identify process conditions that can be utilized in the thin film deposition process performed by the thin film deposition system. Process 700 may include other steps or arrangements of steps other than those shown and described herein without departing from the scope of the present invention.

[0091] Figure 8 is a flowchart of a thin film deposition method 800 according to one embodiment. Method 800 can be implemented using the systems, components, and processes described with respect to Figures 1 to 7 Method 800 can also be implemented using other systems, components, and processes. At 802, method 800 includes depositing a thin film on a patterned wafer in a thin film deposition chamber. One example of a wafer is Figure 1 wafer 106. One example of a thin film deposition chamber is Figure 1 thin film deposition chamber 102. At 804, method 800 includes irradiating the thin film with a laser in the thin film deposition chamber. One example of a laser is Figure 2 excitation laser 208. At 806, method 800 includes collecting photoluminescence spectral data in response to irradiation with the laser using a light sensor. One example of a light sensor is Figure 2 light sensor 210. At 808, method 800 includes analyzing the spectrum using an analysis model trained by a machine learning process. One example of an analysis model is Figure 2The analysis model 214. At 810, method 800 includes determining whether there is an oxygen leak based on a spectrum using the analysis model. At 812, method 800 includes, if there is an oxygen leak, outputting a signal indicating the oxygen leak.

[0092] Figure 9 is a flowchart of a thin - film deposition method 900 according to an embodiment. Method 900 can be implemented using the systems, components, and processes described with respect to Figures 1 to 8 The method 900 can also be implemented using other systems, components, and processes. At 902, method 900 includes depositing a thin film on a patterned wafer in a first chamber. An example of the wafer is Figure 3 the wafer 306. An example of the first chamber is Figure 3 the thin - film deposition chamber 302. At 903, method 900 includes transferring the wafer from the first chamber to a second chamber. An example of the second chamber is Figure 3 the thin - film analysis chamber 334. At 904, method 900 includes irradiating the thin film with a laser in the second chamber. An example of the laser is Figure 2 the excitation laser 208. At 906, method 900 includes collecting photoluminescence spectral data in response to the irradiation with the laser using a light sensor. An example of the light sensor is Figure 2 the light sensor 210. At 908, method 900 includes analyzing the spectrum using an analysis model trained through a machine - learning process. An example of the analysis model is Figure 3 the analysis model 314. At 910, method 900 includes determining whether there is an oxygen leak based on the spectrum using the analysis model. At 912, method 900 includes: if there is an oxygen leak, outputting a signal indicating the oxygen leak.

[0093] Figure 10 is a flowchart of a thin - film deposition method 1000 according to an embodiment. Method 1000 can be implemented using the systems, components, and processes described with respect to Figures 1 to 9 The method 1000 can also be implemented using other systems, components, and processes. At 1002, method 1000 includes depositing a thin film on a patterned wafer in a thin - film deposition chamber. An example of the wafer is Figure 1 the wafer 106. An example of the thin - film deposition chamber is Figure 1 the thin - film deposition chamber 102. At 1004, method 1000 includes irradiating the thin film with X - rays in the thin - film deposition chamber. At 1006, method 1000 includes collecting photoelectron spectral data in response to the irradiation with X - rays using a photoelectron sensor. An example of the photoelectron sensor is Figure 1The emission sensor 110. At 1008, method 1000 includes analyzing a spectrum using an analytical model trained through a machine learning process. An example of the analytical model is Figure 1 The analytical model 114. At 1010, method 1000 includes determining whether there is an oxygen leak based on the spectrum using the analytical model. At 1012, method 1000 includes, if there is an oxygen leak, outputting a signal indicating the oxygen leak.

[0094] Figure 11 Is a flowchart of a thin film deposition method 1100 according to an embodiment. The method 1100 can be implemented using the systems, components, and processes described with respect to Figures 1 to 10 The method 1100 can also be implemented using other systems, components, and processes. At 1102, method 1100 includes depositing a thin film on a patterned wafer in a first chamber. An example of the wafer is Figure 3 The wafer 306. An example of the first chamber is Figure 3 The thin film deposition chamber 302. At 1103, method 1100 includes transferring the wafer from the first chamber to a second chamber. An example of the second chamber is Figure 3 The thin film analysis chamber 334. At 1104, method 1100 includes irradiating the thin film with x-rays in the second chamber. At 1106, method 1100 includes collecting photoelectron spectroscopy data in response to the x-ray irradiation using a photoelectron sensor. An example of the photoelectron sensor is Figure 3 The emission sensor 310. At 1108, method 1100 includes analyzing the spectrum using an analytical model trained through a machine learning process. An example of the analytical model is Figure 3 The analytical model 314. At 1110, method 1100 includes determining whether there is an oxygen leak based on the spectrum using the analytical model. At 1112, method 1100 includes, if there is an oxygen leak, outputting a signal indicating the oxygen leak.

[0095] Figure 12 Is a flowchart of a thin film deposition method 1200 according to an embodiment. The method 1200 can be implemented using the systems, components, and processes described with respect to Figures 1 to 11 The method 1200 can also be implemented using other systems, components, and processes. In one embodiment, at 1202, method 1200 includes depositing a thin film on a wafer in a thin film deposition chamber. An example of the thin film deposition chamber is Figure 1The thin film deposition chamber 102. At 1204, method 1200 includes irradiating the thin film with excitation light. At 1206, method 1200 includes detecting the emission spectrum from the thin film in response to the excitation light. At 1208, method 1200 includes detecting contamination of the thin film by analyzing the emission spectrum using an analysis model of the control system. An example of the analysis model is Figure 6 The analysis model 614. An example of the control system is Figure 6 The control system 612.

[0096] Figure 13 Is a flowchart of a thin film deposition method 1300 according to an embodiment. Method 1300 can be implemented using the systems, components, and processes described with respect to Figures 1 to 12 Method 1300 can also be implemented using other systems, components, and processes. At 1302, method 1300 includes training an analysis model to detect contamination of the thin film using a machine learning process that utilizes data from a plurality of spectra detected under a plurality of contaminated and non - contaminated conditions. An example of the analysis model is Figure 6 The analysis model 614. At 1304, method 1300 includes depositing a thin film on a wafer in a thin film deposition chamber. At 1306, method 1300 includes irradiating the wafer with excitation light. An example of the thin film deposition chamber is Figure 1 The thin film deposition chamber 102. At 1308, method 1300 includes detecting the emission spectrum from the wafer in response to the excitation light. At 1310, method 1300 includes detecting whether the thin film is contaminated by analyzing the emission spectrum using the analysis model.

[0097] In one embodiment, a method includes depositing a thin film on a wafer in a thin film deposition chamber and irradiating the thin film with excitation light. The method includes detecting the emission spectrum from the thin film in response to the excitation light, and detecting contamination of the thin film by analyzing the emission spectrum using an analysis model of the control system.

[0098] In the above - mentioned method, it further includes irradiating the thin film with the excitation light in the thin film deposition chamber.

[0099] In the above - mentioned method, it further includes: transferring the wafer from the thin film deposition chamber to a detection chamber; and irradiating the thin film with the excitation light in the detection chamber.

[0100] In the above - mentioned method, it further includes: transferring the wafer from the thin film deposition chamber to a detection chamber; and irradiating the thin film with the excitation light in the detection chamber; when irradiating the thin film with the excitation light, maintaining the vacuum condition of the thin film deposition chamber in the detection chamber.

[0101] In the above method, detecting contamination includes detecting oxidation of the thin film.

[0102] In the above method, detecting contamination includes detecting oxidation of the thin film, and the method further includes: using the control system to detect a leak in the thin film deposition chamber based on detecting oxidation of the thin film.

[0103] In the above method, it further includes stopping the operation of the thin film deposition chamber in response to detecting contamination of the thin film.

[0104] In the above method, the excitation light includes ultraviolet light, and detecting the emission spectrum includes: detecting a photoluminescence spectrum.

[0105] In the above method, the excitation light includes X-ray light, and detecting the emission spectrum includes: detecting a photoelectron spectrum.

[0106] In the above method, it further includes training the analysis model using a machine learning process to detect contamination of the thin film.

[0107] In one embodiment, a system includes: a thin film deposition chamber configured to deposit a thin film on a wafer; and a radiation source configured to irradiate the thin film with excitation light. The system includes an emission sensor configured to detect an emission spectrum from the wafer in response to the excitation light. The system includes a control system coupled to the radiation source and the emission sensor and configured to detect contamination of the thin film by analyzing the spectrum of photons and to stop the thin film deposition process in the thin film deposition chamber in response to detecting contamination of the thin film.

[0108] In the above system, the radiation source is a laser, and the emission sensor is an ultraviolet light sensor.

[0109] In the above system, the radiation source is an X-ray source, and the emission sensor is a photoelectron spectroscopy detector.

[0110] In the above system, the radiation source is positioned to irradiate the thin film in the thin film deposition chamber.

[0111] In the above system, it further includes a thin film analysis chamber, and the radiation source is configured to irradiate the thin film in the thin film analysis chamber.

[0112] In the above system, the control system includes an analysis model trained using a machine learning process to detect contamination of the thin film based on the emission spectrum.

[0113] In the above system, the control system includes an analysis model trained using a machine learning process to detect contamination of the thin film based on the emission spectrum, and the analysis model includes a neural network.

[0114] In one embodiment, a method includes training an analysis model using a machine learning process to detect contamination of a thin film, the machine learning process utilizing data from a plurality of spectra detected under a plurality of contaminated and non-contaminated conditions. The method includes depositing a thin film on a wafer in a thin film deposition chamber and irradiating the wafer with excitation light. The method includes detecting an emission spectrum from the wafer in response to the excitation light and detecting whether the thin film is contaminated by analyzing the emission spectrum using the analysis model.

[0115] In the above method, the thin film includes titanium nitride, and detecting contamination of the titanium nitride includes: detecting that the emission spectrum includes emission characteristics of titanium oxide.

[0116] In the above method, the thin film includes titanium nitride, and detecting contamination of the titanium nitride includes: detecting that the emission spectrum includes emission characteristics of titanium oxide, and the method further includes depositing the thin film over a semiconductor nanosheet structure.

[0117] Embodiments of the present invention are capable of in-situ detecting contamination of a thin film. Once contamination is detected, the thin film deposition process can be stopped, thereby reducing the number of wafers with defective thin films. This provides many benefits, including increased wafer yield and reduced scrapped wafers. Additionally, oxygen leakage in the deposition chamber can be detected based on the detected contamination.

[0118] The various embodiments described above can be combined to provide other embodiments. All U.S. patent application publications and U.S. patent applications mentioned in this specification and / or listed in the application data sheet are incorporated herein by reference. If desired, aspects of the embodiments can be modified to incorporate concepts from the various patents, applications, and publications to provide other embodiments.

[0119] These and other changes can be made to the embodiments in light of the above detailed description. Generally, in the following claims, the terms used should not be construed as limiting the claims to the specific embodiments disclosed in the specification and claims, but should be construed to include all possible embodiments and the full scope of equivalents. Thus, the claims are not limited by the disclosure.

Claims

1. A method for detecting contamination of a thin film, comprising: Depositing a thin film on a wafer in a thin film deposition chamber, wherein the thin film comprises titanium nitride. After depositing the thin film, an emissive sensor disposed in a tube is located in the thin film deposition chamber, and the tube extends from the outside of the thin film deposition chamber to the inside of the thin film deposition chamber; Outputting excitation light by an excitation laser disposed in the tube, and the excitation light enters the internal volume of the thin film deposition chamber from the tube to irradiate the thin film; Detecting, by the emissive sensor disposed in the tube, an emission spectrum of titanium oxide from the thin film in response to the excitation light; and Detecting a leak in the thin film deposition chamber by analyzing the emission spectrum of titanium oxide using an analysis model of a control system.

2. The method according to claim 1, wherein, An end of the tube is transparent to allow the excitation light to pass through.

3. The method according to claim 1, further comprising: Transferring the wafer from the thin film deposition chamber to a detection chamber; And Irradiating the thin film with the excitation light in the detection chamber.

4. The method according to claim 3, further comprising maintaining a vacuum condition of the thin film deposition chamber in the detection chamber when irradiating the thin film with the excitation light.

5. The method according to claim 1, wherein The analysis model includes a neural network.

6. The method according to claim 1, wherein, The thin film is deposited above a nanostructure of the wafer.

7. The method according to claim 1, further comprising stopping the operation of the thin film deposition chamber in response to detecting a leak in the thin film deposition chamber.

8. The method according to claim 1, wherein, The excitation light includes ultraviolet light, and detecting the emission spectrum includes: detecting a photoluminescence spectrum.

9. The method according to claim 1, wherein The excitation light includes X-ray light, and detecting the emission spectrum includes: detecting a photoelectron spectrum.

10. The method according to claim 1, further comprising training the analysis model using a machine learning process to detect a leak in the thin film deposition chamber.

11. A system for detecting contamination of a thin film, comprising: A thin film deposition chamber configured to deposit a thin film on a wafer, the thin film comprising titanium nitride; A tube, the interior of which includes: A radiation source configured to output excitation light, and the excitation light enters the internal volume of the thin film deposition chamber from the tube to irradiate the thin film; An emissive sensor configured to detect an emission spectrum of titanium oxide from the wafer in response to the excitation light, wherein the tube is configured to partially extend into the thin film deposition chamber such that the radiation source is outside the thin film deposition chamber and the emissive sensor is inside the thin film deposition chamber; and A control system coupled to the radiation source and the emissive sensor and configured to detect a leak in the thin film deposition chamber by analyzing the emission spectrum of titanium oxide and to stop a thin film deposition process in the thin film deposition chamber in response to detecting a leak in the thin film deposition chamber.

12. The system according to claim 11, wherein, The radiation source is a laser, and the emissive sensor is an ultraviolet light sensor.

13. The system according to claim 11, wherein, The radiation source is an X-ray source, and the emissive sensor is a photoelectron spectroscopy detector.

14. The system according to claim 11, wherein, The radiation source is positioned to irradiate the thin film in the thin film deposition chamber.

15. The system according to claim 11 further includes a thin film analysis chamber, wherein, The end of the tube is transparent to allow the excitation light to pass through.

16. The system according to claim 11, wherein, The control system includes an analysis model trained by a machine learning process to detect leaks in the thin film deposition chamber based on the emission spectrum of titanium oxide.

17. The system according to claim 16, wherein, The analysis model includes a neural network.

18. A method for detecting contamination of a thin film, comprising: Training an analysis model by a machine learning process to detect contamination of a thin film, the machine learning process using data from a plurality of spectra detected under a plurality of contaminated and non-contaminated conditions; Depositing a thin film on a wafer in a thin film deposition chamber, the thin film including titanium nitride; Irradiating the wafer with excitation light using a radiation source outside the thin film deposition chamber, wherein the radiation source is located in a tube that extends from the outside of the thin film deposition chamber to the inside of the thin film deposition chamber; Detecting, using an emitter sensor located in the tube and inside the thin film deposition chamber, the emission spectrum of titanium oxide from the wafer in response to the excitation light; and Detecting leaks in the thin film deposition chamber by analyzing the emission spectrum of titanium oxide using the analysis model.

19. The method according to claim 18, wherein The radiation source is a laser.

20. The method according to claim 18, further comprising depositing the thin film over a semiconductor nanosheet structure.

Images