Method, control device and storage medium

The method and control device address the imprecision in controlling multiple actuators by collectively managing actuator changes, ensuring precise and optimal process outcomes despite actuator limitations and time delays.

DE102020124935B4Active Publication Date: 2024-11-07VON ARDENNE ASSET GMBH & CO KG
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
DE102020124935
Authority / Receiving Office
DE · DE
Patent Type
Patents
Current Assignee / Owner
Filing Date
2020-09-24
Publication Date
2024-11-07
Estimated Expiration
2040-09-24

AI Technical Summary

Technical Problem

Conventional methods of controlling multiple actuators in coating processes face challenges due to the risk of imprecise activation and superimposed actions, leading to difficulties in accurately attributing measured values to individual actuators, especially when actuators reach their control limits, resulting in complex interplay of boundaries that are hard to manage.

Method used

A method and control device that consider the changes of all actuators together to identify which actuators can compensate for limitations, adjusting 'free' actuators to achieve the best possible process outcome under restrictions, using a control device with a processor and storage medium to implement this approach.

Benefits of technology

Enhances the precision of actuator control, ensuring that the process achieves optimal results even when actuators approach their limits, by accounting for time delays and interdependencies between control interventions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000000_0000_ABST
    Figure 00000000_0000_ABST
Patent Text Reader

Abstract

Method (100) for operating a process by means of which a substrate is processed, comprising the method (100): • Determining a first digit output (Δy) u ) based on a recorded (101) result of the process, • where the first digit output (Δy) u ) each actuator of a group of actuators (858, 862) by means of which the process is controlled (103), has a component which is assigned to the actuator; • Determining a second digit output (Δy) based on the first digit output (Δy u ) and based on a characteristic value Φ, which is a function of each component of the first position output (Δy u ) is; • Controlling the group of actuators (858, 862) according to the second actuator output (Δy).
Need to check novelty before this filing date? Find Prior Art

Description

[0001] Various embodiments relate to a method, a control device and a storage medium.

[0002] In general, a substrate, such as a glass substrate, a metal substrate, and / or a polymer substrate, can be processed, for example, coated, so that its chemical and / or physical properties can be modified. A coating system allows a substrate to be processed in a sequence of processes. These processes could include, for example, pretreatment, one or more coating processes, and / or a tempering step. Various coating methods can be used to coat a substrate. For example, a vacuum coating system can be used to deposit one or more layers onto a substrate or multiple substrates using chemical and / or physical vapor deposition. See DE 10 2012 109 093 A1 for further information.

[0003] Various processes can be controlled with spatial resolution. For example, the spatial distribution with which gas is supplied to the process can be controlled. Similarly, the spatial distribution of a magnetic field in which the process takes place can be controlled. Depending on the resolution of the spatial distribution, a large number of actuators may need to be controlled, such as multiple valves. Individual control of the actuators can be very complex. To simplify this process, the actuators are conventionally controlled either jointly or completely separately. See DE 10 2013 101 270 A1 for further information.

[0004] According to various embodiments, it has been recognized that conventional control of multiple actuators carries a significant risk of failure or at least imprecision. This can be particularly true when the controlled system superimposes the actions of several actuators, making it impossible to uniquely assign a measured value to a single actuator. This can cause the operating point of the process to diverge, for example, to a state where the actuator reaches a limit.

[0005] In the case of gas distribution, for example, a gas stream can be supplied at spatial intervals (e.g., every 10 centimeters), the rate of which can be varied by means of an actuator. The gas from each stream can spread and thus mix with other gas streams. The pressure resulting from this mixing can then no longer be clearly attributed to the influence of a single gas stream. For example, it is no longer possible to distinguish, based on the pressure alone, whether two gas streams are equal or whether one is zero and the other twice as large. The risk is particularly increased when an actuator has reached the limit of its adjustment range and therefore no longer responds. In this case, the controller figuratively attempts to repeatedly adjust the actuator without producing a response. The risk is also increased if the actuator cannot respond quickly enough to a required change.The more actuators there are, the more difficult it can become to reduce this risk.

[0006] For example, each actuator may be subject to an upper limit and a lower limit of its operating range, as well as a limitation on its rate of change. This complex interplay of limitations results in a high-dimensional space of boundary conditions and is therefore not easily and satisfactorily controlled. If, for instance, the actuation of each actuator is simply limited, the change achieved per actuation is often too small and deviates significantly from the change actually required.

[0007] According to various embodiments, a method, a control device, and a storage medium are provided, enabling more precise control of the actuators. The changes of all actuators are considered together (as a parameter) instead of separately, and based on this, it is determined which actuators are capable of compensating for the limitations that have occurred. If one or more of the actuators reach their respective limit, those actuators that have not reached their limit (also referred to as free actuators) can be adjusted in such a way as to achieve the best possible result of the process under the constraint.

[0008] It shows Fig. 1 a method according to different embodiments in a schematic control diagram; Fig. 2 and Fig. 3. The procedure according to different embodiments is presented in a schematic control diagram; Fig. 4, Fig. 5 and Fig. 6. The method according to different embodiments is shown in different schematic diagrams; Fig. 7 and Fig. 8 each the procedure according to different embodiments in a schematic process diagram; Fig. 9 a processing device according to various embodiments in a schematic diagram; and Fig. 10 and Fig. 11 each a vector space according to different embodiments in a schematic cross-sectional view.

[0009] The following detailed description refers to the accompanying drawings, which form part thereof and illustrate specific embodiments in which the invention can be implemented. In this context, directional terminology such as "top," "bottom," "front," "back," "anterior," "rear," etc., is used with reference to the orientation of the described figure(s). Since components of embodiments can be positioned in a number of different orientations, the directional terminology serves only for illustration and is in no way limiting. It is understood that other embodiments may be used and structural or logical modifications may be made without deviating from the scope of protection of the present invention.It is understood that the features of the various exemplary embodiments described herein can be combined with one another, unless specifically stated otherwise. The following detailed description is therefore not to be interpreted in a limiting sense, and the scope of protection of the present invention is defined by the appended claims.

[0010] Within the scope of this description, the terms "connected," "connected," and "coupled" are used to describe both direct and indirect connections (e.g., resistive and / or electrically conductive, such as an electrically conductive connection), direct or indirect connections, and direct or indirect couplings. In the figures, identical or similar elements are designated with identical reference numerals where appropriate.

[0011] Control can be understood as the intentional influencing of a system. The current state of the system (also called the actual state) can be changed according to a specification (also called the target state). Regulation can be understood as control, but with the additional benefit of counteracting changes in the system's state caused by disturbances. Visually, control can have a forward-directed control path and thus implement a sequential control system that converts an input variable (e.g., the specification) into an output variable. However, the control path can also be part of a control loop, thus implementing a regulation system. In contrast to purely forward-directed sequential control, regulation involves a continuous influence of the output variable on the input variable, which is effected by the control loop (also called feedback).In other words, control can be used as an alternative or additional to control, or alternatively, regulation can be used in addition to control. The state of the system (also called the operating point) can be represented by one or more controlled variables, whose actual value represents the current state of the system and whose setpoint (also called the reference value) represents the desired state of the system. In control, the current state of the system (e.g., determined based on a measurement) is compared with the desired state of the system, and the one or more controlled variables are influenced by a corresponding manipulated variable (using an actuator) in such a way that the deviation of the actual state from the desired state of the system is minimized. The manipulated variable (also called the variable to be controlled) can be the variable that is changed by controlling the actuator.

[0012] The system to be controlled can, for example, have one or more coating processes. The state of the system can thus represent the state of each coating process within the system, and vice versa. The current state of each coating process can be detected by means of a measuring element (e.g., having one or more sensors). For each controlled variable, one or more sensors can be used, configured to detect a quantity (also called a measured variable) that represents the controlled variable or that is the controlled variable itself.

[0013] A sensor (also called a detector) can be understood as a transducer designed to detect a property of its environment (e.g., qualitative or quantitative) corresponding to the sensor type, such as a physical property, a chemical property, and / or a material composition. The measured quantity is the physical quantity (also called the controlled variable) to which the sensor applies. Depending on the complexity of the environment being measured, the sensor may be configured to distinguish between two or more discrete states of the measured quantity (also called a switching sensor) or to detect the measured quantity quantitatively. An example of a quantitatively detected measured quantity is a gas inflow rate (e.g., measured as a flow rate), the actual state of which can be converted into a measured value by the sensor.

[0014] Each sensor can be part of a measurement chain, which includes the necessary infrastructure (e.g., a processor, storage medium, and / or bus system, etc.). The measurement chain can be configured to control the corresponding sensor (e.g., gas sensor, pressure sensor, and / or voltage sensor), process its measured value as an input, and then provide an electrical signal as an output representing the measured input. For example, the output could indicate the measured value. The measurement chain can be implemented using a control device.

[0015] The following discussion refers, among other things, to reactive coating processes in which a chemical reaction occurs. For example, one or more of the coating processes may involve a reactive sputtering process, which is supplied with one or more reactive gases (e.g., m reactive gases) that form a chemical compound with the coating material (also referred to as the target material) that is deposited onto a substrate (where, for example, m = 1, m > 1, or m > 2). It should be understood that the coating process does not necessarily have to be reactive. The quantities that represent materials involved in the chemical reaction in a reactive coating process may represent corresponding materials that do not participate in a chemical reaction in a non-reactive coating process.Accordingly, what has been described for the reactive coating process can apply by analogy to a non-reactive coating process and vice versa.

[0016] In various embodiments, reference is made to physical vapor deposition (PVD) as an exemplary coating process, e.g., one involving a sputtering process. It can be understood that what is described for PVD can apply analogously to chemical vapor deposition (CVD). In contrast to CVD, in PVD a solid material is first converted into the gas phase (also referred to as gaseous phase or vapor), and a layer is formed using this gas phase. In PVD, the gas phase of the target material can optionally react chemically with a reactive gas to form a chemical compound, which is incorporated into the layer or forms it. Thus, in the chemical reaction of PVD, two or more materials are combined to form the chemical compound.

[0017] In chemical vapor deposition (CVD), a gaseous starting compound (also called precursor or reactant) is split into at least two reaction products, of which at least one is incorporated into the layer and optionally one reaction product is removed from the coating process as an excess (e.g., by means of a pump). Optionally, CVD can be carried out using a plasma in which the splitting of the precursor takes place.

[0018] A plasma can be generated using a so-called working gas (also known as a plasma-forming gas). Depending on the specific design, the working gas can be a gaseous material that is inert, meaning it participates in few or no chemical reactions. A working gas can be defined by, or adapted to, the target material used. For example, a working gas can be a gas or a gas mixture that does not react with the target material to form a solid. The working gas can be, for example, a noble gas (e.g., helium, neon, argon, krypton, xenon, radon) or several noble gases. The plasma can be generated from the working gas, which essentially causes the target material to atomize. If a reactive gas is used, it can have a higher chemical reactivity than the working gas, e.g.,Regarding the target material: In other words, the atomized target material can react faster (i.e., form more reaction product per unit time) together with the reactive gas (if present) than together with the working gas (e.g., if it reacts chemically with the working gas at all). The reactive gas and the working gas can be supplied together or separately as a process gas (e.g., as a gas mixture), for example, via the gas supply device.

[0019] The reactive gas can be a gaseous material that reacts with the target material (e.g., with the atomized target material) and / or can be incorporated into the deposited layer via a chemical reaction. For example, if a target material is used that can form a nitride (e.g., aluminum nitride), or if a nitride of the target material is to be deposited, the reactive gas can contain nitrogen or be composed of nitrogen. If, for example, a target material is used that can form an oxide (e.g., aluminum oxide), or if an oxide of the target material is to be deposited, the reactive gas can contain oxygen or be composed of oxygen. The reactive gas can, for example, be a gas mixture (reactive gas mixture) of several gases that react with the target material and / or the deposited layer, e.g., oxygen and nitrogen, for example, when an oxynitride (e.g.,The reactive gas mixture can, for example, consist predominantly (i.e., more than 50%) of oxygen, for example, to deposit an oxide or oxynitride. Examples of reactive gases include: molecular oxygen, molecular nitrogen, nitric oxide, carbon dioxide, hydrogen sulfide, methane, gaseous hydrocarbons, molecular fluorine, molecular chlorine, ozone, or other gaseous materials.

[0020] The following refers to a method that involves controlling one or more actuators (also known as actuator control). In this context, reference is also made to a control device or code segments. The control device (also referred to as a control unit) can be configured to implement one or more of the methods described herein. For this purpose, the control device can include a processor configured to implement the respective method. For example, the processor can be configured to output corresponding control instructions. Alternatively or additionally, the processor can be configured to receive and process corresponding instructions and signals. The instructions received by the processor can, for example, be implemented using code segments stored in non-volatile memory.For example, the code segments can contain at least instructions and / or one or more specifications which, when executed by the processor, cause the processor to perform one of the procedures.

[0021] The term "control device" can be understood as any type of logic-implementing entity that may, for example, have a processor (and corresponding circuitry) capable of executing software stored in a memory medium, firmware, or a combination thereof, and issuing instructions based on that software. The control device may be configured, for example, using code segments (e.g., software). The control device may, for example, be a programmable logic controller (PLC) or be derived from one.

[0022] Depending on the specific embodiment, a data storage device (more generally referred to as a storage medium) can be a non-volatile data storage device. The data storage device can, for example, comprise a hard disk and / or at least one semiconductor memory (such as read-only memory, random access memory, and / or flash memory). The read-only memory can, for example, be a erasable programmable read-only memory (which may also be referred to as an EPROM). The random access memory can be a non-volatile random access memory (which may also be referred to as NVRAM). For example, the data storage device can contain one or more of the following: the code segments representing the process, one or more control elements, one or more tuning parameters, one or more model parameters, or one or more calibration parameters.

[0023] The term "processor" can be understood as any type of entity that allows the processing of data or signals. The data or signals can, for example, be processed according to at least one (i.e., one or more than one) specific function performed by the processor. A processor can be an analog circuit, a digital circuit, a mixed-signal circuit, a logic circuit, a microprocessor, a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), a programmable logic gate array (FPGA), an integrated circuit, or any combination thereof.Any other type of implementation of the respective functions, which are described in more detail below, can also be understood as a processor or logic circuit, for example, virtual processors (or a virtual machine) or a multitude of decentralized processors that are interconnected via a network, arbitrarily geographically distributed, and / or have arbitrary contributions to the implementation of the respective functions (e.g., workload distribution among the processors). The same generally applies to any differently implemented logic for implementing the respective functions. It is understood that one or more of the process steps described in detail herein can be executed (e.g., realized) by a processor through one or more specific functions executed by the processor.

[0024] In this context, a "controller" can be understood as a processor that implements the functions for controlling a process, for example, by converting information about the current state of a process into signals for controlling that process. These signals can then be used, for instance, to control one or more actuators that supply the process. The processor can be configured as a controller using one or more control elements (e.g., via software), or it can provide at least some of the control functions in a hard-coded manner. For example, a control device can incorporate a processor configured as a controller. Therefore, it can be understood that what has been explained regarding the controller can apply analogously to one or more control elements, and vice versa.

[0025] At least one control element (also called a correction element) can be configured to take into account a time interval (also called substrate runtime) between the processing of the substrate by a process and the detection of the substrate. The correction element can, for example, be configured to consider the duration of this time interval or any interventions occurring within that time interval.

[0026] This section explains several control elements in a more easily understandable parallel structure. For example, the control elements can operate independently of each other. It can also be understood that the explained control elements can be implemented together (e.g., in a so-called series structure), for example, using a common transfer function.

[0027] The term "actuator" (e.g., comprising an actuator) can be understood as a transducer designed to influence a state, a process (e.g., a coating process), or a device in response to a control signal. The actuator can convert an input signal (by means of which the control is effected) into mechanical movements or changes in physical quantities such as pressure or temperature. An electromechanical actuator, for example, can be configured to convert electrical energy into mechanical energy (e.g., through movement) in response to a control signal.

[0028] An actuator can be configured to influence the current state (also referred to as the operating point) of the process (e.g., its manipulated variable) that is supplied by the actuator. This influence can be direct or indirect. The manipulated variable and the controlled variable can, for example, differ from each other. The controlled variable (e.g., pressure) can then be a function of one or more manipulated variables (e.g., gas flow).

[0029] For example, the actuator can change an electrical voltage as the manipulated variable, which supplies a plasma, so that a atomization rate or a layer thickness is consequently changed as the controlled variable. For example, the actuator can change a gas flow rate as the manipulated variable, so that a pressure is consequently changed as the controlled variable. For example, the actuator can change a working gas flow rate as the manipulated variable, so that a atomization rate or a layer thickness is consequently changed as the controlled variable. For example, the actuator can change a reactive gas flow rate as the manipulated variable, so that a chemical composition of the plasma or a coating is consequently changed as the controlled variable.

[0030] Examples of actuator components include: a voltage source (if present), a valve (e.g., of a pump assembly and / or gas supply device), a motor (e.g., of a valve or pump), or similar components. The voltage source may, for example, be configured to generate one or more voltages according to a control signal and supply them to the coating process. The pump assembly may, for example, be configured to pump one or more gases according to a control signal, thereby removing them from the coating process. The gas supply device may, for example, be configured to supply one or more gases to the coating process according to a control signal (also referred to as gas supply). The supplied gas may have a gas flow rate (i.e., gas flow per unit of time). The gas flow may, for example, be a standard volumetric flow rate supplied to the coating process.

[0031] Regarding the control of an actuator, reference is made, among other things, to the more easily understood manipulated variable or its manipulated value, which is influenced by the actuator. The descriptions can be applied analogously to the controlled variable or its manipulated value, which is supplied to the actuator for control, and vice versa. Intuitively, the actuator acts as a converter that transforms a control signal into the manipulated variable or its manipulated value, so that the manipulated value is a function of the manipulated value. Modern actuators are, for example, provided as complex assemblies that include an actuator and their own control device (also called the actuator control device). The actuator control device can be configured to receive the manipulated value as input via the control signal and to control the actuator according to the manipulated value.The generation and transmission of the control variable then takes place within the actuator, so that the manipulated variable is supplied to the actuator. Less complex actuators can only process the controlled variable as a control signal, so the controlled variable is supplied to them for control purposes.

[0032] More generally, an actuator can be controlled by means of a control signal, where the control signal can represent the manipulated variable or its setpoint and / or the controlled variable or its control value. Alternatively or additionally, the control signal can contain instructions that specify how the manipulated variable or its setpoint is to be changed (e.g., its relative change).

[0033] The term "process parameter" (also referred to as operating point parameter) describes a physical quantity of a process, e.g., representing its operating point and / or a controlled variable of the process. The process parameter can be measurable, but does not necessarily have to be. For example, the process parameter may be a quantity that is difficult to measure, e.g., not without interfering with the processing of a substrate by the process. According to various embodiments, the process parameter can be a computationally determined quantity, e.g., indirectly determined based on a result of the process. Determining the process parameter can, for example, be based on at least one additional piece of information about the process, e.g., its duration. The process parameter could, for example, be a coating rate, which is determined based on the measured layer thickness and the known coating duration.

[0034] Fig. Figure 1 illustrates a method 100 according to various embodiments in a schematic control diagram. The method 100 can be configured to operate a process group that performs two or more processes (also referred to as coating processes), e.g., at least a first process and a second process. For example, the process group can perform several (e.g., N) processes (where, for example, N > 1, N > 5, or N > 10). For the sake of clarity, the following discussion of method 100 refers, among other things, to two coating processes as exemplary processes. It should be understood that what is described for the two coating processes can apply analogously to more than two processes or to processes of other types.

[0035] The process group can include one or more (e.g., N or more) processing devices, each configured to perform one or more processes. For example, the process group can be configured to process multiple substrates (e.g., N or more) in parallel, e.g., one substrate per processing device. If a processing device is configured to perform a coating process, it is also referred to as a coating device. The current state of the substrate can be used as the control variable(s) of the process group.

[0036] The various coating processes can be similar, for example, in their coating material, reactive gas, and / or the type of coating device (e.g., thermal or atomizing). Each coating process can be configured according to different embodiments to coat at least one substrate (i.e., one or more substrates).

[0037] Method 100 can include, in 101, the detection (also referred to simply as measurement 101) of a substrate 102, which has been processed by means of the first process 152a (also referred to simply as process step A or P1) and then by means of the second process 152b (also referred to simply as process step B or P2). By detecting the substrate 102, the actual state of the substrate (e.g., as a control variable of the process group) is determined.

[0038] The measuring device 101 can include sensing the substrate by means of a measuring element (e.g., having one or more sensors). The sensor(s) can be part of the measuring chain 112, which provides an electrical signal as an output variable 112o (hereinafter also referred to as the result 112o of the sensing) based on one or more measured variables as an input variable 112a. The measuring chain can, for example, be implemented by means of a control device 224.

[0039] Examples of the measured quantity include: the surface resistance of the substrate 102, the layer thickness of the substrate 102, the optical spectrum of the substrate (e.g., in transmission and / or reflection) 102, the transmittance of the substrate 102, and the reflectance of the substrate 102. The measured quantity can, for example, be related to a coating of the substrate. The reflectance or transmittance can be related to a reference wavelength, e.g., 400 nm or 500 nm. Examples of corresponding measuring elements include: a surface resistance measuring element (e.g., comprising one or more voltage or induction sensors) for measuring surface resistance, a layer thickness measuring element (e.g., comprising one or more interferometers) for measuring layer thickness, and an optoelectronic measuring element (e.g., comprising one or more CCD sensors) for detecting electromagnetic radiation.

[0040] The procedure 100 can include, in 103, rules of the process group by means of a first manipulated variable 114a of the first process and by means of a second manipulated variable 114b of the second process based on a result of the acquisition. The rule 103 can include controlling a first actuator, which is configured to influence the first manipulated variable, and controlling a second actuator, which is configured to influence the second manipulated variable.

[0041] The first process can be controlled (e.g., supplied) by the first actuator. More generally, the first actuator can be configured to influence the operating point of the first process, e.g., its first manipulated variable. The second process can be controlled (e.g., supplied) by the second actuator. More generally, the second actuator can be configured to influence the operating point of the second process, e.g., its second manipulated variable.

[0042] Examples of the first control variable 114a include: electrical power (e.g., supplying the first process), fluid flow rate (e.g., supplying the first process), pressure (e.g., to which the first process is subjected), a chemical composition supplied to or exposed to the first process, and a magnetic flux density to which the first process is exposed. The fluid flow rate can, for example, refer to a fluid (e.g., a gas) that is supplied to and / or removed from the first process.

[0043] Examples of the second control variable 114b include: electrical power (e.g., supplying the second process), fluid flow rate (e.g., supplying the second process), pressure (e.g., to which the second process is subjected), a chemical composition supplied to or exposed to the first process, and a magnetic flux density to which the first process is exposed. The fluid flow rate can, for example, refer to a fluid (e.g., a gas) supplied to and / or removed from the second process.

[0044] Controlling the process group using the first manipulated variable 114a can be based on a first correction element, which takes into account a first time interval 111a (also referred to as the first substrate runtime, first time delay 111a, or time delay 111a of the first process) between the processing 152a of the substrate by the first process and the measurement 102. The first time interval 111a can have a duration that begins with the end of the processing 152a of the substrate by the first process and ends with the measurement 102.

[0045] Controlling the process group using the second manipulated variable 114b can be based on a second correction element, which takes into account a second time interval 111b (also referred to as the second substrate runtime, second time delay 111b, or time delay 111a of the second process) between the processing 152b of the substrate by the first process and the measurement 102. The second time interval 111b can have a duration that begins with the end of the processing 152b of the substrate by the second process and ends with the measurement 102.

[0046] The first time interval 111a and the second time interval 111b can differ from each other, e.g., in their duration and / or their start time. The first time interval 111a and the second time interval 111b can coincide in their end time (the time of measurement 101).

[0047] Optionally, measurement 102 can generate a timestamp representing the time of measurement 102 (also referred to as the measurement time). Optionally, measurement 102 can also generate a substrate identifier representing the substrate that was detected at the measurement time.

[0048] The control device 224 can, for example, be configured to perform the control 103 (then also referred to simply as controller 224).

[0049] Fundamentally, the controller 224 can (but does not necessarily have to) be configured such that the control of 103 using the first control variable and the control of 103 using the second control variable are carried out separately and / or independently of each other, e.g., by means of so-called elementary controllers 224a, 224b (visually providing an individual control loop). Several elementary controllers 224a, 224b can, for example, operate independently of each other. For instance, the two elementary controllers can each be a dedicated component of the controller 224.

[0050] For example, each elementary controller can be set up to control exactly one process of the process group, and / or each process can be assigned to exactly one elementary controller which is set up to control the process.

[0051] A first elementary controller 224a can be configured to perform the control 103 using the first control variable, e.g., by controlling the first actuator. A second elementary controller 224b can be configured to perform the control 103 using the second control variable, e.g., by controlling the second actuator.

[0052] Such an elementary controller can be implemented using a programmable logic controller (PLC), which may even be able to implement several such elementary controllers. A so-called PID controller (proportional-integral-derivative controller) used for this purpose feeds back the control error using a P-element (proportional element), an I-element (integral element), and / or a D-element (derivative element). More generally, the elementary controller can have one or more control elements.

[0053] An exemplary implementation according to embodiment 100 is described below. The input variable for the control 103 is a dimensionless deviation from the target state, which is referred to here as the target state deviation x (e.g., a layer defect). If the target state deviation x = 0, the deviation from the target state is zero, or the actual state of the substrate is equal to the target state of the substrate (e.g., equal to the state of a reference product). If x ≠ 0, there is a deviation from the target state, e.g., a quantized deviation from the target state. The description for the target state deviation can also apply analogously to a differently defined input variable for the control 103.

[0054] The deviation from the target state can be determined, for example, based on the recorded actual state of the substrate and / or a process parameter derived from it, as will be described in more detail later.

[0055] Based on a coating rate as a process parameter, the determination of the target state deviation can be carried out, for example, according to the following relationship: x:=(RRRef−1), where x is the target state deviation, R is the determined actual coating rate, and R ref The target coating rate (e.g., which leads to the reference product) is referred to. In this case, the target state deviation corresponds to a relative target value deviation, e.g., the percentage deviation from the target state.

[0056] Based on a color index of the substrate as the measured quantity (also referred to as the measured quantity), the determination of the target state deviation can be carried out, for example, according to the following relationship: x:=c⋅(a*−aref*), where c is a constant, a* is the determined actual color index according to the Lab color space (also known as CIELAB) and aref* The target color index is defined according to the Lab color space. In this case, the target state deviation corresponds to an absolute target value deviation, e.g., the distance from the target state. The constant c can be freely chosen. It is advantageous to choose c such that x again corresponds to a percentage change in layer thickness. The same can apply analogously for another point on the a*b* coordinate plane or for another color space. This can be advantageous, for example, if the corresponding color index is only influenced by a single process within the process group. By controlling this process, the desired value of the color index can be stabilized. This allows for a uniform interpretation of target state deviations obtained through different methods.

[0057] Based on the refractive index of the substrate as the measured quantity, the target state deviation can represent an absolute refractive index deviation in analogy to the color index.

[0058] Just like the target state deviation, the control output Δy can also be written as a dimensionless quantity, which is then converted into a physical control variable.

[0059] The position output Δy is represented as a relative scaling or shift of a control value. For example, a relative change in process performance could be calculated from this: Pneu=Pist⋅expΔy

[0060] This means that the procedure 100 can, for example, also be applied to several related processes with different actual states.

[0061] In the case of the position output Δy, the rule can be based on a momentary change Δy. The following refers, among other things, to the more intuitive process setting y. What is described for the process setting y (also called the operating point) can be applied analogously to the position output Δy = f(y).

[0062] The following text also refers to a vector notation for the quantities used, where the vector components denote measured quantities, characteristic values, or manipulated variables used in parallel to one another. This facilitates understanding the case where the process group (e.g., its operating point) cannot be readily described by a single number, for example, when the control 103 is applied to several (different) setpoints. If, for instance, the measurement 101 involves capturing the substrate at several points in parallel, then instead of a one-dimensional (e.g., scalar) actual state of the substrate, one obtains a tuple of measured values ​​as the actual state (also referred to as a multi-component actual state) of the substrate, which can be written as components of a vector. Consequently, the deviation from the setpoint can also be represented as a vector. x→ (also called deviation vector) which consists of individual deviation values ​​x i , i = 1 ... N i (also referred to as error values).

[0063] The following explains the rules for a scalar deviation from the desired state using a single elementary controller. A generalization of the scalar deviation from the desired state to a vector deviation from the desired state will be explained later. What is described for the single elementary controller can be applied analogously to multiple elementary controllers.

[0064] Fig. Figure 2 illustrates the method 100 according to various embodiments 200 in a schematic control diagram, explaining an exemplary implementation (also referred to as feedback correction) of the correction element 202 using the first elementary controller 224a. It can be understood that what is described for the first elementary controller 224a can apply analogously to the second elementary controller 224b or to one or more additional elementary controllers 224b, to a differently configured control device 224 by means of which the method is carried out, or to a differently implemented method 100.

[0065] The first elementary controller 224a can be configured to process the measurement result 112o (also referred to as measurement result 112o). The measurement result 112o can, for example, include a first process parameter 212a, which is fed to the first elementary controller 224a. This process parameter can intuitively represent the virtual (e.g., difficult or impossible to measure) controlled variable of the corresponding process, which is not measured but determined based on the actual state of the substrate, as will be described in more detail later.

[0066] The elementary controller 124 can implement the correction element 202. Each elementary controller 124 can implement its own correction element 202. For example, two elementary controllers 124, or rather their correction elements 202, can differ from each other. Each correction element 202 can, for example, be assigned to (e.g., only) one process within the process group and be configured to take into account the time span of the process to which the correction element 202 is assigned. Accordingly, the correction element 202 can take into account the time delay of the process (also referred to as substrate runtime) to which the correction element 202 is assigned.

[0067] The elementary controller 124 can further implement one or more control elements 210, which are connected in parallel to the correction element 202 according to the feedback correction. The one or more control elements can be a proportional (P) control element, an integral (I) control element, and / or a derivative (D) control element, or be composed of these. For example, the one or more control elements can have two or more different control elements.

[0068] The output 222o of the correction element 202 (also referred to as correction or, in formulas, as Δx̂) can be superimposed on the measurement result 112o, which is supplied to the control element 210. This ensures that the measurement result 112o supplied to the control element 210, e.g., the first process characteristic 212a, is corrected based on the time delay of the process. As a result of the superimposition, a corrected measurement result 212o is provided, which is based on the measurement result 112o and the correction 222o. The corrected measurement result 212o can, for example, have a corrected process characteristic that is supplied to the control element 210.

[0069] The correction element 202 can be configured to provide the correction 222o based on the time delay of the process and based on an output of the control element(s) 210. For example, the correction element 202 can feed back the output of the control element(s) 210, taking into account the time delay of the process.

[0070] As an output of one or more than one control element 210 (which provides the position output of the feedback correction), a position value can be obtained according to which the first actuator, which is set up to influence the first position variable 114a of the first process, is controlled.

[0071] The following is an exemplary implementation according to embodiments 200.

[0072] The elementary controller 224a, for example, only has the integral element and therefore operates as a pure integral controller. This ensures that a deviation detected once leads to a permanent correction of the control output. The elementary controller 224a stores the output corrections 222o and the output time in order to take into account the time-of-flight effects between the control intervention 103 (also referred to as the control intervention) and the measurement 101 using a predictor (e.g., a Smith predictor).

[0073] Rule 103, for example, specifies: input of the target state deviation (from a previous calculation); a prediction (also referred to as the predictor output) which shows that the sum of the corrections already output but not yet effective for the measurement is taken into account in order to calculate a virtual target state deviation x' from the target state deviation x; determining an unlimited actuator output Δy uby multiplying the deviation vector by a proportional factor; checking the unlimited actuator output for violations of one or more limits (e.g., actuator limits and / or rate limit) to obtain the limited actuator output Δy; optionally applying a watchdog to detect instability using Δy and the configured rate limit; converting the dimensionless value Δy into a change to be performed on the operating point (e.g., a control variable). The virtual target state deviation x' intuitively represents the expected measured value after applying the corrections already made.

[0074] The following are exemplary implementations of models for determining the deviation from the target state according to embodiments 200.

[0075] Determining the deviation from the target state can, for example, involve using a spectral model (also known as singular value decomposition or SVD). The output of the singular value decomposition then directly corresponds to the deviation from the target state.

[0076] Determining the deviation from the target state can, for example, involve the use of so-called thin-film simulation software (e.g., featuring a neurocognitive model of numerical processing). The thin-film simulation software can implement a fit parameter f with a reference value f. ref to determine the deviation from the target state, for example according to the following relation: x:=ffref−1.

[0077] Determining the deviation from the target state can, for example, involve using a scalar model to determine the deviation from the target state, for example according to the following relation: x:=∑ci(ai−ai,ref), where the index i references arbitrary scalar parameters, such as a*, b*. For example, the deviation from the target state can be derived from a single parameter or from a linear combination of parameters.

[0078] Determining the deviation from the target state can, for example, involve using a resistance model, such as the following relation: x:=ρRsq1dref−1, where ρ is the measured specific resistance of the layer, d ref the target thickness of the layer and R sq denotes the target resistance of the layer.

[0079] Optionally, the target state deviation (e.g., deviation in layer thickness) can be scaled so that it corresponds to the relative deviation of the actual coating rate from the target state of the coating rate (also referred to as target coating rate).

[0080] Fig. Figure 3 illustrates the method 100 according to various embodiments 300 in a schematic control diagram, explaining an exemplary implementation of the correction element (also called parallel correction) using the first elementary controller 224a.

[0081] According to the parallel correction, the output 222o of the correction element 202 (also referred to as the correction) can be superimposed on the output of each control element 210 to obtain the actuator output. This ensures that the output provided by each control element 210 is corrected based on the time delay of the process. As a result 312o of the superposition, for example, a corrected output 312o is provided, which is based on the measurement result 112o and the correction 222o.

[0082] The result 312o of the superposition (which provides the position output of the parallel correction) can be a position value, according to which the first actuator, which is set up to influence the first position variable 114a of the first process, is controlled.

[0083] The correction element 202 can be configured to provide the correction 222o based on the time delay of the process and based on the output of the control element(s) 210. For example, the correction element 202 can correct the output of the control element(s) 210 taking into account the time delay of the process.

[0084] Fig. Figure 4 illustrates the method 100 according to various embodiments 400 in a schematic diagram, when the time delay is not taken into account; and Fig. Figure 5 illustrates the method 100 according to various embodiments 500 in a schematic diagram when the time delay is taken into account.

[0085] During the time interval between the processing of the substrate by the process and the detection of substrate 102 (also referred to as the time delay), several position interventions can occur, each position intervention having the control 101 of the process group (e.g., one or more processes of the process group). After the processing of the substrate, one or more position interventions can occur until the detection 101 of substrate 102 takes place.

[0086] For example, the frequency a / t=1 / ΔT of the position interventions (in number “a” per time “t”) can be greater than the reciprocal of the time delay T. n of the nth process (e.g., the first and / or the second process). Alternatively or additionally, a frequency a / t = 1 / ΔT with which a substrate is detected can be greater than the reciprocal of the time delay T. n of the nth process (e.g., the first and / or the second process).

[0087] The time interval ΔT here intuitively represents the time interval between immediately successive events in which a measurement or a control takes place (then also referred to as a measurement event or control intervention).

[0088] The processing of the substrate using these processes takes place in a defined temporal sequence, which results from the position s n of the nth process along the path (also called the transport path) along which substrate 102 is transported, and the velocity v (also called the transport velocity) at which substrate 102 is transported. For example, the first process (i.e., n=1) ends at t1=v·s1 and the second process (i.e., n=2) ends at t2=v·s2. The time interval ΔT can, for example, satisfy the relation ΔT≥t1-t2. For example, the regulating 103 and / or the measuring 101 can be performed with the clock 1 / ΔT according to which the substrates are transported.

[0089] The duration ΔT and / or the duration t1-t2 can, for example, be in a range from approximately 5 seconds to approximately 30 seconds.

[0090] For example, 1 / ΔT > p / T n , where the scalar p can be larger than 2, e.g. than 3, e.g. than 4, e.g. than 5, e.g. than 6, e.g. than 7, e.g. than 8, e.g. than 9, e.g. than 10, e.g. than 20, e.g. than 50, e.g. than 100, e.g. than 200, e.g. than 1000.

[0091] As can be seen, the time delay causes an iteratively delayed effect of the adjustment, which only becomes visible after the time delay has elapsed, whereby at this point ("now") further adjustments have already taken place, which subsequently lead to an overreaction.

[0092] As can be seen, Δy is used for each positional intervention. uNot only the measurement result x, but also a correction Δx̂ is taken into account. The correction Δx̂ can be a function of the time delay. For example, Δy u = K·(x + Δx̂). The same can apply analogously to the parallel correction, where Δy then u = K·x + Δx̂ is.

[0093] This ensures that the control intervention (also referred to as the control event) already takes into account that its effect will only be reflected in the measurement result after the time period has elapsed.

[0094] Fig. Figure 6 illustrates the method 100 according to various embodiments 600 in schematic diagrams that compare the results when the time delay is disregarded (in diagram 500) and when the time delay is taken into account (in diagram 600). Intuitively, taking the time delay into account allows for a prediction of the measurement result as a consequence of processing the substrate using the process.

[0095] Without taking the time delay into account (in diagram 500), the same deviation from the target state leads to the same change in the manipulated variable.

[0096] Taking the time delay into account (in diagram 600), the same deviation from the target state leads to different changes in the manipulated variable, e.g., as a function of what portion of the substrate runtime has already elapsed. This provides a visual record of the manipulated variable changes that have already taken place.

[0097] Further exemplary implementations of the correction element are explained below.

[0098] Fig. Figure 7 illustrates the process 100 according to various embodiments 700 in a schematic process diagram.

[0099] The process group can have several processing devices, of which the nth processing device is set up to carry out the nth process P. nto be carried out (1≤n≤N). Examples of a process include: a pretreatment, one or more coating processes and / or a tempering step.

[0100] The multiple processing devices can be arranged in a vacuum chamber 802. The substrate 102 can be transported along a path 111, past each of the multiple processing devices, at a velocity v.

[0101] The processing of the substrate using these processes takes place in a defined temporal sequence, which results from the position s n of the nth process along path 111 (also called the transport path) along which substrate 102 is transported, and the velocity v (also called the transport velocity) at which substrate 102 is transported. For example, the first process (i.e., n=1) at s1=t 1 / v, the second process (i.e., n=2) at s2=t2 / v and the last process (i.e., n=N) at sN =t N / v arranged.

[0102] After the substrate has been processed using the N processes of the process group, the substrate 102 is measured (also known as ex-situ measurement). This allows for the measurement of the resulting product properties, such as color and / or film resistance. Optionally, after the substrate 102 has been measured, it can be processed using one or more additional processes. For example, one or more measuring elements can record the current state of the substrate in a partially processed state using the N processes (also known as in-situ measurement).

[0103] Consequently, there is a time delay between the nth process and measurement 101. For example, in an architectural glass coating plant, this time delay can range from approximately 600 seconds (first process) to approximately 60 seconds (nth process). This could correspond, for example, to 10 processed substrates being treated by the first process before measurement 101 of the first of the 10 processed substrates takes place (illustratively, before the result of the first process becomes visible in the measurement result).

[0104] According to various embodiments, the control variables of the individual processes can be set using a control loop (also known as a closed-loop controller) in order to minimize the deviation of the actual state of the layer properties from a target state. For example, the layer thickness of the coating processes could be influenced to achieve a defined color appearance.

[0105] A closed loop (also known as a closed-loop system) eliminates the need for manual intervention. For example, the evaluation of measurement data, the calculation of appropriate corrective actions, and the execution of physical adjustments can be fully automated. While operator control and intervention remain optional, they are not strictly required.

[0106] According to various embodiments, method 100 achieves simultaneous control of all relevant processes with the shortest possible settling times. Depending on the process and actuator, the time delay (also called delay time) between the control intervention and the measurement of the product properties can vary and be taken into account.

[0107] Method 100 can be configured so that the controller's interventions can cooperate with other control systems. For example, an operator can manually make corrections, such as balancing the output of multiple coating sources. As a second example, another process automation system could simultaneously adjust the coating rate and substrate feed speed to maximize the instantaneous plant throughput.

[0108] Controlling 103 and / or measuring 101 can be performed discretely over time. For example, controlling 103 can involve controlling the actuator (also referred to as a positioning action) at intervals of t = ΔT. Alternatively or additionally, measuring 101 can be performed at intervals of t = ΔT. For example, each positioning action can be triggered by the input of the measurement result (intuitively, by the arrival of measurement data).

[0109] If the measurement 101 is performed at several points (for example, using multiple sensor stations) within the system, e.g., ex-situ and in-situ, the method described herein can be applied to each of these sensor stations as an independent sub-controller. Each sub-controller then operates in sync with the respective measurement.

[0110] In a first embodiment, the processing of a first substrate is carried out (or ends) by the first process P1 at time t1 and by the second process P2 at time t2. During the processing of the first substrate, the process group can be at a first operating point A1. Subsequently, the measurement 101 of the first substrate and the first control 103 take place at time t3. As a result of the control 103 at time t3, the process group, e.g., its first process P1 and its second process P2, is at a second operating point A2 after time t3. Immediately after the first control 103, in the first embodiment, the processing of a second substrate is carried out (or begins) by the first process P1 at time t4 and by the second process P2 at time t5. During the processing of the second substrate, the process group can be at the second operating point A2.The second substrate is then measured (101) and the second control (103) is performed at time t6. As a result of the control (103) at time t6, the process group, e.g., its first process P1 and its second process P2, is at a third operating point A3 after time t6.

[0111] As can be seen from the first embodiment, the result of the first rule 103 can only be determined at time t6. Additional substrates processed between t4 and t6 may therefore be rejected if operating point A2 does not meet the requirements.

[0112] In various applications, the current state of the substrate can exhibit several measured values ​​as components, each of which can be uniquely assigned to a process within the process group. This can apply, for example, to the case where the substrate is coated with a reflective layer and subsequently with a wavelength-filtering layer. In this case, the intensity and wavelength of the radiation reflected by the substrate can be recorded as components, with the intensity being attributable to the reflective layer and the wavelength to the wavelength-filtering layer. More generally, each component of the measurement result can only be influenced by one of the processes.

[0113] In other, different application scenarios, the substrate's current state can have multiple components, each of which can be influenced by more than one process (also referred to as the complex case). In this complex case, a clear assignment between component and process is not always readily possible. This can be the case, for example, when the substrate is coated with multiple layers of different refractive indices. The interaction of the multiple layers with each other can lead to a complex optical spectrum, which is completely altered if the operating point of just one process is changed.

[0114] To facilitate regulation in this complex case, the recorded actual state of the substrate can be converted into one or more process parameters, as described in more detail below.

[0115] If the number of layers is small, a numerical layer model can be varied until it approximately predicts the actual state of the substrate (e.g., the optical spectrum). Analogous to layer planning, the parameters of the layer model are varied using least-squares adjustment until the simulated state of the substrate (also referred to as the simulated substrate) and the actual state (e.g., measured optical properties) of the substrate correspond as closely as possible. The parameters of the layer model then correspond to the actual layers of the substrate, such as their thickness and / or refractive index, which are stored as process parameters 212a. Based on a comparison of the process parameters 212a with the target state of the substrate, it can be determined how the process group should be controlled. Control can then be performed based on the process parameters 212a as a measurement result.

[0116] However, the complexity increases considerably with the number of layers, so that even more than five layers lead to an almost insurmountable computational effort and a high risk that the simulated substrate will not match the actually manufactured substrate, i.e., the simulation will fail. This is due, among other things, to the aforementioned fact that there is a vast and difficult-to-manage variety of layer systems that correspond to a recorded actual state of the substrate (e.g., optical properties). Therefore, a large variety of convergence scenarios and, depending on the chosen ranking conditions, associated parameter sets may be required for the adjustment calculation to converge.

[0117] In this complex case, a mapping function (e.g., a functional relationship) can be determined that translates the recorded actual state of the substrate into process parameters. The mapping function can, for example, transform a deviation of the spectrum from a target spectrum into one or more process parameters per process. The mapping function can, for example, be the inverse (or pseudo-inverse) of a transformation that translates a change in process parameters into a change in the spectrum. The transformation can be determined, for example, based on a variational analysis or actual measurements. The variational analysis can, for example, use the numerical layer model.

[0118] More generally speaking, in complex applications, the recorded current state of the substrate is converted into several process parameters using one of the mechanisms described above, in order to draw conclusions about the individual process steps from the resulting product properties.

[0119] For example, a measured optical spectrum of the substrate can be translated into the thickness of individual layers or groups of layers as process parameters. Similarly, a resistance measurement of the substrate can be translated into the conductivity of a layer in the layer stack as a process parameter. Finally, an X-ray fluorescence (XRF) measurement can be translated into the mass coverage of multiple layers of different materials as a process parameter.

[0120] As a result, several process parameters are obtained, each of which is uniquely assigned to a process within the process group. This allows individual corrections (e.g., for the respective control variable) to be determined for each process.

[0121] In a second embodiment, the first process can involve the atomization of a material (also referred to as the target material) using a plasma (also referred to as sputtering), which is then used to form a layer on the substrate (also referred to as coating). The average coating rate can be proportionally varied, for example, by changing the electrical power supplied to the first process as a primary control variable. Alternatively or additionally, the lateral distribution of the coating rate can be modified by changing the spatial distribution of the plasma used for atomization.Examples of a control variable for influencing the spatial distribution of the plasma include: a spatial distribution with which a gas is supplied to the plasma (also called gas supply distribution); a spatial distribution of a magnetic field strength (also called magnetic field distribution) by means of which the formation of the plasma takes place.

[0122] In a third embodiment, the first process can involve the thermal evaporation of a material by means of an electron beam (also referred to as electron beam evaporation), which is then used to form a layer on the substrate (also referred to as coating). The average coating rate can be controlled by varying the beam power. The transverse distribution of the coating rate can be altered by changing the spatial distribution with which the material is irradiated by the electron beam (i.e., and thus the applied power density).

[0123] As described above, the control 103 can be achieved using several independently operating elementary controllers. Each of these elementary controllers, for example, can be assigned to only one process within the process group, whereby the elementary controller is supplied with one or more process parameters that are unique to that process (also referred to as individual control). Individual control means that it is sufficient for each elementary controller to consider only one time delay, namely the time delay between the measurement 103 and the processing 101 by the process assigned to the elementary controller.

[0124] Taking time delay into account can be implemented in various ways, two of which are explained below as examples.

[0125] Time delays can be accounted for using a Smith predictor, which, for example, maps to the process parameter. The Smith predictor then outputs the corrected process parameter 212o, which clearly reflects the expected effect of the interventions made in the meantime. Feedback correction can be implemented using the Smith predictor, for example.

[0126] Alternatively or additionally, taking time delays into account can involve continuously storing the temporal progression of the process group's operating point, e.g., its control variables (e.g., in a FIFO buffer). A FIFO buffer is designed to compensate for differences in processing speed and allows for the tolerance of a bottleneck in processing by temporarily storing the processing objects for later processing. For example, the control variables used in control can be continuously stored. In this way, past control variable values ​​and change times are stored in the FIFO buffer. With each measurement (also referred to as a measurement event), a correction is then calculated based on the control variable value that was valid when the substrate was being processed (i.e., before the measurement) and applied to the current control variable value.For example, feedback correction can be implemented by saving the temporal progression of the current state of the process group.

[0127] The control of a scalar manipulated variable (e.g., a coating rate) can be achieved using an integral controller. In this case, once a control input has been made, it remains active even if the deviation from the setpoint becomes zero, unlike with a proportional controller. A proportional or differential component (e.g., a full PID controller) is not necessarily required for all types of manipulated variables. For example, the proportional control element can be omitted. In this case, the proportional control element can be replaced, for instance, by a cascaded controller.

[0128] The control of a vectorial manipulated variable (e.g., a gas supply distribution) can be achieved analogously using a generalized vectorial integral controller. This will be described in more detail later.

[0129] The reaction rate of each elementary controller can be configured to be shorter than the time delay. This ensures that when a fault is detected at time t3, only those substrates processed in the interim (between t1 and t3) are found to have been processed incorrectly.

[0130] Optionally, the procedure 100 can implement cooperative behavior, for example, by instructing only relative changes to the manipulated variable as the control input. For example, a sputtering or electron beam power can be changed by a relative factor. The control of the process group then involves, for example, determining a first control factor and a second control factor (e.g., using the correction element), whereby the control of the process group using the first manipulated variable is based on the first scalar control factor by which the setpoint of the first manipulated variable is changed, and where the control of the process group using the second manipulated variable is based on the second scalar control factor by which the setpoint of the second manipulated variable is changed.

[0131] For example, the old target value of the electrical power supplied to the process (e.g., electron beam power or plasma power) can be multiplied by the scalar control factor to obtain the new target value. As a second example, a gas flow change can be made additively. Then, the old target value of the gas flow supplied to the process (e.g., reactive gas flow or working gas flow) can be added by the scalar control factor to obtain the new target value. Optionally, the total gas flow can also be scaled.

[0132] To put it more simply, independent adjustments to the control variables can be made between control interventions. However, at the moment of a control intervention, the current actual state can also be recorded and the control factor applied accordingly.

[0133] More generally, the control of the process group using the first manipulated variable can also be based on a state (e.g., actual state or target state) of the first manipulated variable at the time of measurement 101. Alternatively or additionally, the control of the process group using the second manipulated variable can also be based on a state (e.g., actual state or target state) of the second manipulated variable at the time of measurement 101.

[0134] This simultaneously enables the control of multiple similar processes. For example, a layer formed by sputtering can be produced sequentially by several identical processing devices to achieve a sufficient coating thickness. Through cooperative behavior, these multiple similar processes can be controlled together, even if they are set to individually different control variable values ​​(also referred to as setpoints).

[0135] Alternatively or additionally, the control of multiple processes can be weighted differently, or individual processes can be selectively included or excluded from the control during operation, for example, in the case of a technical defect. The same can apply to individual actuators, for example, if they are defective.

[0136] It may happen that measurement 101 (which can serve as a trigger for control 103) occasionally deviates from the predetermined fixed cycle. For example, the recorded current state of the substrate may be fed in at a temporarily higher rate because additional measurements were triggered manually. Measurement 101 may also be skipped for one or more cycles because a substrate is missing or has already been identified as defective. If the transport speed (also referred to as feed rate) and / or the distance between the substrates transported in immediate succession is changed, e.g., to alter the throughput, the cycle time of measurement 101 may even be permanently changed.

[0137] The controller can be configured to account for this deviation from the clock signal and handle it without interference. The most common and dangerous misconfiguration of PID-type controllers is excessively high gain values. This causes random disturbances to lead to increasing oscillations in the actuator output. As a result, defective components or even physical damage can occur. The basic controller can therefore incorporate self-monitoring (also known as a watchdog timer) that detects and compensates for this type of misconfiguration, as will be described in more detail later.

[0138] Fig. Figure 8 illustrates the process 100 according to various embodiments 800 in a schematic process diagram. Similar to the description above, several substrates are processed sequentially, which are here referred to as substrate 1 to substrate 6.

[0139] To prevent previously executed control steps from being repeated, a so-called predictor can be used. In simpler terms, a predictor uses information about the control loop model to predict future controlled variables. The Smith predictor is explained below using this as an example. The descriptions for the Smith predictor can also be applied analogously to other types of predictors.

[0140] Assume that a substrate 2, detected at time t4, was processed at a previous time t2 = t4 - t by the first process P1. The corresponding measured value is denoted as x, which was generated by the process setting y = A1 of the first process P1.

[0141] At measurement time t4, the process setting of the second process P2 had already been changed by Δy to y' = A2. To account for this, the virtual target state deviation x' can be determined.

[0142] Determining the virtual target state deviation x' can involve using a control loop model to simulate the process group in parallel with the actual controlled process group. The control loop model receives the actuator output y as input and calculates the expected control loop response Δx̂. The expected system response Δx̂ can be added to the measured value, providing a prediction as if the expected response had already occurred. This can be expressed as: x'=x+Δx^

[0143] The virtual target state deviation x' (visually a virtual measured value x') can be fed as a corrected measurement result 212o to one or more than one control element (e.g. PID control element) (compare the feedback correction).

[0144] The control loop model can be used intuitively to predict the response of the process group (also referred to as the system response). An exemplary control loop model can implement the system response as a step function (i.e., a step or instantaneous change) with a time delay Δt (also referred to as the predictor delay); and that the size of the step function is determined based on the control gain setting, for example, according to the following relationships: Δx^:=s⋅∑−tt0Δys:=14K

[0145] The Δy term intuitively reflects the sum of all control inputs Δy applied in the time interval from -t until now. The choice of the so-called system gain s is based on the experience that, with a sensible controller setting, the gain K typically lies in a range of approximately 0.2 to approximately 0.5 times the ideal correction. K, for example, denotes only the integral gain of the integral control element.

[0146] For an average value of K=14Kideal It gives exactly the correct amplitude of change.

[0147] Of course, a more precise model can also be used as the control loop model. The control loop model described above achieves this by adding only one additional parameter (namely, the time delay). In many applications, this is sufficient to achieve very good control performance.

[0148] Alternatively or additionally to correcting the target state deviation x (for example, to provide the corrected measurement result 212°), the output of one or more control elements can be corrected. This can be expressed as: Δy':=Δy+y(−t)−y(t0).

[0149] The expression y(-t), which represents the individual setpoint values ​​of the relevant actuators, can optionally be captured at the moment of processing 101 and stored in relation to the substrate, for example, assigned to a substrate identifier. This allows for an alternative or additional assignment using a substrate identifier (also known as a product ID), which is more robust.

[0150] More generally, the current state of the process group during processing of the substrate by the first process and / or during processing of the substrate by the second process can be assigned to the substrate (also referred to as substrate assignment), e.g., to an identifier (also referred to as substrate identifier) ​​of the substrate. The substrate identifier can, for example, have a sequential substrate number. The correction element can then take into account the assignment of the current state of the process group to the substrate (also referred to as substrate assignment).

[0151] Substrate allocation ensures that the heuristic, and potentially inaccurate, control loop model is not necessarily required. For example, if a layer is built up sequentially using several identical processes (e.g., coating processes each carried out using magnetrons), the assumption of a sudden change at time t0 is no longer entirely accurate. A similar situation arises if the rate change occurs with a finite ramp rate.

[0152] Optionally, the substrate allocation can be configured to better account for variable time delays (changed transport speed, temporary standstill).

[0153] The following section describes further exemplary implementations that take one or more constraints (also called limitations or position limits) into account. It should be noted that the description refers to the displacement by Δy'. The dynamic behavior may be essentially the same or similar to the implemented configuration.

[0154] For example, it can be taken into account that several actors are or will be set up in different ways. Then, an individual difference y(-t)-y(t0) will be assigned to each actor, which can be considered, for example, by defining how the individual results are combined.

[0155] The elementary controller then determines the relative change of 222° in the manipulated variable as follows: Δyu:=Kwt⋅x' In it they describe Δy u the change in the control variable before the limit (unlimited control output); K the rule gain (e.g. having a scalar constant); W t the clock weighting factor (e.g. having a time-dependent scaling), which will be described in more detail later; x' represents the virtual target state deviation, for example, the process characteristic value corrected by the output of the predictor.

[0156] Taking into account the clock weighting factor w t This clearly improves the stability of the control system when deviations occur in the actual measurement cycle. The base cycle ΔT corresponds to the nominal or average measurement cycle of the process group under production conditions. The actual elapsed time Δt (also known as the actual cycle interval) since the last control intervention can be determined by measuring the time. Based on the actual cycle interval, the cycle weighting factor w can be calculated. t determine, for example according to the relation: wt:=min{ΔtΔT,1.0}

[0157] If the actual clock interval Δt decreases (easily visualized as insufficient measurement intervals), the control intervention is dampened, and the result corresponds to an increase in the sampling rate of a conventional controller. The dynamic behavior as a function of (real) time then remains essentially the same.

[0158] If the actual clock interval Δt increases (intuitively, measurement intervals become too short), an overreaction is inhibited.

[0159] Consecutive measurement intervals that are too long slow down (with respect to real time) the reaction speed of the elementary controller, which at most increases the settling time. The relationship for the clock weighting factor w mentioned above already illustrates this. t This results in good control behavior, even when the actual clock interval fluctuates completely irregularly.

[0160] More generally speaking, the clock weighting factor w tbe a (e.g. non-linear) function of the actual clock interval.

[0161] The (still unlimited) position output Δy u Optionally, a check for limit violations can be performed. Checking the position output Δy u indicates that instead of the position output Δy u The limit value is used when the position output Δy u outside predefined limits (e.g. Δy) low , Δy high ) is located; and / or that a limitation of the position output Δy u to a predefined rate limit Δy rate This occurs when the amount of Δy u is greater than the rate limit Δy rate .

[0162] Optionally, the limits can be determined dynamically, as explained below. For example, for user-friendliness, the parameterization is performed in actuator units (actuator unit). The Δy limit values ​​are then dynamically calculated from the actuator limits.

[0163] For example, to regulate generator power, the parameterization is done as an absolute limit of the power: Plow≤P≤Phigh.

[0164] By rearranging P neu := P ist exp Δy is obtained logPlowPist≤Δy≤logPHighPist.

[0165] The rate limit P rate is specified as the change in rate within a beat: |Pneu−Pist|≤!wtPrate. The clock weighting factor is applied here, as otherwise rapidly successive steps could override the rate limit. Rearranging the formula yields the following result: |Δy|≤log(Pist+PratePist)≤−log(Pist−PratePist)

[0166] The properties of the logarithm mean that positive changes lead to a narrower limit than negative ones; consequently, the following is used as the limit: Δyrate:=log(Pist+PratePist).

[0167] If several generators are controlled in parallel as actuators, the limits are determined for each generator and the strongest limit is applied in each direction.

[0168] For other actuators (e.g., gas inlet, magnetic system), different equations apply for determining the actuator value from y. The dynamic limit is calculated analogously from these equations by solving for Δy.

[0169] The limited output Δy can be determined as explained below. Determining the limited output Δy involves an absolute limit being applied, followed by the application of the rate limit, and optionally, another absolute limit being applied. The absolute limitation can be determined, for example, according to the following relation: Δy1:={Δyunlif Δylow≤Δyunt≤Δyhigh 0if Δylow>ΔyhighΔylowif Δyunl<ΔylowΔyhighif Δyhigh<Δyuni.

[0170] If |Δy1| > Δy rateFurthermore, the following can be done: Determining an overdraft factor. v:=|Δyunl|Δyrate;

[0171] Determining a damped reaction Y2 := y unl ·1 / v; and if a limit was applied during the absolute limitation, repeat the absolute limitation with the damped response y2 as input.

[0172] The limited position output is then Δy:= Δy1, where this is optionally the value after repeating the absolute limitation.

[0173] This results in the following properties. Since the absolute limit is always applied last, it is ensured that it is not violated. In the case of unfulfillable boundary conditions, 0 (no intervention) is output. The rate limit is only applied if it is not already satisfied by the absolute limit. The damping method (scaling of y) unland renewed absolute limitation) prevents unnecessary distortion in the vectorized case. y→.

[0174] As a byproduct, the overdraft factor v is generated, which can be used in the subsequent self-monitoring.

[0175] Self-monitoring is configured to detect and counteract unstable controller behavior. Any control intervention can trigger self-monitoring. In principle, self-monitoring reacts to exceedances of the rate limit Δy. rate . The user's configuration of the rate limit can form the basis for the functioning of self-monitoring (also known as a watchdog).

[0176] In each intervention, the watchdog receives the overdraft factor. v:=|Δyunl|Δyrate.

[0177] This overrun factor is classified, for example qualitatively according to v < 1.0 (“OK” or “no overrun”) and v ≥ 1.0 (“not OK” or “overrun”).

[0178] During normal operation, transient deviations are possible, e.g., as a step response after a setpoint change. These are subject to a tolerance period measured in a number of control interventions for which they are acceptable.

[0179] To ensure this, the watchdog can be set up as a state machine that can switch between one or more of the following states.

[0180] The watchdog's default state (also called sleeping) can be configured so that if a value of "OK" is received, the default state remains, and if a value of "exceeded" is received, the state changes to the transition state.

[0181] The transition state (also called awakening) can have the following configuration: after a maximum of 10 position interventions, the state changes to the reaction state; an incoming value "OK" triggers a change to the reaction state if 3 consecutive values ​​are OK (for example, it can be assumed that the transient is over; and a newly arriving exceedance triggers an action); and an incoming value "exceedance" only increments the counter.

[0182] The response state (also called wake) can have the following characteristics: an incoming "OK" value triggers a return to the ground state if 25 consecutive "OK" values ​​occur; and an incoming "exceed" value triggers a watchdog response.

[0183] The behavior in the transition state (also called the awakening state) leads to a rapid switch to the desired case of oscillating behavior. The initial rate exceedance occurs during the maximum displacement of x, whereas during the subsequent return oscillation, "OK" values ​​occur, causing the rapid transition to the response state.

[0184] If an exceedance occurs in the reaction state (also called the wake state), the gain factor K of the controller is reduced as a response. This reduction is based, for example, on the overshoot factor v, according to the following relationship: K:=max(K1.1v,K2),

[0185] To illustrate, the amplification is then reduced to such an extent that, upon repetition of the step, the rate limit is only utilized to approximately 90%; however, by a maximum factor of 2.

[0186] After returning to the default state (also known as sleep state), K can optionally be left in the new setting. However, it can also be specified that no permanent, automatic changes to user-defined values ​​should be possible. In this case, the implemented configuration might cause the gain value to gradually increase back to its original value while in sleep state, potentially causing the unstable behavior to reappear after some time.

[0187] The following are exemplary implementations for converting the actuator output Δy into an actuator variable (according to which the corresponding actuator is controlled).

[0188] The manipulated variable for an actuator can be determined based on the manipulated output Δy (intuitively a dimensionless change in the manipulated variable) and the actual state of the process group.

[0189] If the actuator has a generator, the generator power P can be changed as the manipulated variable (e.g. exponentially), for example according to the following relation: Pneu=Pist⋅expΔy

[0190] In the case of multiple generators j, the same relation can be applied to all generators, optionally with a scaling factor a for each generator. j can be set up: Pj,new=Pj,is⋅exp(ajΔy).

[0191] The described behavior can also apply to other types of control variables, such as an electrical voltage as a control variable or the total gas flow (e.g., total reactive gas flow) as a control variable.

[0192] The gas supply distribution (also called trim gas distribution) can be represented as follows. In the gas supply distribution, the manipulated variable can be written as a vector, e.g., according to: Δy→≡yi,i=1..N.

[0193] The control value is then the gas flow. fi≡f→ per gas inlet channel (also referred to as individual gas inlet), and can be specified, for example, in sccm (standard cubic centimeters per minute) or slm (standard liters per minute). Control is achieved, for example, additively and relative to the sum of the individual gas inlets: fi,neu=fi,ist+aj⋅Δyi⋅∑ifi,ist.

[0194] Similar to multiple generators, a scaling factor a can also be optionally used here. j can be used for each of the gas supply channels.

[0195] In the case of the magnetic field distribution, the manipulated variable can also be written as a vector, e.g. as: Δy→≡yi,i=1..N.

[0196] The setpoint is then, for example, the displacement of at least one of several magnets that provides the magnetic field distribution, for example, the vertical stroke. hi≡h→ the individual support points, specified, for example, in millimeters or micrometers.

[0197] The control variable can then be determined, for example, according to the following relationship: hi,neu=hi,ist+aj⋅Δyi.

[0198] The following section will discuss multi-component quantities in more detail, whose components can be, for example, measured values, characteristic values ​​or control values ​​used in parallel with each other.

[0199] Fig. Figure 9 illustrates a processing device 900 according to various embodiments in a schematic diagram, which is configured to carry out a coating process as an exemplary process. The coating process can comprise the formation of a layer on a substrate using a coating material. For this purpose, the coating process can be supplied with at least the coating material or a reaction product of the coating material, with one or more gases, with electrical power, and optionally with heat energy supplied to the substrate. The layer can, for example, comprise a reaction product of the reactive gas and the coating material, e.g., an oxide or a nitride.

[0200] The coating process 902d can incorporate a gas-phase deposition. The following refers to sputtering as an exemplary physical vapor deposition (PVD), although the description can be applied analogously to other PVD processes or chemical vapor deposition (CVD). The atomized coating material can be supplied by means of a magnetron (also referred to as a sputtering device).

[0201] The processing device 900 can perform several subprocesses, which interact along a chain of interactions and each of which can influence the coating process. A first subprocess 902a of the processing device 900 can involve generating a plasma. A second subprocess 902b of the processing device 900 can involve atomizing the coating material using the plasma. An optional third subprocess 902c of the processing device 900 can involve reacting the atomized coating material with a reactive gas.

[0202] Fundamentally, various measurable variables can be used as manipulated or controlled variables, depending on how the coating process is regulated. Therefore, the following discussion focuses on the more easily understood actuators that influence the operating point of the coating process.

[0203] An exemplary first actuator 852 is configured to influence the electrical supply to the plasma. For example, the first actuator may include a generator and / or be configured to supply an electrical voltage, an electrical power, and / or an electrical current to the plasma. Changing the electrical supply to the plasma may alter the rate at which the coating material is atomized.

[0204] One or more exemplary second actuators 854 are configured to influence the magnetic field in which the plasma is generated, e.g., its spatial distribution. For example, each second actuator 854 may have a lifting device and / or be configured to move a magnet of the magnetic system. Changing the magnetic field can alter the plasma and thus the rate at which the coating material is atomized (also referred to as the atomization rate).

[0205] An exemplary third actuator 856 is configured to influence the supply of working gas to the plasma, e.g., the total working gas flow. The working gas can be figuratively understood as the gas that is ionized to form the plasma. For example, the third actuator 856 may have a valve and / or be configured to change the total working gas flow. Changing the total flow can alter the pressure to which the plasma and / or the atomized coating material are exposed. Higher pressure can cause the atomized coating material to disperse more widely, which affects the transfer of the coating material to the substrate, and / or can influence plasma formation.

[0206] One or more exemplary fourth actuators 858 are configured to influence the spatial distribution of the working gas. For example, each fourth actuator 858 may have a valve and / or be configured to divide the total flow of working gas into several gas supply channels (also referred to as gas channels). Changing the spatial distribution of the working gas supply to the plasma can alter a pressure gradient to which the plasma and / or the atomized coating material are exposed.

[0207] The second subprocess 902b, e.g., the atomization itself, cannot always be directly influenced. An exemplary fifth actuator (not shown) is configured to purge the solid coating material with a purge gas. The purge gas displaces, for example, the plasma or optionally present reactive gas, thus influencing the atomization rate. If the magnetron is a tubular magnetron, an exemplary additional fifth actuator (not shown) can be configured to influence the rotational speed of the tubular coating material. This affects the duration that a surface area is exposed to the plasma, and therefore its temperature and / or chemical composition.

[0208] More generally, the first and second subprocesses provide the means to convert the solid coating material into the gas phase. If plasma is not used for this purpose, appropriate actuators can be installed to influence the conversion of the solid coating material into the gas phase.

[0209] An exemplary sixth actuator 860 and one or more exemplary seventh actuators 862 are configured analogously to the third / fourth actuator(s), with the difference that these are configured to influence the supply of a reactive gas to the atomized coating material, e.g., a spatial distribution of the reactive gas supply. This influences the chemical composition of the formed layer and affects the atomization rate.

[0210] Naturally, more or fewer of the aforementioned actuators can be used, or other manipulated variables can be employed. It should also be understood that the above separation of subprocesses is merely for clarity. For example, the aforementioned subprocesses need not be strictly separated from one another. Alternatively or additionally, an actuator can influence multiple variables, or several actuators can interact with each other.

[0211] Therefore, the following section will more clearly refer to the entirety of subprocesses 902a, 902b, and 902c as a process, where the process is supplied, for example, with one or more gases, coating material, electrical energy, and so on. The actuators can then include the components by which the process is supplied and / or maintained. Supplying the process can, for example, involve providing electrical power, one or more gases, thermal energy, a magnetic field, etc.

[0212] More generally speaking, those actuators that influence the operating point of the process can be controlled.

[0213] Fig. Figure 10 illustrates a vector space 1000 according to various embodiments in a schematic cross-sectional view. As explained above, the actuator output can have several components in the case of multiple actuators, which is noted here in the more easily understandable vector notation. The resulting vector can, for example, be used as a characteristic value Φ or at least more easily transformed into a characteristic value Φ that represents the changes of all actuators together.

[0214] Generally speaking, the characteristic value can be a function of the vector components of the vector. In the following, the easily understood magnitude of the vector will be used as an example of a characteristic value. Similarly, the easily understood position coordinate towards which the vector is directed will also be used as an example of a characteristic value. It can be understood that the characteristic value can also be determined using a different operation, for example, according to the applied norm of the vector space.

[0215] For example, the intersection point of the vector with a plane can also be used as a characteristic value.

[0216] For example, the parameter can be scalar or at least based on a scalar product. The magnitude of the vector can be defined as its scalar product with itself.

[0217] The following discussion refers to the easily understood Cartesian vector space. It should be understood that the multi-component output can also be notated differently, e.g., not necessarily in a vector space or not necessarily in Cartesian coordinates.

[0218] The Cartesian vector space is spanned by the corresponding Cartesian coordinates, each of which is assigned to at least one of the multiple actuators. The value on the coordinate then represents the control input, according to which the actuator to which the coordinate is assigned is controlled. Each component (also called a vector component) of the actuator output, denoted as a vector, can analogously be assigned to one of the actuators.

[0219] A measurable physical property of a substrate coating, such as layer thickness and / or optical refractive index, can be used as a measurement parameter. This physical property can be optimized with respect to its homogeneity (also referred to as uniformity) across a large substrate area. For example, the measured value of the property at each measured location on the substrate (e.g., the test area) should exhibit the smallest possible deviation from the mean value of all locations. This can be expressed as: maxi|yi−y¯|→min., with y¯:=1N∑iyi

[0220] A coating system can have several groups of similar actuators that allow for localized changes in the coating properties. Examples include multiple fourth actuators, multiple seventh actuators, and / or multiple second actuators. For instance, several spatially distributed gas channels at different locations within the coating device can cause a local change in the coating thickness. For example, at least for some coating materials, the refractive index can be changed via the local reactive gas flow. For example, a magnetron can have a magnet system with several second actuators, each of which is configured to change the spatial distribution of the magnets in the magnet system, thus locally altering the magnetic field strength on the coating material to be atomized and thereby causing a change in the coating thickness.The magnetic system (in the longitudinally extended case also referred to as a magnetic bar) can have a support and several segments which are supported by the support (also referred to as support points), each of which has a second actuator (e.g. a lifting device) and one or more magnets, wherein the second actuator is configured to move the one or more magnets relative to the support, e.g. away from it or towards it (herefore simply referred to as a stroke).

[0221] The manipulated variable is the output (visually, the position) of the actuator used (for example, a mass flow controller or a servo motor). Examples of manipulated variables include a gas flow (e.g., in sccm) or a servo position (e.g., in mm), which can be represented as a scalar (also called a numerical value). A group of similar actuators, for example, has one manipulated variable per actuator. Such a list of manipulated variables is referred to here as a vector manipulated variable (also called a manipulated variable vector). Each component of the manipulated variable vector contains the manipulated variable for the actuator assigned to the respective index position of the component.

[0222] According to various embodiments, the method can provide 100 one or more than one of the following: The input consists of the measured or calculated property value at several (at least two) locations on the substrate (also referred to as measuring points). The number and location of the measuring points can be a function of the number and spatial arrangement of the actuators. - The output is the change in the vector control variable in order to achieve optimal or at least improved uniformity. - If one or more actuators reach an assigned actuation limit, the remaining free actuators are controlled in such a way as to achieve the best possible uniformity under the constraint. - Crosstalk of the positioning effect to spatially distant areas of the substrate is suppressed.

[0223] Inhibiting crosstalk is advantageous, for example, when applied to a gas supply system with multiple gas channels. With a native 1:1 mapping of measuring point to actuator, a change in a control value would significantly alter the stratification properties at neighboring and more distant measuring points as well, resulting in undesirable feedback effects and oscillations in a closed-loop control system.

[0224] The following explains a so-called conversion of the measurement grid (also referred to as grid conversion or resampling), which can be done optionally if, for example, the dimension of the vector measurement quantity and the dimension of the control vector differ from each other.

[0225] The raster conversion can involve calculating zone-wise partial averages based on the acquired property values ​​to obtain an input in the dimension (e.g., number of channels) of the control vector. For each actuator, a spatial area (visually a "zone," e.g., a spatially extended section along the coating device) is specified, and all measurement points within this zone are averaged. This type of raster conversion provides one or more of the following: - The measuring grid can be changed between two position interventions, e.g. to capture more or fewer measuring points, without affecting the control. - The failure of individual measuring points or sensors has less of an impact on the control system, for example as long as each zone has at least one measuring point. - If necessary, several zones can overlap at least partially or even completely.

[0226] The raster conversion can alternatively or additionally include interpolation and / or extrapolation of the measured values, e.g. by means of a polynomial process.

[0227] The processing chain of the elementary controller is briefly explained below and then discussed in more detail. The elementary controller can, for example, operate as an integral controller. As described above, the elementary controller stores, for instance, output corrections and the output time (time of the control intervention) in order to compensate for the substrate propagation times between control intervention and measurement using the Smith predictor.

[0228] The system can be configured to detect only relative changes based on the current state of the substrate. This allows for collaboration with other systems that control the actuators (also known as cooperative behavior).

[0229] The control output can be formed as a linear scaling of the deviation vector (integral control). If a limit is exceeded, a limited control vector can be determined that satisfies all boundary conditions and has a minimal Euclidean distance from the ideal (unlimited) control vector.

[0230] As an interim result, the so-called correlated output is obtained. Δy→ This signal can optionally be filtered using a linear transformation (also known as decorrelation) to compensate for or at least inhibit crosstalk between the control channels. The result of the decorrelation is the actuator output. Δz→, whose components are uniquely assigned to the actuators.

[0231] A boundary can be represented analogously in vector notation. In Hessian normal form, it can be written as follows: y→⋅c→≤ξ

[0232] If this relation is satisfied, the criterion of the limitation is met, or the limitation is not violated. This is intuitively represented. y→ the position vector in vector space, c→ a (e.g. normalized) normal vector of a plane, which specifies the position of the boundary, and ξ the distance of the plane from the origin of the vector space.

[0233] In this way, every boundary along a coordinate of the vector space can be defined by a normal vector. c→ (with |c→|=1 and express the scalar value ξ uniquely. The boundary along a coordinate (visually, an individual control channel) can then be expressed as c→=(0…010…0)T.

[0234] This representation makes the processing of boundary conditions accessible to linear algebra. This allows boundary conditions to be combined and constrained optimization to be performed under the influence of the applicable boundary condition(s).

[0235] Furthermore, other boundary conditions can also be formulated as linear equations with respect to the vector control variable (also called the control vector). For example, an additional constraint can be defined that the sum of all vector elements should take on a specific value.

[0236] Furthermore, this allows for a linear transformation of position outputs, for example by performing an inverse transformation of the limits.

[0237] Optionally, in the event of repeated violations of a limit (e.g., a rate limit), the watchdog can be configured to reduce the control gain in order to inhibit oscillation of the elementary controller.

[0238] To compensate for or at least inhibit crosstalk between adjacent or even distant actuators, a decorrelation of the actuator output can be implemented. In a simpler implementation, this decorrelation involves multiplication by a decorrelation operator D, for example, a matrix or one derived from a matrix. The decorrelation operator is, in principle, freely selectable. For instance, one can experimentally determine the crosstalk factors of the actuators and derive the decorrelation operator from them by (pseudo)inversion.

[0239] However, decorrelation is not necessarily required for setting up a magnetic system.

[0240] For adjusting one of the spatial gas supply distributions, a decorrelation can optionally be implemented, for example, by adjusting immediately adjacent gas supply channels in opposite directions. Without a decorrelation, feedback (e.g., triggering oscillations) can occur between the gas supply channels.

[0241] The following refers to the control of multiple actuators, which can be done analogously to what has been described previously.

[0242] Representing the actual state of the substrate, the dimensionless setpoint deviation, i.e., the setpoint deviation x, can be used and fed to the elementary controller. In the case x = 0, the actual state is equal to the setpoint of the substrate. The case x ≠ 0 can indicate a quantized deviation from the setpoint.

[0243] For vector notation, the target state deviation x described above becomes the target state deviation vector. x→ (also referred to as the deviation vector) is generalized. It contains as many components as there are control variables (i.e., control channels).

[0244] Analogous to the deviation from the target state, the position output can also be Δz→ It is recorded as a dimensionless quantity, which is then converted into a physical control variable.

[0245] The issue Δz→ can vividly represent the relative scaling (in the case of multiplication) or shifting (in the case of addition) of a position value.

[0246] This ensures that the procedure can be applied to several related processes with different actual states. The elementary controller, for example, works exclusively with relative changes. Δz→.

[0247] In the event of a limit violation, a limited control vector (also called a restricted control vector) is determined that has the smallest possible Euclidean distance to the ideal (unlimited) control vector.

[0248] An exemplary implementation of procedure 100 may include one or more of the following: - For all zones, the measuring points contained within them are identified and their layer thickness values ​​are averaged. The zone averages are then incorporated into the deviation vector. x→ transferred. - The sum of corrections already applied but not yet effective in the measurement is used to calculate a virtual deviation vector from the measured deviation vector. The virtual deviation vector represents the expected measured values ​​after applying the corrections already made (intuitively, a prediction). - There will be an unlimited correlated position output Δy→u (e.g. as a key figure) is calculated by multiplying the deviation vector by a proportional factor. Δy→u is a vector whose components are assigned to the actuators. - The correlated position output is checked for violations of limits (e.g., position limits and rate limit). In case of a violation, a new correlated position output is generated. Δy→ calculated, which is based on the previously determined output Δy→u as close as possible. - Using Δy→u The watchdog is optionally applied based on the configured rate of change limit (also known as the rate limit limit). - The correlated position output is converted to the decorrelated position output through linear transformation. Δz→ calculated. - The vector control variable is increased by the vector Δz→ changed: Z→1:=Z→0+a⋅Δz→, where optionally a scaling factor a ≠ 1 can be applied, e.g. a scaling according to total gas flow or a relative division among several similar actuators.

[0249] An exemplary implementation of calculating zone averages is explained below. Each actuator is assigned a spatial region (visually referred to as a zone, e.g., a spatially extended section along the coating device), whereby all measuring points of the substrate that lie within this zone are averaged. A position interval [x] can be used as a criterion for the zone. start ,x end ] and / or be assigned a list of track identifiers.

[0250] One or more of the zone criteria can also be empty (intuitively "not set"), i.e., omitted. All measuring points whose location lies within the position interval [x] assigned to the zone are assigned to the zone. start ,xend ] is located and / or its identifier is included in the list of track identifiers assigned to the zone. If one or more of the two criteria are not set, their check can be skipped.

[0251] The spatial allocation of the zones does not necessarily have to correspond exactly to the location of the actuators. For example, the measurable substrate width in an inline sputtering system may be narrower than the width of the coating device and / or the spatial distance between the outer gas supply channels (also referred to as the gas supply width). In this case, the detectable area on the substrate can optionally be divided according to the proportionally scaled width of the gas supply.

[0252] The result of the measurement at the specified time is the target deviation x. For example, the target deviation x can represent the deviation of the actual layer thickness of the substrate or its coating from the target layer thickness, expressed, for instance, in relative units. For example, x = -0.05 can indicate that the actual layer thickness is 95% of the target layer thickness (more generally referred to as the reference value of the layer thickness).

[0253] The zone mean value for each zone can, for example, be an arithmetic mean of the individual measurements. xi:=(xj)j∈Zi'

[0254] Of course, another relation can also be used to combine the individual measured values ​​per zone into a target state deviation x representing the i-th zone. i to transfer.

[0255] Two zones can also overlap, e.g. by their position intervals [x start, x end They overlap. This allows one or more measuring points to be included in the target state deviation x. i It incorporates multiple zones, thus achieving a more robust finish.

[0256] The individual deviations from the target state x i (i=1 to I) form the components of the deviation vector x→.

[0257] The mechanism described above for regulating based on the scalar deviation from the target state x (also called scalar regulating) can be applied component-wise to the deviation vector. x→ to be applied. The virtual target state deviation x→' This results from anticipating the expected system reaction to the corrective steps taken in the meantime. x→^, which can be expressed as: x→'=x→+x→^.

[0258] The expected system response can be determined based on the correlated actuator output. Δy→ to be determined. Intuitively, it can be assumed that the effect of the decorrelation (i.e., the transformation of Δy→→Δz→ The system reverses (e.g., inverts) the operation. As with the scalar quantity, a step function can be implemented as the system response, which is executed with a time delay Δt and an estimated step height. 14K which can be expressed as follows: x→^:=14K∑−tt0Δy→

[0259] In this case, x→' serve as input for the integral control step.

[0260] This can be expressed as: Δy→u:=Kwt⋅x→', where K is the control gain (for example, a scalar constant) and w tThis refers to the clock weighting factor (e.g., causing a time-dependent scaling). It can then be expressed that: wt:=min{ΔtΔT,1.0}, where Δt denotes the elapsed time since the immediately preceding position intervention and ΔT denotes the configured base clock (input parameter).

[0261] As described above, one or more than one positioning limit (for example, having an upper limit and / or a lower limit) can be expressed as: y→⋅c→≤ξ

[0262] This makes it easier to see that the scalar product y→⋅c→ The scalar parameter Φ can be used to determine whether the parameter satisfies the criterion Φ ≤ ξ. In this case, the parameter is a function of the vector components of y→ (also as y i (designated) and the vector components of c→ (also known as c i(designated) . The parameter Φ can, for example, be expressed as: Φ=∑iciyi

[0263] The parameter set (c→,ξ) is referred to herein as a constraint. The constraint represents a limitation of c→y→=ξ. to the half-space on one side of the hyperplane y→ The hyperplane is referred to here as the boundary surface of the constraint.

[0264] The following is an example implementation that allows the constraint to be updated.

[0265] To determine the appropriate limits for the current actuator action, the process for calculating the output value can be reversed. This is explained below using several fourth actuators 858 as examples, although the description can also be applied analogously to other actuators.

[0266] Each exemplary fourth actuator 858 is configured to influence the spatial distribution by which the plasma is supplied with working gas. For example, each fourth actuator 858 may have one or more valves configured to divide the total flow of working gas into several gas supply channels (also referred to as gas supply channels).

[0267] For the i-th channel of the gas supply, for example, the limits f low , f high It must be predetermined. Formulated as constraints, it can be written as follows: c→f,i,low=(0…0(−1)0…0)T,ξf,low=fi,ist−flow cf,i,high=(0…010…0)T,ξf,high=fhigh−fi,is

[0268] In the following, the subscripts "i,low" and "i,high" are omitted to simplify the notation. The described behavior applies analogously to each gas supply channel and its respective constraints.

[0269] By rearranging f i,neu = f i,ist+ aj · Δz i · Σ i f i,ist The transformed constraint for Δz is obtained as follows: c→z=c→f;ξz=ξfaj∑ifi,ist.

[0270] The optional decorrelation is then achieved through a linear transformation. Δz→:=D_⋅Δy→.

[0271] To get out of the constraint (c→z,ξz) for Δz→ the corresponding constraint (c→y,ξy) for Δy→ To determine, the defining equation can be used. czΔz→<ξz can be used. The left side is identical to c→zDΔz→, which results in: c˜y:=c→z⋅D_;ξ˜y:=ξz. The established formula is: c→y:=c˜y|c˜y|;ξy:=ξ˜y|c˜y|, where c→y In general, it can have several components other than zero, and in this case, it can no longer be uniquely assigned to a gas channel.

[0272] The same principle applies to the second actuator 854, which is configured to influence the magnetic field generated by the magnetic system. Only the transformation of the displacement h i of the magnet (e.g. having a stroke) in the change Δz i The investigation is conducted somewhat differently. From hi,neu=hi,ist+aj⋅Δyi it follows that: C→z=C→h;ξz=ξhaj

[0273] Fig. Figure 11 illustrates a vector space according to various embodiments in a schematic cross-sectional view 1100. The application of the boundary (also called limiting) is explained below.

[0274] Determining the limited issue Δy→ (also referred to as second-place output) can exhibit: absolute limitation by one or more orthogonal projections onto one or more boundary planes; checking the rate limit based on Δy→1, where, if this limit is exceeded, a transfer of Δy→u→Δy→2 this occurs; and optionally a renewed absolute limitation by orthogonal projection onto one or more than one boundary plane, which is a transformation of Δy→2→Δy→3 provides.

[0275] This sequence vividly illustrates one or more of the following: - This avoids unnecessarily shortening the output vector, which can occur, for example, if the rate limit is applied first and then the projections are applied. - This avoids an unnecessarily large misalignment of the positioning vector, which can occur when the projected vector Δy→1 would be shortened, as illustrated in cross-sectional view 1100. - The unlimited vector is vividly reduced to the magnitude Δy rate shortened and the result Δy→2 again limited, which changes the order explained above. Δy→u→Δy→2→Δy→3 results.

[0276] Cross-sectional view 1100 shows an exemplary case in which two vector transformations are performed and the results are checked, with all three processes leading to a changed intermediate result. The resulting vector Δy→3 lies as close as possible to the ideal vector 31. The deviation between Δy→3 and vector 31, for example, can be accepted in order to require as little computing power as possible.

[0277] The absolute limitation mentioned above is explained below. The absolute limitation is based, for example, on any number of constraints. (c→y,n,ξy,n).

[0278] A minimum position vector is determined. e→, which leads to the sum Δy→u+e→ The condition of all constraints is met.

[0279] The given constraints for Δy→ are in constraints (c→n,ξn) with reference to e→ transformed, which can be written as: cy,n(Δy→u+e→)≤ξy,n

[0280] This suggests that: (c→n,ξn):=(c→y,n,ξy,n−c→y,n⋅Δy→u) is.

[0281] Based on this, two cases can be distinguished. In case 1, all ξ n ≥ 0m or in other words, fulfills e→=0→ All constraints. In case 1, no limitation is needed. In case 2, there is at least one n such that ξ n < 0.

[0282] This boundary value problem can be solved using quadratic optimization or other computer science methods. Quadratic optimization (also known as quadratic programming) and the quadratic programs used for it belong to a special area of ​​mathematical optimization that intuitively uses particularly simple functions.

[0283] Other methods also exist for determining the optimal positioning vector. e→ can be determined.

[0284] Examples of solutions include: the gradient descent method; the interior-point method; active-set methods, such as the primal dual active-set algorithm; Lemke's algorithm; and unique sink orientations. If the boundary value problem conditions lead to a linear complementarity problem, more specialized algorithms such as Lemke's algorithm or unique sink orientations are available.

[0285] The gradient descent method can be characterized by progressing from a starting point along a descent / ascent direction until the numerical improvement plateaus. For example, if the negative gradient is chosen as the descent direction, i.e., the direction of the locally steepest descent, the result is the steepest descent method. In general, the gradient descent method refers to an optimization method in which the descent direction is determined by gradient information, and is therefore not necessarily limited to the negative gradient.

[0286] Active-set methods comprise a class of iterative algorithms for solving quadratic optimization problems. The Primal Dual Active Set algorithm is a method for solving a quadratic optimization problem over a convex subset of a Hilbert space over a corresponding set. The Interior Points method comprises a class of algorithms for solving optimization problems. Its main application area is, for example, linear or quadratic programming. Compared to active-set methods (e.g., those containing a simplex method), the Interior Points method exhibits better theoretical properties (e.g., polynomial complexity) and faster convergence for very large, sparsely populated problems. The Primal Dual Active Set algorithm uses the Lagrange multiplier to arrive at a solution that is both feasible and as optimal as possible.

[0287] Optionally, it can be determined whether there are conflicting, i.e., unfulfillable, constraints. If these exist, the corresponding intervention can be skipped, and no change can be made to the process's operating point, or none of the process's actuators can be set.

[0288] The result of the iterative method is the position vector e→ received, or Δy→q=Δy→u+e→or Δy→3=Δy→3=Δy→2+e→.

[0289] The rate limit mentioned above is explained below. The rate limit is applied when |Δy→1|>Δyrate This shows that |Δy→1| can be used as a scalar parameter Φ to determine whether the parameter Φ fulfills the criterion that Φ > Δy race The parameter Φ can be expressed, for example, as: Φ=∑iyiyi

[0290] Applying the rate limit can involve: Determining the overdraft factor v:=|Δy_u|Δyrate; Determining a dampened response as Δy→2:=Δy→u⋅1 / v; If case 2 applied, i.e., absolute limitation had already been applied; applying absolute limitation to the dampened response Δy→2.

[0291] The limited position output is then Δy:= Δy1, where this is optionally the value after applying the absolute limitation to the damped response. Δy→2 may be.

[0292] The balancing behavior is explained below. As a byproduct of the mechanisms described above, the elementary controller can be configured to balance (also called a balancing elementary controller). The balancing elementary controller can be configured to provide a mean-adjusted control intervention. For this purpose, a so-called sum constraint can be formulated, for example, according to the relation: (11…1)T⋅Δy→=!0→c→=1N(1…1)T,ξ=0

[0293] The change vector Δy→u and all real constraints can be projected onto the boundary of the sum constraint.

[0294] An exemplary implementation of the decorrelation mentioned above is explained in more detail below. The decorrelation can exhibit the correlated output. Δy→ into the decor-related output Δz→ to transfer, which can be expressed, for example, as a transformation that reads: Δz→:=D_⋅Δy→, where D denotes the decorrelation operator.

[0295] If the magnetic field or the configuration of the magnetic system is to be changed, the decorrelation is not necessarily required. Instead, the decorrelation, or the decorrelation operator D, can be such that the z is mean-corrected. Intuitively, the decorrelation operator D, or the transformation it performs, can ensure that all vector elements are changed by a constant.

[0296] If the spatial gas supply distribution is to be changed, the decorrelation operator D, written as a tensor, can, for example, be expressed as follows: D_:=12[1−10…0−12⋱⋱⋮0−1⋱−10⋮⋱⋱2−10…0−11]

[0297] Intuitively, the decorrelation operator D causes a change in one actuator and an equal and opposite change in the immediately adjacent actuator. A change in the inner actuators is compensated by an opposite change of half the magnitude in the adjacent actuators. This means that fewer or no experimentally determined parameters are required.

[0298] Alternatively, the decorrelation operator D can also be determined experimentally, which improves its effectiveness. This involves determining how an actuator affects more than one controlled variable or measured variable, or how directly adjacent manipulated variables interact or depend on each other (also known as crosstalk).

[0299] In the case of gas supply distribution, the gas supplied by one gas supply channel can spread and overlap with the gas flow of an immediately adjacent gas supply channel. This overlap can be detected, for example, by slightly adjusting the gas flow of each gas supply channel and then taking measurements. The result is a so-called crosstalk matrix C, whose pseudoinverse can be used as a decorrelation operator D. However, the pseudoinversion of the crosstalk matrix can be very susceptible to amplifying small measurement errors.

[0300] The following section explains further exemplary implementations for changing the operating point of the process group, or its processes. Controlling an actuator can be based on the dimensionless actuator vector. Δz→ and based on the current state of the process. The dimensionless control vector can be used for this purpose. Δz→ based on the current state of the process, the input is converted into a control variable or its setpoint, which is then supplied to the actuator for control purposes.

[0301] In the case of gas supply distribution, Δz→≡Δzi (i=1 to N). The gas flow can be used as the control variable. fi≡f→ The individual gas supply channels are used, for example, specified in sccm or slm. The adjustment is additive and relative to the sum of the individual gas flows, which can be expressed as: fi,new=fi,ist+aj⋅Δzi⋅∑ifi,ist

[0302] The optional scaling factor a j This is used for relative weighting when using multiple gas supply channels j (also referred to as gas channels).

[0303] In the case of magnetic field distribution, Δz→≡Δzi (i=1 to N). The vertical stroke can be used as the setpoint. hi≡h→ per support point, specified, for example, in mm or µm. The adjustment is additive, which can be expressed as: hi,neu=hi,ist+aj⋅Δzi According to various embodiments, a measurement is converted into controlled variables, whereby an individual delay time is taken into account for each controlled variable.

[0304] According to various embodiments, the measurement or the controlled variable is corrected taking into account the individual delay time (e.g. by means of a Smith predictor).

[0305] According to various embodiments, the output of the control element is corrected (e.g. alternatively or additionally to the correction of the measurement or the controlled variable) taking into account the time-delayed state of the process in the past.

[0306] According to various embodiments, the measurement or the controlled variable is corrected taking into account past control interventions.

[0307] According to various embodiments, the operating point of the process at the time of measurement is predicted based on past control interventions.

[0308] The following are various examples that refer to what has been described previously and depicted in the figures.

[0309] Example 1 is a method for operating a process by which a substrate is processed, comprising the following: determining a first actuator output based on a recorded result of the process, wherein the first actuator output has (e.g., exactly) one component per actuator of a group of actuators by which the process is controlled, which is (e.g., only) assigned to that actuator; determining a second actuator output based on the first actuator output and based on a parameter that is a function of each component of the first actuator output; controlling the group of actuators by which a process is controlled (e.g., by which a supply and / or maintenance of the process takes place) according to the second actuator output; wherein, for example, the second actuator output has (e.g., exactly) one component per actuator of the group of actuators by which the process is controlled, which is (e.g.,only) is assigned to the actuator (for example, according to which the actuator is controlled).

[0310] Example 2 is the procedure according to Example 1, wherein determining the second position output shows that several constraints of the process (e.g., constraints of supply) are taken into account, where, for example, a number of the multiple constraints is greater than a number of the multiple components.

[0311] Example 3 is the procedure according to Example 1 or 2, wherein determining the second position output shows that the characteristic is compared with at least one limitation of the process control (e.g., supply limitations), where the multiple limitations, for example, show at least one limitation.

[0312] Example 4 is a method (e.g., according to one of Examples 1 to 3) for operating a process by which a substrate is processed, comprising the following procedure: determining a second actuator output based on a captured result of the process; controlling a group of actuators by which a process is regulated (e.g., by which the process is supplied and / or maintained) according to the second actuator output; wherein the second actuator output has (e.g., exactly) one component per actuator of the group of actuators by which the process is regulated, which is (e.g., only) assigned to the actuator (e.g., according to which the actuator is controlled); wherein, in determining the second actuator output, several limits of the control of the process (e.g., limits of supply) are taken into account, wherein the number of the several limits is greater than the number of the several components.

[0313] Example 5 is the procedure according to Example 4, wherein determining the second actuator output involves determining a first actuator output based on the recorded result of the process and mapping the first actuator output to the second actuator output based on a characteristic value which is a function of each component of the first actuator output, wherein, for example, the first actuator output has several components, each of which is assigned to at least (e.g., exactly) one actuator of a group of actuators.

[0314] Example 6 is the procedure according to Example 5, where determining the second position output shows that the parameter is compared with at least one of the several limitations of the process control (e.g., supply limitations).

[0315] Example 7 is the procedure according to one of Examples 1 to 6, wherein the multiple limits per component of the position output have one or more limits on a position value (e.g., a supply rate).

[0316] Example 8 is the procedure according to Example 7, wherein one or more than one limit has a lower limit of the setpoint (e.g. the supply rate).

[0317] Example 9 is the procedure according to Example 7 or 8, wherein one or more than one limit has an upper limit of the setpoint (e.g. the supply rate).

[0318] Example 10 is the procedure according to one of Examples 7 to 9, wherein the multiple constraints include at least one constraint on the rate of change of the process (e.g., a constraint on the rate of change of supply).

[0319] Example 11 is the procedure according to Example 10, taking into account the limitation of the change rate of the process (e.g. change per position intervention or change per cycle) before taking into account the limitation of the setpoint.

[0320] Example 12 is the procedure according to one of Examples 1 to 11, wherein each component of the first and / or second position output has an indication of the extent (e.g. by what value) to which the at least one actuator assigned to the component is deflected (e.g. by what value its actuator variable is changed).

[0321] Example 13 is the procedure according to one of Examples 1 to 12, wherein the number of multiple boundaries is greater than twice the number of multiple components.

[0322] Example 14 is the procedure according to one of Examples 1 to 13, wherein the process is supplied with material by means of the group of actuators by means of which the process is controlled.

[0323] Example 15 is the procedure according to Example 14, wherein each actuator in the group of actuators supplies the process with the same material and / or regulates the spatial distribution with which the supply is carried out.

[0324] Example 16 is the method according to one of Examples 1 to 15, wherein the group of actuators by means of which the process is controlled controls a magnetic field to which the process is exposed and / or controls a spatial distribution of the magnetic field.

[0325] Example 17 is the procedure according to one of Examples 1 to 16, wherein the group of actuators by means of which the process is controlled regulates a pressure to which the process is subjected and / or regulates a spatial distribution of the pressure.

[0326] Example 18 is the procedure according to one of Examples 1 to 17, wherein a spatial distribution is provided by means of the group of actuators, with which the process is supplied.

[0327] Example 19 is the procedure according to one of Examples 1 to 18, wherein, to determine the second position output, one or more than one limit is taken into account such that the second position output (e.g., each of its components) lies in a range of values ​​which is bounded by the one or more than one limit, e.g., enclosed by the multiple limits.

[0328] Example 20 is the procedure according to one of Examples 1 to 19, wherein the position output has a distance of at least one limit (e.g. the limit of the rate of change) of the several limits.

[0329] Example 21 is the procedure according to one of Examples 1 to 20, wherein the result of the process that is recorded is the substrate (e.g. its current state) after it has been processed by means of the process.

[0330] Example 22 is the process according to one of Examples 1 to 21, wherein the substrate is processed in a vacuum using the process.

[0331] Example 23 is the procedure according to one of Examples 1 to 22, wherein the first position output and the second position output have the same number of components.

[0332] Example 24 is the procedure according to one of Examples 1 to 23, where the first output represents a target state of the result of the process.

[0333] Example 25 is the procedure according to one of Examples 1 to 24, wherein the second position output represents a smaller deflection of the process than the first position output, for example, wherein the second position output has a smaller amount than the first position output.

[0334] Example 26 is the procedure according to one of Examples 1 to 25, wherein determining the first position output involves determining, for each component of the first position output, a target position variable of the actuator assigned to the component, which represents a target state of the result of the process.

[0335] Example 27 is the procedure according to Example 26, wherein determining the second position output involves mapping the target position variable to the second position output, taking into account the multiple limitations or at least one limitation, where, for example, each component of the first position output is mapped to (e.g., exactly) one component of the second position output.

[0336] Example 28 is the procedure according to Example 27, wherein mapping the setpoint variable involves projecting the setpoint variable onto a surface that spans one or more than one boundary of the multiple boundaries.

[0337] Example 29 is the procedure according to Example 27 or 28, wherein mapping the setpoint variable involves decorrelating the setpoint variable component-wise.

[0338] Example 30 is the procedure according to one of Examples 27 to 29, wherein the mapping has the feature of mapping the setpoint variable component-wise to the second position output.

[0339] Example 31 is the procedure according to one of Examples 27 to 30, wherein the mapping involves changing at least one component of the target control variable according to a limitation of the several limitations or according to the at least one limitation.

[0340] Example 32 is the procedure according to one of Examples 1 to 31, wherein, for determining the second position output, at least one limit (e.g., of the rate of change) is only conditionally taken into account, e.g., only if a comparison of the characteristic with one of the several limits (e.g., upper and / or lower limit) fulfills a criterion, whereby the criterion is fulfilled, for example, if the characteristic violates the limit.

[0341] Example 33 is the procedure according to one of Examples 1 to 32, wherein the characteristic is based on a scalar product, wherein the scalar product multiplies the first position output by a vector, where the vector, for example, has the first position output or a normal vector of a plane representing a boundary.

[0342] Example 34 is the procedure according to one of Examples 1 to 33, whereby the multiple limits are taken into account one after the other to determine the position output.

[0343] Example 35 is the procedure according to one of Examples 1 to 34, wherein the group of actuators are controlled simultaneously.

[0344] Example 36 is the method according to one of Examples 1 to 35, wherein the group of actuators comprises two or more actuators, e.g. 3 or more actuators, e.g. 5 or more actuators, e.g. 10 or more actuators, e.g. 20 or more actuators, e.g. 50 or more actuators, e.g. 100 or more actuators, e.g. 200 or more actuators.

[0345] Example 37 is a control device (e.g., having a processor) which (e.g., its processor) is configured to perform the procedure according to one of Examples 1 to 36.

[0346] Example 38 are code segments that are set up, when executed by a processor, to cause it to execute the procedure according to one of Examples 1 to 36.

[0347] Example 39 is a non-volatile storage medium that stores instructions (e.g., as code segments according to Example 38) that are set up, when executed by a processor, to cause it to perform the procedure according to one of Examples 1 to 36.

Claims

[1] Method (100) for operating a process by means of which a substrate is processed, the method (100) comprising: • Determine a first control output (Δy u ) based on a recorded (101) result of the process, • where the first control output (Δy u ) for each actuator of a group of actuators (858, 862) by means of which the process is controlled (103), a component is associated with the actuator; • Determine a second control output (Δy) based on the first control output (Δy u ) and based on a parameter Φ, which is a function of each component of the first control output (Δy u ) is; • Controlling the group of actuators (858, 862) according to the second control output (Δy). [2] Method (100) according to claim 1, wherein determining the second control output (Δy) comprises comparing the characteristic variable with at least one limitation of the control (103) of the process. [3] Method (100) according to claim 2, wherein the at least one limit per component comprises a lower limit of a control value and an upper limit of the control value. [4] Method (100) according to claim 2 or 3, wherein the at least one limitation is a limitation of a rate of change (Δy rate ) of the process. [5] Method (100) according to one of claims 1 to 4, wherein each component of the first control output (Δy u ) contains an indication of the value by which the at least one actuator assigned to the component is deflected. [6] Method (100) according to one of claims 1 to 5, wherein a spatial distribution is controlled (103) by means of the group of actuators (858, 862), with which the process is supplied with a material. [7] Method (100) according to one of claims 1 to 5, wherein the group of actuators (858, 862) controls (103) a spatial distribution of a magnetic field to which the process is exposed. [8] Method (100) according to one of claims 1 to 7, wherein the first control output (Δy u ) represents a target state of the result of the process. [9] Method (100) according to one of claims 1 to 8, wherein the second control output (Δy) has a smaller value than the first control output (Δy u ). [10] Method (100) for operating a process by means of which a substrate is processed, the method (100) comprising: • Determining a control output (Δy) based on a recorded result of the process, • controlling a group of actuators (858, 862) by means of which the process is controlled (103) according to the control output (Δy), wherein the control output (Δy) has a component for each actuator of the group of actuators (858, 862) according to which the actuator is controlled; • wherein, to determine the control output (Δy), a plurality of limitations of the control (103) of the process are taken into account, wherein a number of the plurality of limitations taken into account is greater than a number of the plurality of components. [11] Control device (224) configured to carry out the method (100) according to one of claims 1 to 10. [12] A non-volatile storage medium storing instructions arranged, when executed by a processor, to cause it to carry out the method (100) according to any one of claims 1 to 10.

Citation Information

Patent Citations

  • Controlling energy for reactive sputtering, comprises e.g. applying voltage between target and counter electrode, where energy supply takes place at constant voltage and controller receives actual energy and is supplied as control variable

    DE102012109093A1

  • Arrangement for control of lateral distribution of coating rate at magnetron sputtering in vacuum coating system, has a gas adjusting unit for adjusting gas flow from gas channel segment into vacuum chamber, which is set as an actuator

    DE102013101270A1