Method for determining the surface energy of a sample surface

Abstract

A method (200) for determining a surface energy of a sample surface (110) is disclosed, comprising: receiving spectral data (414) characterizing the sample surface (110), which represent a corresponding optical spectrum for each position of a plurality of positions (112) on the sample surface (110); determining spectral features at at least one position of the plurality of positions (112) using the spectral data (414); and determining a predicted contact angle (θ') representing the surface energy of the sample surface (110) at the at least one position using the spectral features.

Description

[0001] Several embodiments relate to a method for determining the surface energy of a sample surface.

[0002] In general, the surface energy of a sample surface can be specified by a contact angle between the sample surface and a liquid droplet applied to the sample surface.

[0003] In conventional contact angle measurement, a liquid droplet is applied to a specific position on the sample surface, and the contact angle between the sample surface and the liquid droplet is determined (e.g., by optical observation). This type of contact angle measurement has several disadvantages. For example, the liquid droplet can interact with or contaminate some samples, leading to alterations (e.g., damage). Furthermore, this method is very time-consuming, as a liquid droplet must be applied to each position to be examined, and the corresponding contact angle measured. Therefore, to investigate the lateral distribution of the contact angle on the sample surface, a comparatively large number of individual measurements are required, and the lateral resolution of this distribution is limited by the size of the liquid droplets.

[0004] The inventors recognized that (multi- or hyper-)spectral data of the sample surface enable the prediction of the contact angle. This spectral data can be acquired non-contact and non-destructively, without damaging the sample, as it is not necessary to apply a liquid droplet to the sample surface. Therefore, the effort required to determine the contact angle is significantly reduced, and it is possible to analyze a large number of positions on the sample surface with comparatively little effort. This allows the lateral distribution of the contact angle to be determined with significantly reduced effort and increased lateral resolution.

[0005] According to various embodiments, a method is therefore provided by which the surface energy of a sample surface can be determined without contact, non-destructively and with significantly reduced effort.

[0006] Furthermore, with conventional contact angle measurement it is not possible to measure the contact angle in-situ for various processes, whereas contact angle prediction using (multi- or hyper-) spectral data enables such in-situ analysis.

[0007] Several exemplary embodiments are given below. It is understood that aspects described in relation to a device may also apply to the methods, and vice versa. For example, the device may be configured to perform one or more of the methods.

[0008] Example 1 is a computer-implemented method for (e.g., in-situ) determining the surface energy of a sample surface of a (e.g., semiconductor) sample, comprising: receiving spectral data characterizing the sample surface, which represent a corresponding optical spectrum for each position of a plurality of (lateral) positions on the sample surface; determining (e.g., extracting) spectral features (e.g., represented by a spectral vector) at at least one position of the plurality of positions using the spectral data (e.g., using at least the optical spectrum associated with the at least one position); and determining a predicted contact angle, which represents the surface energy of the sample surface at the at least one position, using the spectral features.

[0009] Example 2 is set up according to Example 1, wherein the predicted contact angle is determined using a model that is set up to map spectral features to a predicted contact angle.

[0010] Example 3 is set up according to Example 2, where the model is a machine learning model.

[0011] Example 4 is set up according to Example 2, wherein the model has expected spectral features for each contact angle of a plurality of contact angles; and wherein determining the predicted contact angle involves comparing the spectral features (e.g. the spectral vector) at the at least one position of the sample surface with the spectral features expected for the plurality of contact angles.

[0012] Example 5 is set up according to one of Examples 1 to 4, wherein the spectral features are determined by one or more than one of the following (feature extraction) methods: principal component analysis, extraction (e.g. parameterization) of texture information, determination of a mean, determination of a standard deviation, and / or by means of an encoder (e.g., based on machine learning).

[0013] Example 6 is set up according to one of Examples 1 to 5, wherein the predicted contact angle is a prediction of the contact angle a liquid droplet (a drop of liquid) would have to the sample surface if the liquid droplet were applied at the at least one position on the sample surface.

[0014] Example 7 is set up according to one of Examples 1 to 6, wherein determining the spectral features at the at least one position of the sample surface comprises: determining respective spectral features at several positions of the plurality of positions on the sample surface; and wherein, at each position of the multiple positions, a predicted contact angle representing the surface energy at the position of the sample surface is determined using the respective spectral features.

[0015] Example 8 is the method according to Example 7, further comprising: determining a homogeneity value representing a homogeneity of surface energy at the sample surface, using the contact angle predicted at each of the multiple positions.

[0016] Example 9 is a method for controlling a (e.g., physical or chemical) process (e.g., a manufacturing process), comprising the following: generating (e.g., forming and / or modifying) a surface structure according to first process parameters; acquiring spectral data which, for each position of a plurality of (lateral) positions on a sample surface of the generated surface structure, represent a corresponding optical spectrum; determining at least one predicted contact angle (e.g., a respective predicted contact angle at several positions of the sample surface) using the method according to one of Examples 1 to 8; determining second process parameters (e.g., different from the first process parameters) using the at least one predicted contact angle; and modifying the surface structure according to the second process parameters (e.g.,an additional cleaning process) and / or the creation of a different surface structure (e.g. on a different substrate) according to the second process parameters.

[0017] Example 10 is set up according to Example 9, wherein the surface structure is generated on a substrate in a process chamber (e.g. vacuum chamber) and wherein the spectral data are acquired while the substrate is arranged in the process chamber (e.g. vacuum chamber).

[0018] Example 11 is set up according to Example 9 or 10, wherein generating the surface structure involves forming the surface structure (e.g. by means of a coating process) according to the first process parameters; or wherein generating the surface structure involves processing a layer (e.g. its surface) by means of a layer processing process (e.g. a cleaning process, an etching process, etc.) according to the first process parameters.

[0019] Example 12 is a method for generating a model comprising: for each position of a plurality of positions on a sample surface of one or more samples, determining a respective contact angle by applying a liquid droplet to the sample surface and determining the respective contact angle of the liquid droplet to the sample surface; acquiring spectral data which, at least for each position of the plurality of positions, represent a corresponding optical spectrum; for each position of the plurality of positions, determining (e.g., extracting) respective spectral features (e.g., represented by a respective spectral vector) using the spectral data (e.g.,using at least the optical spectrum associated with the position); and generating the model using the respective contact angle and spectral features of each position of the plurality of positions such that the model maps spectral features to a predicted contact angle.

[0020] Example 13 is set up according to Example 12, wherein the model has a correlation matrix that maps spectral features to the predicted contact angle.

[0021] Example 14 is set up according to Example 12, wherein the model is a machine learning model which is set up to output a value of the predicted contact angle in response to an input of spectral features; and wherein the generation of the model comprises: for each position of the plurality of positions, training the model using the respective spectral features such that the model outputs the respective contact angle as the predicted contact angle in response to an input of the respective spectral features.

[0022] Example 15 is a method for training a machine learning model, comprising: for each position of a plurality of (lateral) positions on a sample surface, determining a respective contact angle by applying a liquid droplet to the position on the sample surface and determining the respective contact angle of the liquid droplet to the sample surface; acquiring spectral data which represent a corresponding optical spectrum for at least each position of the plurality of positions; generating, in response to an input of the spectral data into the machine learning model, a lateral distribution of a predicted contact angle on the sample surface by means of the machine learning model;Adapting the machine learning model based on a comparison between the lateral distribution of the predicted contact angle and the contact angles determined at the multitude of positions on the sample surface.

[0023] Example 16 is a device comprising: a process chamber (e.g., a vacuum chamber); a control device configured to carry out the method according to any one of Examples 1 to 8; at least one radiation source configured to irradiate the sample arranged in the process chamber (e.g., a vacuum chamber) with electromagnetic radiation; and at least one (e.g., hyper-)spectral camera configured to record the intensity of the electromagnetic radiation scattered and / or reflected by the sample as spectral data.

[0024] Example 17 is set up according to Example 16, wherein the control device is set up to carry out the method according to one of Examples 8 to 11; and wherein the device further comprises a coating unit for forming the surface structure and / or a processing unit for processing the surface structure.

[0025] Example 18 is a computer-readable medium (e.g., a computer program product, a non-volatile storage medium, a non-transitory storage medium, a non-volatile storage medium, etc.) which stores instructions which, when executed by a processor, cause the processor to perform the procedure according to one of Examples 1 to 11.

[0026] Example 19 is a computer program that stores instructions which, when executed by a processor, cause the processor to perform the procedure according to one of Examples 1 to 11.

[0027] They show Fig. 1 a schematic representation of a contact angle; Fig. 2 a schematic flowchart of a method for determining the surface energy of a sample surface according to different embodiments; Fig. 3. An exemplary representation of spectral data as a data cube; Fig. 4 a top view of the surface of an exemplary wafer with a multitude of positions on the surface, at each of which a respective surface energy of the surface can be determined according to various aspects; and Fig. 5 A schematic flowchart for controlling a process according to different embodiments.

[0028] The following detailed description refers to the accompanying drawings, which form part of this document and in which specific embodiments of the invention are shown for illustration.

[0029] Several examples are described in more detail below.

[0030] Fig. Figure 1 shows a schematic representation 100 of a contact angle θ (also called dihedral angle, contact angle, or wetting angle). To measure the contact angle θ, a liquid droplet 108 can conventionally be applied to a sample surface 110 of a sample 102, and (as shown) the contact angle θ can then be determined by lateral observation.

[0031] The sample 102 can generally have a surface structure 106 on the sample surface 110. The surface structure 106 can be arranged on a substrate 104 (which can have one or more layers, structures, etc.).

[0032] The surface structure 106 can be any type of structure on the surface 110 of the sample 102. In some aspects, the surface structure 106 can have one (e.g., planar) layer (or several laterally arranged layers). The surface structure 106 can have several laterally arranged structures, which can optionally consist of different materials.

[0033] For example, the sample can be a semiconductor sample (e.g., a (semiconductor) wafer) in any (e.g., backend) manufacturing step. To illustrate various aspects, the semiconductor sample is used as an example in several embodiments below. It is understood that this serves to illustrate these aspects and their advantages, and that the sample 102 can also be any other type of sample from which a contact angle θ is to be determined.

[0034] The contact angle θ can characterize the surface energy of the sample surface 110 at the investigated position.

[0035] The liquid droplet 108 can, for example, be a water droplet. In this case, the surface energy described herein can, for example, characterize the hydrophobicity of the sample surface 110 at the investigated position. Thus, a contact angle θ < 90° can represent a hydrophobic sample surface 110, a contact angle θ ≈ 90° an amphiphilic sample surface 110, and a contact angle θ > 90° a hydrophilic sample surface 110. It is understood that water is merely an example and that the liquid droplet 108 can also be a droplet of any other liquid, such as an organic liquid (e.g., diiodomethane).

[0036] According to various aspects disclosed herein, this contact angle θ can also be predicted without applying a liquid droplet 108 to the sample 102, using (multi- or hyper-) spectral data of the sample surface 110.

[0037] Fig. Figure 2 shows a schematic flowchart of a method 200 for determining a surface energy of a sample surface according to different embodiments.

[0038] Method 200 can (in 202) include receiving spectral data characterizing the sample surface.

[0039] The spectral data can represent a corresponding optical spectrum for each position of a multitude of lateral positions on the sample surface 110. For example, the spectral data can characterize the entire sample surface 110 with a corresponding optical spectrum for each point on the sample surface 110 according to a lateral resolution.

[0040] The optical spectrum associated with a given position can exhibit a recorded intensity for a multitude of wavelengths. In some aspects, the spectral data can be multispectral, exhibiting a recorded intensity for a multitude of wavelength bands (e.g., three to ten). In this case, the optical spectrum associated with a given position can be a discrete optical spectrum. In other aspects, the spectral data can be hyperspectral, exhibiting a continuous optical spectrum (with intensity as a function of wavelength) for each position within the multitude of lateral positions. In the case of hyperspectral data, the accuracy of the contact angle prediction disclosed herein can be increased. The number of distinguishable wavelengths or wavelength bands can also be referred to as spectral channels.Depending on various aspects, the number of spectral channels can be greater than or equal to 100.

[0041] Fig. Figure 3 shows an exemplary representation of spectral data as a data cube 300 (also called a hypercube). Here, the data cube 300 can exhibit a corresponding (discrete or continuous) optical spectrum for each position (x*, y*) (in the direction of wavelength λ). Visually, the data cube 300 can form a three-dimensional data structure consisting of one wavelength-resolved dimension (λ) and two position-resolved dimensions (x, y). Within this structure, the * notation can define a specific number for the corresponding variable, e.g., a specific x* value for the variable x. It is understood that if the sample 102 is, for example, a wafer (see, for example, Fig. 4) some of the x and y values ​​may be assigned to sample 102 and other parts of the x and y values ​​may not be assigned to sample 102.

[0042] The spectral data can be acquired in various ways (e.g., different spectral imaging techniques). The method disclosed herein for determining the surface energy can be independent of how the spectral data are or were acquired.

[0043] For example, the surface of sample 102 (110) can be irradiated with (at least) one radiation source (e.g., a broadband radiation source, also called an illumination unit) (with a predefined wavelength range), and a spectral camera can record the intensity of the electromagnetic radiation scattered and / or reflected by the surface of sample 102 as spectral data (wavelength-resolved). The spectral camera can be configured, for example, to record a specific optical spectrum along a line (e.g., for all x-values ​​at a specific y*-value). To record the respective optical spectrum for all y-values, sample 102 and the spectral camera can be moved relative to each other (e.g., the spectral camera can be moved and / or sample 102 can be moved).

[0044] The wavelength range can depend on how the spectral data 204 are or were acquired. For example, every optical spectrum can have a wavelength range from approximately 400 nm to approximately 1000 nm.

[0045] Method 200 (in 204) can involve determining (e.g., extracting) spectral features at at least one position (x*, y*) using spectral data. The spectral features can be determined, for example, from the optical spectrum associated with the at least one position. The spectral features can be represented by a spectral vector.

[0046] The spectral features can be determined using a feature extraction method or a combination of several feature extraction methods. Examples of feature extraction methods for spectral features include principal component analysis, extraction (e.g., parameterization) of texture information, determination of static parameters (such as a mean and / or standard deviation), an encoder (e.g., based on machine learning), etc.

[0047] Method 200 (in 206) can include determining a predicted contact angle θ' using spectral features (e.g., using the spectral vector). The predicted contact angle θ' can indicate the contact angle θ that the liquid droplet 108 would have with the sample surface 110 if the liquid droplet 108 were applied at the at least one position (x*, y*) on the sample surface 110. Intuitively, the predicted contact angle θ' can represent the surface energy of the sample surface 110 at the at least one position (x*, y*).

[0048] According to various aspects, (in 204) respective spectral features can be determined for several positions (e.g., each position on the sample surface 110), and (in 206) a corresponding predicted contact angle θ' can then be determined for each of these several positions using the respective spectral features. In this way, a lateral distribution of the predicted contact angle θ' can be determined. This shows Fig. 4 To illustrate, a wafer as sample 102 with several positions 112 on its sample surface 110.

[0049] Optionally, the method 200 can also include determining a homogeneity value using the contact angle predicted at each of the several positions. This homogeneity value can represent a homogeneity of the surface energy at the sample surface 110.

[0050] The procedure 200 of Fig. 2. This can be performed by one or more computers with one or more data processing units. The term "data processing unit" can be understood as any type of entity that enables 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) special function performed by the data processing unit. A data processing unit can include or be composed of an analog circuit, a digital circuit, a logic circuit, a microprocessor, a microcontroller, a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), an integrated circuit, a programmable gate array (FPGA), or any combination thereof.Any other way of implementing the respective functions described in more detail herein can also be understood as a data processing unit or logic circuit arrangement. One or more of the process steps described in detail here can be executed (e.g., implemented) by a data processing unit through one or more special functions performed by the data processing unit. Thus, according to various interpretations, the process is specifically computer-implemented.

[0051] To determine the predicted contact angle θ' (in 206), the data processing unit can, for example, implement a model that is set up to map spectral features to a predicted contact angle θ'.

[0052] In some aspects, the model can exhibit expected spectral features for each contact angle θ of a multitude of contact angles. The data processing unit can then determine the predicted contact angle θ' by comparing the determined spectral features with the spectral features expected for the multitude of contact angles. In this case, the predicted contact angle θ' can, for example, be the contact angle to whose expected spectral features the determined spectral features show the greatest similarity (according to a predefined similarity metric). Visually, the model can include a database of expected spectral features as a function of the contact angle. For example, the model can include a correlation matrix that maps spectral features to the predicted contact angle.

[0053] In other aspects, the model can be a machine learning-based model (also referred to as a machine learning model). The machine learning model can be configured to output a value of the predicted contact angle θ' in response to an input of spectral features.

[0054] To generate the model, a corresponding contact angle can be determined for a large number of positions according to conventional contact angle measurement (see Fig. 1) can be measured and the respective optical spectrum can be recorded.

[0055] These positions can be different locations on the surface of a single sample or even multiple samples. Based on the respective optical spectrum, the specific spectral features can be determined for each position. Ideally, data can be obtained that exhibit a bijective mapping between spectral features and expected contact angles. It is understood that this involves using positions (e.g., creating corresponding samples) for which different contact angles are expected. For example, a sample can be created that exhibits a surface energy gradient (e.g., a hydrophobicity gradient).

[0056] The model can then be generated using the respective contact angle and spectral characteristics of each position among the multitude of positions, such that these spectral characteristics map to the predicted contact angle. In this way, for example, the correlation matrix can be generated or the machine learning model can be trained accordingly. The assignment of spectral characteristics to predicted contact angles (or contact angle ranges) can be performed, for example, using cluster analysis.

[0057] In some aspects, the machine learning model can also be trained to directly map the spectral data (in response to spectral input into the machine learning model) to predicted contact angles (e.g., a lateral distribution of the predicted contact angle on the sample surface). However, significantly smaller amounts of data are required to generate (e.g., train) the machine learning model if it is trained on previously extracted spectral features.

[0058] A model (e.g., a machine learning-based model (also referred to as a machine learning model)) may, for example, be a reinforcement learning model (e.g., using Q-Learning, Temporal Difference (TD), Deep Adversarial Networks, etc.) and / or a classification model (e.g., a linear classifier (e.g., a logistic regression classifier or a Naive Bayes classifier), a support vector machine, a decision tree, a boosted tree classifier, a random forest classifier, a neural network, or a nearest neighbor model). A neural network can be or exhibit any type of neural network, such as a convolutional neural network (CNN), a variational autoencoder network (VAE), a thinned autoencoder network (VAE), or a sparsely populated autoencoder network.: sparse autoencoder network (SAE), a recurrent neuroanal network (RNN), a deconvolutional neural network (DNN), a generative adversarial network (GAN), a forwardthinking neural network, a sum-product neural network, etc. According to various aspects, the model can be a statistical model.

[0059] In various aspects, it is explained that the model can map spectral features to the predicted contact angle. It is understood that the predicted contact angle is not necessarily a single, fixed value, but rather that the model can assign spectral features to several different contact angle ranges. In a simple case, the model can, for example, use the spectral features to determine whether the contact angle θ < 90° (e.g., in the example case of a water droplet, whether the sample surface is hydrophobic), whether the contact angle θ ≈ 90° (e.g., in the example case of a water droplet, whether the sample surface is amphiphilic), or whether the contact angle θ > 90° (e.g., in the example case of a water droplet, whether the sample surface is hydrophilic). It is understood that this is just one example and that arbitrary ranges can be defined.

[0060] As explained herein, the method 200 described herein, for example, enables in-situ contact angle prediction. To illustrate this, Figure 200 shows... Fig. 5 A schematic flowchart for controlling a process according to different embodiments.

[0061] An arrangement 400 can include a process chamber 404 in which the process takes place. The process can be any type of technical process (e.g., a physical or chemical process). The process chamber 404 can, for example, be a vacuum chamber. A negative pressure (e.g., a vacuum), as described herein, can be a pressure in the range of approximately 10 mbar to approximately 1 mbar (in other words, a rough vacuum) or less, for example, a pressure in the range of approximately 1 mbar to approximately 10 -3 mbar (in other words, a fine vacuum) or less, for example, a pressure in the range of approximately 10 -3 mbar to approximately 10 -7mbar (in other words, a high vacuum) or less, for example, a pressure in the range of less than 10 -7 mbar (in other words, an ultra-high vacuum). Optionally, process chamber 404 can, for example, be a continuous flow system.

[0062] To carry out the process, the process chamber 404 can include a corresponding processing unit 410. The processing unit 410 can be configured to generate the surface structure 106. In some aspects, the surface structure 106 can be generated by forming a layer. In this case, the processing unit 410 can, for example, be a coating unit. In other aspects, the surface structure 106 can be generated by processing a layer. In this case, the processing unit 410 can, for example, be a layer processing unit. Processing a layer can include, for example, etching the layer, cleaning the layer, etc.

[0063] The arrangement 400 can include a control device 402. The control device 402 can include the data processing unit for performing the method 200. For example, the control device 402 can include a memory in which data for implementing the model is stored. A memory used in the embodiments can be volatile memory, for example, a DRAM (dynamic random-access memory), or non-volatile memory, for example, a PROM (programmable read-only memory), an EPROM (erasable PROM), an EEPROM (electrically erasable PROM), or flash memory, such as a floating-gate memory device, a charge-swapping memory device, an MRAM (magnetoresistive random-access memory), or a PCRAM (phase-change random-access memory).

[0064] The control device 402 can implement a control model 406. The control model 406 can be configured to generate process parameters 408 and control the processing unit 410 according to the process parameters 408. For example, the control model 406 can be configured to determine appropriate control parameters for controlling the processing unit 410 based on the process parameters 408.

[0065] The process chamber 404 may further include a spectral data acquisition unit 412. The spectral data acquisition unit 412 may be configured to acquire the spectral data described herein. For example, the spectral data acquisition unit 412 may include at least one radiation source and at least one (e.g., hyper)spectral camera. The process chamber 404 may also include a unit that enables relative movement between the sample 102 and the spectral camera. This movement may be provided, for example, by the transport device on which the sample (e.g., a wafer) is transported in a continuous flow system.

[0066] According to various aspects, the control device 402 can control the spectral data acquisition unit 412 to acquire the spectral data 414 after the creation of the surface structure 106.

[0067] These spectral data 414 can then be entered into the control model 406 of the control device 402. The control model 406 can be configured, according to method 200, to predict the contact angle θ' at at least one position (e.g., at several positions) of the sample surface 110 and can then adjust the process parameters 408 using the predicted contact angle θ'. In some aspects, the process parameters 408 can remain unchanged (e.g., if the predicted contact angle θ' corresponds to a desired contact angle). In other aspects, the process parameters 408 can be changed (e.g., if the predicted contact angle θ' does not correspond to the desired contact angle).

[0068] For example, the process parameters 408 can be adjusted to perform post-processing of the sample surface 110 to achieve the desired contact angle. The process parameters 408 can also be adjusted to achieve the desired contact angle for a subsequent execution of the process (e.g., to generate the surface structure on a different substrate).

[0069] It is clearly demonstrated that, for example, Method 200 enables in-situ determination of the surface energy during the processing of (semiconductor) wafers and therefore also the immediate detection of process defects that lead to a change in the contact angle. Compared to conventional (ex-situ) contact angle measurement, the method disclosed herein does not require the wafer to be removed from the process chamber. Furthermore, the time lag between the fabrication of the surface structure and the detection of potential defects is significantly reduced, so that the process parameters can be adjusted for the immediately following wafer. This reduces the number of defective wafers and thus significantly lowers costs. Method 200 disclosed herein is also more reliable than conventional (ex-situ) contact angle measurement.

Claims

[1] Computer-implemented method (200) for determining a surface energy of a sample surface (110) of a sample (102), comprising the method (200): • Receiving (202) spectral data (414) characterizing the sample surface (110), which represent a corresponding optical spectrum for each position of a plurality of positions (112) on the sample surface (110); • Determining (204) spectral features at at least one position of the plurality of positions (112) using the spectral data (414); and • Determine (206) a predicted contact angle (θ') representing the surface energy of the sample surface (110) at at least one position using the spectral features. [2] Method (200) according to claim 1, wherein the predicted contact angle (θ') is determined using a model which is configured to map spectral features to a predicted contact angle. [3] Method (200) according to claim 2, where the model is a machine learning model; or wherein the model has expected spectral features for each contact angle of a plurality of contact angles, and wherein determining the predicted contact angle involves comparing the spectral features at the at least one position of the sample surface (110) with the spectral features expected for the plurality of contact angles. [4] Method (200) according to any one of claims 1 to 3, wherein determining the spectral features at the at least one position of the sample surface (110) comprises: determining respective spectral features at several positions (112) of the plurality of positions (112) on the sample surface (110); and wherein at each of the multiple positions (112) a predicted contact angle, which represents the surface energy at the position of the sample surface (110), is determined using the respective spectral features. [5] Method (200) according to claim 4, further comprising: • Determining a homogeneity value representing a homogeneity of surface energy at the sample surface (110) using the contact angle predicted at each of the multiple positions (112). [6] Procedure for controlling a process, comprising the procedure: • Generating a surface structure (106) according to first process parameters; • Acquisition of spectral data (414) which represent a corresponding optical spectrum for each position of a plurality of positions (112) on a sample surface (110) of the generated surface structure (106). • Determining at least one predicted contact angle (θ') using the method (200) according to any one of claims 1 to 5; • Determining second process parameters using at least one predicted contact angle; and • Processing the surface structure (106) according to the second process parameters and / or creating a different surface structure (106) according to the second process parameters. [7] Method according to claim 6, wherein the surface structure (106) is produced in a process chamber (404) on a substrate (104) and wherein the spectral data (414) are acquired while the substrate (104) is arranged in the process chamber (404). [8] Method (200) according to claim 6 or 7, wherein generating the surface structure (106) comprises forming the surface structure (106) by means of a coating process according to the first process parameters; or wherein the creation of the surface structure (106) involves processing a layer by means of a layer processing process according to the first process parameters. [9] Device comprising: • a trial chamber (404); • a control device (402) configured to perform the method (200) according to any one of claims 1 to 5; • at least one radiation source configured to irradiate the sample (102) arranged in the process chamber (404) with electromagnetic radiation; and • at least one spectral camera set up to record the intensity of the electromagnetic radiation scattered and / or reflected by the sample (102) as spectral data (414). [10] Device according to claim 9, • wherein the control device (402) is configured to perform the method according to any one of claims 6 to 8; and • wherein the device further comprises a coating unit for forming the surface structure (106) and / or a layer processing unit for processing the layer.

Images