A method and system for controlling the densification of inkjet-printed films

Abstract

This application relates to a method and system for densification control of inkjet-printed thin films. In this application, by constructing a predictive model between laser process parameters and the microstructure state of the thin film, and combining it with closed-loop feedback control, online, dynamic, and precise control of the microstructure state of the inkjet-printed thin film is achieved. This significantly improves the densification quality and process stability of the inkjet-printed thin film, and enables predictable and repeatable control of the microstructure state of the thin film under low ambient temperature conditions. This effectively solves the problems of local overheating, poor controllability, and low repeatability in traditional heat treatment and open-loop laser sintering, and is particularly suitable for fields such as flexible electronics and high-temperature sensors where the consistency of thin film performance is critical.

Description

Technical Field

[0001] This application belongs to the field of thin film preparation and microstructure control technology, and in particular relates to a method and system for densification control of inkjet printed thin films. Background Technology

[0002] Inkjet printing technology has attracted much attention in the field of functional thin film preparation due to its advantages such as non-contact, patternability, and suitability for flexible substrates. However, thin films obtained by inkjet printing usually have problems such as loose structure and insufficient densification, requiring subsequent heat treatment to improve their performance.

[0003] While the currently commonly used overall heat treatment methods (such as hot plate sintering and conventional hot annealing) can achieve a certain degree of densification, they have obvious limitations: First, the required ambient temperature is high (such as above 400°C), making it difficult to be compatible with temperature-sensitive flexible substrates; second, the process is uncontrollable, making it impossible to precisely control the microstructure of the film, resulting in uneven film performance and poor repeatability.

[0004] In recent years, laser sintering has been introduced into this field as a localized, precise energy input method in order to achieve local densification at low ambient temperatures (such as below 400°C). However, existing laser sintering processes mostly adopt an open-loop mode of "preset parameters - process execution - offline detection". This mode cannot overcome problems such as fluctuations in the initial state of the thin film and the complex and unpredictable relationship between process parameters and the final structure, resulting in unstable processing quality and a lack of effective online process control methods. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for densification control of inkjet-printed films. The method and system for densification control of inkjet-printed films provided by this application realize online, dynamic and precise control of the microstructure state of inkjet-printed films, which can significantly improve the densification quality and process stability of inkjet-printed films, and achieve predictable and repeatable control of the microstructure state of films under low ambient temperature conditions.

[0006] This application provides a method for densification control of inkjet-printed films, including:

[0007] A prediction model is established to characterize the relationship between laser process parameters, intermediate physical states, and the microstructure state of the thin film.

[0008] The current laser process parameters are determined based on the preset microstructure state.

[0009] Based on the current laser process parameters and the prediction model, the corresponding microstructure state prediction value is obtained.

[0010] The predicted microstructure state is compared with the preset microstructure state, and the laser process parameters are adjusted based on the comparison results.

[0011] Based on the adjusted laser process parameters, predictions and comparisons are performed again until the deviation between the newly obtained microstructure state prediction value and the preset microstructure state meets the preset conditions.

[0012] Optionally, the prediction model includes a first prediction sub-model and a second prediction sub-model. The step of predicting the corresponding microstructure state based on the current laser process parameters and the prediction model includes:

[0013] Based on the current laser process parameters and the first prediction sub-model, a prediction of the corresponding intermediate physical state is obtained.

[0014] Based on the predicted intermediate physical state, the current laser process parameters, and the second prediction sub-model, the corresponding microstructure state prediction value is obtained.

[0015] Optionally, the laser process parameters include linear energy density, which is obtained based on laser power and scanning speed.

[0016] Optionally, the intermediate physical state is the maximum melt depth formed within the thin film during laser sintering.

[0017] Optionally, the microstructure state is porosity.

[0018] Optionally, comparing the predicted microstructure state with the preset microstructure state and adjusting the laser process parameters based on the comparison result includes:

[0019] The prediction error is obtained by comparing the predicted microstructure state with the preset microstructure state.

[0020] The laser power and / or the scanning speed are adjusted according to the magnitude and direction of the prediction error to adjust the linear energy density.

[0021] Optionally, the deviation between the re-obtained predicted microstructure state and the preset microstructure state satisfies a preset condition, including:

[0022] The absolute value of the deviation between the re-obtained microstructure state prediction value and the preset microstructure state is less than the preset threshold.

[0023] Optionally, establishing the prediction model includes:

[0024] At least one set of intermediate physical state calibration measured values ​​are obtained, which are obtained by laser sintering experiments on thin films on an experimental platform using at least one set of laser process parameters.

[0025] Input the at least one set of laser process parameters into the numerical simulation model to obtain the corresponding intermediate physical state calibration simulation values;

[0026] The error between the measured calibration value and the calibration simulation value is compared. If the error exceeds the preset error threshold, the parameters in the numerical simulation model are adjusted, and the calibration simulation value is recalculated based on the adjusted numerical simulation model for comparison, until the error is less than the preset error threshold, and the calibrated numerical simulation model is obtained.

[0027] Using the calibrated numerical simulation model, the intermediate physical states under multiple different laser process parameters are simulated and calculated to obtain multiple intermediate physical state simulation values.

[0028] The coefficients of the first prediction sub-model are obtained by fitting multiple laser process parameters and multiple intermediate physical state simulation values.

[0029] Obtain measured values ​​of the microstructure state of thin films after laser treatment under multiple different laser process parameters;

[0030] The coefficients of the second prediction sub-model are obtained by fitting multiple laser process parameters, multiple intermediate physical state simulation values, and multiple microstructure state calibration measured values.

[0031] Optionally, the method further includes:

[0032] Offline measured values ​​of the microstructure state of the laser-treated thin film were obtained.

[0033] Based on the laser process parameters corresponding to the offline measured values ​​of the microstructure state, the corresponding intermediate physical state reference values ​​are obtained using numerical simulation or the first prediction sub-model.

[0034] Based on the offline measured values ​​of the microstructure state, the corresponding laser process parameters, and the corresponding intermediate physical state reference values, the coefficients of the second prediction sub-model are updated to obtain the optimized second prediction sub-model.

[0035] This application also provides a densification control system for inkjet-printed films, comprising:

[0036] A module is established to build a prediction model, which is used to characterize the relationship between laser process parameters, intermediate physical states and the microstructure state of the thin film.

[0037] The determination module is used to determine the current laser process parameters based on the preset microstructure state;

[0038] The prediction module is used to make predictions based on the current laser process parameters and the prediction model to obtain the corresponding microstructure state prediction values.

[0039] The first adjustment module is used to compare the predicted microstructure state with the preset microstructure state, and adjust the laser process parameters based on the comparison result.

[0040] The second adjustment module is used to re-predict and compare based on the adjusted laser process parameters until the deviation between the newly obtained microstructure state prediction value and the preset microstructure state meets the preset conditions.

[0041] Compared with existing technologies, this application provides a method and system for densification control of inkjet-printed thin films. This method establishes a prediction model to characterize the correlation between laser process parameters, intermediate physical states, and the microstructure state of the thin film. Based on a preset microstructure state, the current laser process parameters are determined. A prediction is then made based on these parameters and the prediction model to obtain the corresponding predicted microstructure state value. The predicted microstructure state is compared with the preset microstructure state, and the laser process parameters are adjusted based on the comparison results. The prediction and comparison are repeated based on the adjusted laser process parameters until a new predicted microstructure state value is obtained. The deviation between the microstructure state and the preset microstructure state meets the preset conditions. In this application, by constructing a predictive model between laser process parameters and the microstructure state of the thin film, and combining it with closed-loop feedback control, the online, dynamic, and precise control of the microstructure state of inkjet-printed thin films is realized. This significantly improves the densification quality and process stability of inkjet-printed thin films, and enables predictable and repeatable control of the microstructure state of the thin film under low ambient temperature conditions. This effectively solves the problems of local overheating, poor controllability, and low repeatability in traditional heat treatment and open-loop laser sintering, and is especially suitable for fields such as flexible electronics and high-temperature sensors where the consistency of thin film performance is critical. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1This is a schematic flowchart of a method for controlling the densification of inkjet-printed films disclosed in an embodiment of this application;

[0044] Figure 2 This is a schematic diagram of numerical simulation results under different combinations of power and speed disclosed in the embodiments of this application;

[0045] Figure 3 This is a schematic diagram illustrating the physical meaning of the maximum melting depth disclosed in the embodiments of this application;

[0046] Figure 4 This is a schematic diagram of the densification control system for inkjet-printed films disclosed in an embodiment of this application. Detailed Implementation

[0047] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0048] It should be noted that when a component is referred to as being "fixed to" or "set on" another component, it can be directly on or indirectly set on the other component; when a component is referred to as being "connected to" another component, it can be directly connected to or indirectly connected to the other component.

[0049] It should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0050] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "a plurality of" or "several" means two or more, unless otherwise explicitly specified.

[0051] It should be noted that the structures, proportions, sizes, etc., shown in the accompanying drawings of this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed in the specification, and are not intended to limit the conditions under which this application can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size should still fall within the scope of the technical content disclosed in this application, provided that they do not affect the effects and purposes that this application can produce.

[0052] like Figure 1 As shown, this application provides a method for controlling the densification of inkjet-printed films, including:

[0053] S11. Establish a prediction model, which is used to characterize the relationship between laser process parameters, intermediate physical states and the microstructure state of the thin film.

[0054] In this embodiment, the laser process parameters can be linear energy density E, which can be obtained based on laser power P and scanning speed v; the intermediate physical state can be the maximum melt depth MPD formed in the film during laser sintering; and the microstructure state can be porosity Φ.

[0055] In this embodiment, the correlation can be specifically manifested as follows: there is a first correlation between the laser process parameters and the intermediate physical state, and a second correlation between the intermediate physical state, the laser process parameters, and the microstructure state.

[0056] The prediction model can be established by designing calibration experiments covering different combinations of laser power and scanning speed, obtaining corresponding actual porosity data, and using a regression method that includes physical constraints for fitting, so that the prediction model can reliably deduce the microstructure state of the thin film from the process parameters, thereby providing a basis for subsequent online control.

[0057] The prediction model constructed in this embodiment is essentially a physical-guided data-driven model that integrates theoretical simulation and experimental data. It avoids the lack of interpretability of pure black-box models and overcomes the limitation of pure physical models in accurately describing the complexity of actual processes. Thus, it achieves predictable and interpretable online estimation of film porosity, providing a reliable state feedback signal for subsequent closed-loop control.

[0058] S12. Determine the current laser process parameters based on the preset microstructure state;

[0059] In this embodiment, the implementation logic of this step is as follows: taking the preset microstructure state (target porosity) as the control target, and combining the prediction model, a set of laser process parameters (laser power and scanning speed) that are expected to make the final microstructure state of the thin film approach the target are calculated through model inverse solution or iterative optimization algorithm, and used as the parameter basis for the current control cycle.

[0060] In other embodiments, the current laser process parameters can also be determined in other ways, such as by matching a parameter lookup table corresponding to the target porosity based on historical process data; or by directly calculating the parameters by back-calculation based on the simplified relationship of the prediction model. The parameters obtained in these ways can be used as the initial values ​​for closed-loop control, or used independently in scenarios where the control accuracy requirements are relatively relaxed.

[0061] Regardless of how the initial parameters are determined, the system will enter a subsequent closed-loop comparison and adjustment process based on the prediction model to ensure that the final process parameters meet the accuracy requirements.

[0062] S13. Based on the current laser process parameters and prediction model, make predictions to obtain the corresponding microstructure state prediction values;

[0063] In this embodiment, the prediction process is implemented by substituting the current laser process parameters determined in step S12 into the prediction model to calculate the corresponding microstructure state prediction value.

[0064] Specifically, the process involves two successive mapping calculations:

[0065] 1. Calculate the linear energy density: First, calculate the linear energy density based on the laser power and scanning speed. This parameter serves as the energy input characterization of the prediction model.

[0066] 2. Perform model prediction: Subsequently, the calculated linear energy density is input into the prediction model. Based on the established correlation, the prediction model deduces the predicted value of the corresponding intermediate physical state (e.g., maximum melting depth) based on the input linear energy density value, and finally calculates the predicted value of the target microstructure state (e.g., porosity).

[0067] The predicted microstructure state output in this step acts as a real-time reading of a "virtual sensor." It does not rely on offline or destructive microscopic detection, but rather estimates the microstructure state of the thin film in real time through model calculation based solely on online process parameters before or during laser processing. This provides crucial online state feedback signals for subsequent real-time comparison and closed-loop control.

[0068] S14. Compare the predicted microstructure state with the preset microstructure state, and adjust the laser process parameters based on the comparison results.

[0069] In this embodiment, the "comparison and adjustment" process constitutes an online control loop. The predicted microstructure state (such as the predicted porosity value) is compared with a preset target value, and the deviation between the two is calculated. Based on the direction and magnitude of this deviation, the system automatically adjusts the laser process parameters. Its core control logic is: when the predicted microstructure density is lower than the target, the system increases the laser energy input; conversely, it decreases the laser energy input. Specifically, this can be achieved by adjusting the laser power or scanning speed. This adjustment process aims to drive the predicted value towards the preset target and compensate for uncertainties that may occur during processing.

[0070] S15. Based on the adjusted laser process parameters, perform prediction and comparison again until the deviation between the newly obtained microstructure state prediction value and the preset microstructure state meets the preset conditions.

[0071] In this embodiment, the control process is an iterative optimization closed loop. After adjusting the laser process parameters according to S14, the system restarts a control cycle based on the new parameters: that is, it predicts the microstructure state again using the updated parameters (S13), and compares the new predicted value with the preset target again (S14). This "prediction-comparison-adjustment" cycle continues until the deviation between the latest microstructure state prediction value and the preset target value falls within the preset allowable error range, or the preset maximum number of iterations is reached. When any termination condition is met, the system determines that the current process parameters are the optimized parameters that can achieve the target microstructure, and locks these parameters for laser sintering. Through this iterative closed loop, this method achieves high-precision, adaptive online control of the thin film microstructure, ensuring the repeatability and stability of the process results.

[0072] Compared with existing technologies, this application provides a method and system for densification control of inkjet-printed thin films. This method establishes a prediction model to characterize the correlation between laser process parameters, intermediate physical states, and the microstructure state of the thin film. Based on a preset microstructure state, the current laser process parameters are determined. A prediction is then made based on these parameters and the prediction model to obtain the corresponding predicted microstructure state value. The predicted microstructure state is compared with the preset microstructure state, and the laser process parameters are adjusted based on the comparison results. The prediction and comparison are repeated based on the adjusted laser process parameters until a new predicted microstructure state value is obtained. The deviation between the microstructure state and the preset microstructure state meets the preset conditions. In this application, by constructing a predictive model between laser process parameters and the microstructure state of the thin film, and combining it with closed-loop feedback control, the online, dynamic, and precise control of the microstructure state of inkjet-printed thin films is realized. This significantly improves the densification quality and process stability of inkjet-printed thin films, and enables predictable and repeatable control of the microstructure state of the thin film under low ambient temperature conditions. This effectively solves the problems of local overheating, poor controllability, and low repeatability in traditional heat treatment and open-loop laser sintering, and is especially suitable for fields such as flexible electronics and high-temperature sensors where the consistency of thin film performance is critical.

[0073] As one implementation method, in this embodiment of the application, the prediction model includes a first prediction sub-model and a second prediction sub-model. Step S13 includes:

[0074] S131. Based on the current laser process parameters and the first prediction sub-model, a prediction is made to obtain the corresponding intermediate physical state prediction value;

[0075] In this embodiment, the first prediction sub-model establishes a quantitative relationship between laser process parameters (linear energy density) and intermediate physical states (maximum melt depth, MPD). Specifically, the first prediction sub-model takes the linear energy density E as input and outputs the predicted value of the maximum melt depth. Its mathematical expression can be shown in Formula 1 below:

[0076] Formula 1: ,

[0077] Where b0, b1, and b2 are model coefficients obtained through numerical simulation and experimental data calibration.

[0078] Therefore, the execution process of step S131 is as follows: calculate the linear energy density E based on the currently determined laser power P and scanning speed v, and then substitute it into the first prediction sub-model shown in Formula 1 to calculate the corresponding intermediate physical state prediction value, i.e., the maximum melting depth. .

[0079] S132. Based on the predicted intermediate physical state, the current laser process parameters, and the second prediction sub-model, a prediction is made to obtain the corresponding microstructure state prediction value.

[0080] In this embodiment, the second prediction sub-model establishes an intermediate physical state (maximum melting depth). The model establishes a quantitative relationship between laser process parameters (linear energy density E) and microstructure (porosity Φ). This model will determine the maximum melting depth predicted by S131. Using the current linear energy density E as input, the output is the predicted value of porosity. Its mathematical expression can be shown in Formula 2:

[0081] Formula 2: ,

[0082] Where c0, c1, c2, c3, and c4 are model coefficients obtained by calibrating experimental data and applying physical monotonicity constraints.

[0083] Physical monotonicity constraints can be specifically defined as Φ / E≤0 and Φ / ≤0, where, Φ represents the minute change in porosity (Φ); E is a small change in the linear energy density (E); Φ / E is the rate of change of porosity with respect to linear energy density, used to characterize how porosity (Φ) changes when other conditions such as laser scanning speed (v) and penetration depth (MPD) remain unchanged, with only a small increase in linear energy density (E);

[0084] The model imposed constraints Φ / E ≤ 0, which means that porosity monotonically does not increase (i.e., decreases or remains unchanged) with increasing energy density. This aligns with basic physical intuition: the more energy input (the greater the power or the slower the speed), the more fully the material melts, the easier it is for the pores to close, and therefore the porosity should decrease. This constraint ensures that the mathematical model behaves in accordance with physical laws, preventing meaningless solutions that contradict common sense, such as "increasing energy leads to increased porosity".

[0085] For the maximum melting depth ( The minute change; Φ / The porosity is the rate of change of porosity with respect to the maximum melting depth, used to characterize how the porosity (Φ) changes when the depth of the molten region increases slightly, while other conditions such as linear energy density (E) remain constant.

[0086] The model also imposed constraints. Φ / MPD ≤ 0, meaning that porosity monotonically increases with increasing penetration depth. The underlying physics is that a greater penetration depth indicates that the laser energy penetrates deeper into the material, resulting in a higher degree of melting and densification from the surface to the interior, thus reducing porosity. This constraint is also to ensure the physical validity of the model.

[0087] Therefore, the execution process of step S132 is as follows: The result obtained in S131... Substituting the current E into the second prediction sub-model shown in Formula 2, the final predicted value of the microstructure state, i.e., porosity, can be calculated. This porosity This is the "virtual sensor" reading used for subsequent comparison and closed-loop control.

[0088] As one implementation method, in this embodiment of the application, the laser process parameters include linear energy density, which is obtained based on laser power and scanning speed.

[0089] In this embodiment, the linear energy density E is a key process parameter characterizing the energy applied by the laser to the scanning trajectory of a thin film per unit length. It is defined as the ratio of laser power P to laser spot scanning speed V, and the specific calculation formula is E=P / v.

[0090] The linear energy density E comprehensively considers both the energy input rate (power) and the energy interaction time (represented by the reciprocal of the velocity, 1 / v), enabling a more direct characterization of the energy conditions for film heating and melting. In the laser sintering process of this application, the linear energy density E is used as the core control variable because, compared to adjusting the power P or the velocity v alone, adjusting E allows for more stable and consistent control of the heat input, thereby achieving predictable control of the film densification process.

[0091] In one embodiment of this application, the intermediate physical state is the maximum melting depth formed within the thin film during laser sintering.

[0092] In this embodiment, the maximum melting depth reflects the penetration and melting degree of laser energy in the thin film, serving as a crucial physical bridge connecting external process parameters and the final microstructure state. By using the maximum melting depth as an intermediate variable, the prediction model possesses a clear physical causal chain: the laser energy input (linear energy density) first determines the melting depth, which in turn directly dominates the pore closure and densification process (porosity). This physical mechanism-based modeling strategy enables the model not only to fit data but also to reliably predict and extrapolate operating conditions for microstructures that cannot be observed online, significantly improving the model's robustness and practicality.

[0093] In one implementation method, the microstructure state in this application embodiment is porosity.

[0094] In this embodiment, porosity is a key microstructural state indicator that characterizes the degree of densification and internal structural integrity of inkjet-printed functional films. It is defined as the percentage of the pore volume inside the film to the total volume of the film.

[0095] In the laser sintering densification process of inkjet-printed thin films, porosity directly determines and comprehensively reflects several key properties of the film:

[0096] 1. Electrical properties: Porosity, acting as defects and scattering centers in the conductive path, significantly increases the resistivity of the thin film and reduces the carrier mobility. Therefore, porosity is a core structural parameter affecting the performance of thin films as conductive functional layers (such as transparent conductive films and sensor electrodes).

[0097] 2. Mechanical and Interfacial Stability: Porosity weakens the mechanical strength of the film and acts as a stress concentration point, easily inducing crack propagation under thermal cycling or mechanical bending. Simultaneously, high porosity reduces the effective contact area between the film and the substrate, lowering interfacial adhesion and leading to delamination or failure during service.

[0098] 3. Optical and Gas Sensing Performance: For transparent conductive films, porosity causes light scattering, reducing transmittance. For gas-sensitive or high-temperature sensors, porosity directly affects the active surface area and gas diffusion channels, thus determining the sensor's response sensitivity, selectivity, and stability.

[0099] Therefore, this application selects porosity as the control target for the microstructure state, aiming to fundamentally optimize and stabilize the electrical, mechanical, interfacial, and functional properties of inkjet-printed functional films through precise and repeatable control. Using porosity, a comprehensive microstructure index, as the controlled variable in closed-loop control is a direct manifestation and effective guarantee for the high-performance, high-reliability film preparation achieved by this method.

[0100] As one implementation method, in this embodiment of the application, step S14 includes:

[0101] S141. Based on the predicted microstructure state and the preset microstructure state, the prediction error is obtained;

[0102] In this embodiment, the prediction error is the core feedback signal in closed-loop feedback control, used to quantify the deviation between the expected result and the ideal target under the current process parameters. Its calculation method is as follows: the predicted value of the microstructure state (e.g., the predicted porosity value) calculated in step S13 based on the prediction model and the current process parameters is used as the basis for the prediction. ) and the pre-set target microstructure state (target porosity) Direct algebraic subtraction is performed, and the calculation formula is: = This error value It contains information on the magnitude and direction of the deviation. Its physical meaning is as follows: when e>0, it indicates that the predicted structure is too dense (porosity is lower than the target), and the energy input needs to be reduced; when e<0, it indicates that the predicted structure is too sparse (porosity is higher than the target), and the energy input needs to be increased. This error value provides a direct and quantitative basis for subsequent parameter adjustments.

[0103] S142. Adjust the laser power and / or scanning speed according to the magnitude and direction of the prediction error to adjust the linear energy density.

[0104] In this embodiment, the core principle of parameter adjustment is: by adjusting the laser power P and / or scanning speed v, the linear energy density E=P / v is changed, thereby correcting the prediction error e and causing the system's prediction output to converge towards the target value. The adjustment process follows the control logic driven by the following physical laws:

[0105] 1. Adjustment direction: If the predicted structure is sparse (e<0), the system increases the linear energy density E by increasing the laser power P and / or decreasing the scanning speed v, thereby promoting further densification of the thin film; conversely, if the predicted structure is dense (e>0), the system decreases E by decreasing the power P and / or increasing the speed v, thereby preventing over-sintering.

[0106] 2. Adjustment range: The adjustment range is proportional to the magnitude of the prediction error e. For example, a proportional control law ΔE = -κ can be used. The adjustment amount ΔE for the linear energy density is calculated using e (where κ>0 is the control gain), and the values ​​of P and v are redistributed according to the new energy density Enew=E+ΔE.

[0107] 3. Implementation Method: This adjustment process is automatically executed by the control system, forming a real-time closed loop targeting the microstructure state. In this way, the system can adaptively compensate for process fluctuations and dynamically optimize process parameters until the prediction error is eliminated or reduced to an acceptable range, thereby ensuring the accuracy and repeatability of the final film performance.

[0108] As one implementation method, in this embodiment of the application, the deviation between the re-obtained predicted microstructure state and the preset microstructure state satisfies a preset condition, including:

[0109] The absolute value of the deviation between the re-obtained microstructure state prediction value and the preset microstructure state is less than the preset threshold.

[0110] In this embodiment, as one implementation method, the deviation between the re-obtained predicted microstructure state and the preset microstructure state satisfies a preset condition, including:

[0111] The absolute value of the deviation between the re-obtained microstructure state prediction value and the preset microstructure state is less than the preset threshold.

[0112] In this embodiment, the "preset threshold" is a pre-set control precision parameter used to determine whether the predicted value is sufficiently close to the target value so that the iteration can be terminated.

[0113] The setting of this threshold usually takes into account the following factors:

[0114] 1. Performance requirements of the final product: For example, for high-performance sensors, the porosity of the thin film needs to be strictly controlled, and the threshold is usually set to be relatively small (e.g., the absolute deviation of porosity is less than 0.5%); for ordinary conductive films with relatively relaxed uniformity requirements, the threshold can be appropriately relaxed.

[0115] 2. Accuracy of model prediction and measurement techniques: The threshold setting should not be less than the systematic error of the model's prediction capability or the final detection method (such as FIB-SEM image analysis) to ensure the rationality and feasibility of the stopping conditions.

[0116] 3. Trade-off between process efficiency and stability: A threshold that is too small may lead to too many iterations, affecting production efficiency; a threshold that is too large may cause the results to deviate from the target, affecting product consistency. Therefore, the threshold needs to seek the optimal balance between efficiency and stability while meeting performance requirements.

[0117] In practical applications, the preset threshold can be a fixed value or a relative value that is dynamically adjusted according to the target value (e.g., set to 1%-2% of the target porosity). When the system detects that the absolute value of the difference between the predicted microstructure state value (such as the predicted porosity value) obtained from the latest iteration calculation and the preset target value is less than this preset threshold, it determines that the deviation has met the preset condition. The currently adjusted laser process parameters are then the optimized parameters to achieve control accuracy. The closed-loop iteration process terminates, and the system can lock the parameter and enter the execution or ready state. This stopping mechanism ensures that the control process can automatically and accurately converge to the target, avoiding endless iterations or over-adjustment.

[0118] As one implementation method, in this embodiment of the application, establishing a prediction model includes:

[0119] S111. Obtain at least one set of intermediate physical state calibration measured values. The calibration measured values ​​are obtained by laser sintering experiments on the thin film using at least one set of laser process parameters on the experimental platform.

[0120] In this embodiment, the specific process of obtaining the measured values ​​of intermediate physical state calibration includes: selecting 3–5 representative combinations of laser process parameters, covering low, medium, and high linear energy density ranges (for example, laser power of 4 W, 8 W, and 12 W, and scanning speed of 400 mm / s, 500 mm / s, and 600 mm / s, respectively). According to these parameter combinations, inkjet-printed thin films are processed on an actual laser sintering experimental platform. In the experiment, synchronous X-ray imaging and infrared thermal imaging technologies are used to monitor and record the evolution of the melt pool inside the thin film in real time during the laser action. Through image post-processing and analysis, the measured values ​​of the maximum melt depth corresponding to each set of process parameters are extracted as the benchmark data for subsequent numerical simulation model calibration.

[0121] S112. Input at least one set of laser process parameters into the numerical simulation model to obtain the corresponding intermediate physical state calibration simulation values.

[0122] In this embodiment, the same set of typical laser process parameters (combination of laser power and scanning speed) used in step S111 are input into the numerical simulation model. The numerical simulation model is a multiphysics coupling model based on the finite element or finite volume method, which can simulate the heat conduction, melting flow and phase change behavior during the interaction between the laser and the thin film. Through simulation calculation, the spatiotemporal distribution of the temperature field in the thin film and the geometry of the melt pool under each set of parameters are obtained, and then the corresponding maximum melting depth simulation value is extracted from it.

[0123] S113. Compare the error between the measured calibration value and the calibration simulation value. If the error exceeds the preset error threshold, adjust the parameters in the numerical simulation model and recalculate the calibration simulation value based on the adjusted numerical simulation model for comparison until the error is less than the preset error threshold, and obtain the calibrated numerical simulation model.

[0124] In this embodiment, the error between the calibrated measured value and the calibrated simulated value is compared. Specifically, the relative error between the simulated value and the measured value for each set of parameters is calculated. The relative error is obtained by calculating the ratio of the absolute value of the difference between the calibrated simulated value and the calibrated measured value to the calibrated measured value. If the relative error of all parameter groups does not exceed the preset error threshold (e.g., 10%), the numerical simulation model is deemed valid. If the relative error of any group exceeds the threshold, the key physical parameters in the numerical simulation model (such as the absorption coefficient of the material to laser, thermal conductivity, etc.) need to be adjusted, and the simulation calculation is rerun until the relative error of all parameter groups meets the threshold requirement. Finally, the calibrated numerical simulation model is obtained, and its prediction results are physically consistent with the experimental data, possessing reliable prediction capabilities.

[0125] S114. Using the calibrated numerical simulation model, simulate and calculate the intermediate physical states under multiple different laser process parameters to obtain multiple intermediate physical state simulation values.

[0126] In this embodiment, a calibrated numerical simulation model is used to systematically simulate and calculate multiple different combinations of laser process parameters. The simulation is based on solving a multiphysics coupling model. Specifically, the laser heating process is described based on the Fourier heat conduction equation, and the possible fluid flow in the molten region is described by combining the Navier-Stokes equation. The evolution of the solid-liquid interface in the molten pool is tracked using methods such as level sets or VOF. By inputting different combinations of laser process parameters (such as power P and scanning speed v, or directly inputting their derived linear energy density E), the spatiotemporal distribution of the temperature field in the thin film and the evolution of the molten pool are simulated and calculated under the corresponding energy input. The key intermediate physical state variable, namely the maximum melt depth, is extracted from the simulation. The simulation values ​​are used to establish a database of the physical relationship between process parameters and melt pool depth in an efficient and low-cost manner, providing sufficient and physically consistent data support for subsequent model training.

[0127] S115. Fit the simulation values ​​of multiple laser process parameters and multiple intermediate physical states to obtain the coefficients of the first prediction sub-model;

[0128] In this embodiment, the fitting process aims to establish the relationship between laser process parameters (linear energy density E) and simulated values ​​of intermediate physical states (maximum melting depth). The quantitative mathematical relationship between them, that is, determining the first predictive sub-model. The specific coefficients are determined. Multiple sets of data obtained from S111 will be used as the training set, and fitting methods such as multinomial regression will be employed for model training. For example, the coefficients b0, b1, and b2 that minimize the prediction error will be solved using optimization algorithms such as weighted least squares. The specific method is as follows:

[0129] Suppose there are N training samples in total, and the linear energy density of the i-th sample is: The corresponding maximum melt depth simulation value is The first predictive sub-model is in quadratic polynomial form:

[0130] ;

[0131] 1. Rewrite in standard regression form:

[0132] For all N samples, the model can be written in vector form:

[0133] ;

[0134] 2. Design Matrix:

[0135] Define the design matrix X as an n×3 matrix:

[0136] ;

[0137] 3. Parameter vector:

[0138] Define parameter vector ;

[0139] 4. Solving for coefficients using the weighted least squares method:

[0140] If each sample i is assigned a weight Let the weight matrix The objective of weighted least squares estimation is to minimize the sum of squared weighted errors, which is equivalent to solving:

[0141] ,

[0142] Therefore, the obtained b0, b1, and b2 are the optimal coefficient estimates that minimize the model's prediction error.

[0143] The first prediction sub-model obtained after fitting can quickly predict the corresponding maximum melting depth based on any given linear energy density E. .

[0144] S116. Obtain the measured values ​​of the microstructure state calibration of the thin film after laser treatment under multiple different laser process parameters;

[0145] In this embodiment, the calibration and measurement values ​​are obtained through a combination of offline experiments and microscopic analysis. Specifically, the inkjet-printed thin film samples are laser-sintered on actual equipment according to multiple sets of laser process parameters (the same P and v combination) corresponding to the numerical simulation parameters. Subsequently, the cross-section or surface morphology of the processed samples is observed using precision microscopic analysis equipment such as focused ion beam scanning electron microscopy (FIB-SEM). The obtained microscopic images are quantitatively analyzed using image processing and analysis algorithms (such as threshold segmentation and pore identification algorithms) to calculate the actual measured value of the microstructural state (porosity) for each sample. These measured values ​​correspond one-to-one with the simulation parameters, constituting the dataset necessary for training the second predictive sub-model.

[0146] S117. The coefficients of the second prediction sub-model are obtained by fitting multiple laser process parameters, multiple intermediate physical state simulation values ​​and multiple microstructure state calibration measured values.

[0147] In this embodiment, the fitting process aims to establish an intermediate physical state (maximum melting depth). The quantitative relationship between laser process parameters (linear energy density E) and the final microstructure state (porosity Φ) is used to determine the second prediction sub-model. The specific coefficients, the training data consists of three parts: laser process parameters (Ei), intermediate state prediction values ​​calculated by the first prediction sub-model based on Ei, and... ), and the corresponding measured porosity calibration values ​​obtained in S113 ( The model is trained using fitting methods such as multinomial regression. For example, optimization algorithms such as weighted least squares are used to solve for the coefficients c0, c1, c2, c3, and c4 that minimize the prediction error. During the fitting process, physical monotonicity constraints can be applied to ensure that the model's prediction trend conforms to the basic physical law that "increased energy or melting depth leads to decreased porosity." The specific methods are as follows:

[0148] Suppose there are N training samples in total, and the linear energy density of the i-th sample is: Maximum melting depth is ( (This can be the output value of the first predictive sub-model or the output value under simulation), and the corresponding measured porosity calibration value is... The second predictive sub-model is in quadratic polynomial form:

[0149] ;

[0150] 1. Rewrite in standard regression form:

[0151] For all N samples, the model can be written in vector form:

[0152] ;

[0153] 2. Design Matrix:

[0154] Define the design matrix X as an n×5 matrix:

[0155] ;

[0156] 3. Parameter vector:

[0157] Define parameter vector ;

[0158] 4. Solving for coefficients using the weighted least squares method:

[0159] If each sample i is assigned a weight Let the weight matrix The objective of weighted least squares estimation is to minimize the sum of squared weighted errors, which is equivalent to solving:

[0160] ;

[0161] 5. Physical constraints

[0162] Apply physical monotonicity constraints:

[0163] ;

[0164] ;

[0165] In implementation, "discrete point constraints" are typically used: for each calibration sample Apply the inequality (or further encrypt some points) on:

[0166] ;

[0167] ;

[0168] The solution obtained from this That is, the optimal coefficient estimate that minimizes the model's prediction error.

[0169] After fitting is complete, a result that can simultaneously input E and The second prediction sub-model is used to predict the porosity Φ, thus forming a complete prediction chain of "process parameters - intermediate state - microstructure state" together with the first prediction sub-model.

[0170] As one implementation method, in this embodiment of the application, the method further includes:

[0171] S21. Obtain offline measured values ​​of the microstructure state of the thin film after laser treatment;

[0172] In this embodiment, obtaining offline measured values ​​of the microstructure state is a separate step used for model post-hoc calibration and optimization. After laser sintering a batch of thin films using the current prediction model and closed-loop control system, representative samples are extracted from the batch. High-precision offline microscopic analysis techniques (e.g., focused ion beam scanning electron microscopy, i.e., FIB-SEM) are used to observe the cross-section or surface of the thin film samples. The acquired microscopic images are quantitatively analyzed using image processing and analysis algorithms (such as pore segmentation based on grayscale thresholds, morphological analysis, etc.) to accurately calculate the actual porosity (or other microstructural parameters) of the thin film. This data serves as the offline measured value of the microstructure state formed by the sample under specific laser process parameters. These measured data are the "gold standard" for evaluating model prediction accuracy and performing iterative model optimization. The specific method for quantitatively analyzing the acquired microscopic images using image processing and analysis algorithms to accurately calculate the actual porosity of the thin film is as follows:

[0173] 1. Convert the input image to grayscale to simplify subsequent processing steps and highlight structural contrast;

[0174] 2. Define image units (such as pixels or field of view areas) and delineate the recognition area of ​​the thin film sample, excluding invalid areas such as the substrate or edges;

[0175] 3. Enhance the contrast of the recognition area to obtain an enhanced image, thereby strengthening the grayscale difference between the pores and the substrate;

[0176] 4. To suppress image noise and smooth edges, a Gaussian blur algorithm is applied to filter the enhanced image, resulting in a filtered image.

[0177] 5. Use an image segmentation algorithm (such as an adaptive thresholding algorithm) to perform image binarization segmentation to obtain the segmented image;

[0178] 6. Perform morphological operations (such as opening and closing operations) on the segmented image to remove small noise, fill in the internal breaks of pores, or separate the adhesion pores to obtain the image after morphological operation;

[0179] 7. Divide the number of pixels in the pore region of the binary image by the total number of pixels in the recognition region to calculate the porosity. Output the calculated porosity data as the offline measured value of the microstructure state of the sample. The final processed image can be visualized for result verification and recording.

[0180] S22. Based on the laser process parameters corresponding to the offline measured values ​​of the microstructure state, obtain the corresponding intermediate physical state reference values ​​using numerical simulation or the first prediction sub-model.

[0181] In this embodiment, the purpose of obtaining intermediate physical state reference values ​​is to establish the relationship between the offline measured microstructure and the physical state under corresponding process conditions. There are two specific methods:

[0182] 1. Numerical simulation reproduction path: Based on the specific laser process parameters (laser power P and scanning speed v) corresponding to the sample described in S21, rerun the same thermo-fluid multiphysics coupling numerical simulation as in step S111, such as... Figure 2 As shown, this simulation can obtain the temperature field and molten pool evolution results under different process parameters, such as... Figure 3 As shown, the maximum penetration depth (MPD) of the key physical quantity can be extracted from the simulation results. From this, the simulated maximum penetration depth value under the precise process parameters can be directly calculated as an intermediate physical state reference value. This method is the most direct, but the computational cost is relatively high.

[0183] 2. First Prediction Sub-model Prediction Path: Input the laser process parameters (or the calculated linear energy density E) corresponding to the sample in S21 into the currently calibrated first prediction sub-model. The model calculates the corresponding maximum melt depth prediction value, which serves as an intermediate physical state reference value. This method is highly efficient and suitable for rapid online model updates. Its accuracy depends on the accuracy of the first prediction sub-model.

[0184] By using any of the above methods, an intermediate physical state reference value (maximum melting depth reference value) that strictly corresponds to the offline measured porosity can be obtained, providing a complete data tuple for the next step of updating the second prediction sub-model.

[0185] S23. Based on the offline measured values ​​of the microstructure state, the corresponding laser process parameters, and the corresponding intermediate physical state reference values, the coefficients of the second prediction sub-model are updated to obtain the optimized second prediction sub-model.

[0186] In this embodiment, the model update process aims to use newly acquired offline measured data to correct and optimize the second prediction sub-model, making its predictions closer to the actual process results. The specific steps are as follows:

[0187] 1. Data integration: The newly obtained data tuples, namely laser process parameters (linear energy density), intermediate physical state reference values ​​(maximum melting depth reference values) obtained from S22, and offline measured values ​​of microstructure state (measured porosity values) obtained from S21, are added to the historical dataset used to train the second prediction sub-model.

[0188] 2. Model Refitting / Incremental Update: Using the expanded complete dataset as the new training set, the coefficients of the second predictive sub-model are refitted. To improve efficiency, online update algorithms such as incremental learning or recursive least squares can be used to fine-tune the original model coefficients based only on the new data. During the fitting process, physical monotonicity constraints must also be applied.

[0189] 3. Model Deployment: The updated coefficients are deployed to the prediction module of the control system, replacing the old coefficients. This results in an optimized second prediction sub-model with improved prediction accuracy and adaptability to fluctuations in external conditions, thereby continuously optimizing the closed-loop control performance of the subsequent laser sintering process. This mechanism enables the prediction model to learn and continuously optimize itself, which is crucial for ensuring long-term process stability and product consistency.

[0190] like Figure 4 As shown, this application embodiment also provides a densification control system for inkjet-printed films, including:

[0191] Module 21 is established to build a prediction model, which is used to characterize the relationship between laser process parameters, intermediate physical states and the microstructure state of the thin film.

[0192] The determination module 22 is used to determine the current laser process parameters based on the preset microstructure state;

[0193] Prediction module 23 is used to make predictions based on the current laser process parameters and prediction model to obtain the corresponding microstructure state prediction values;

[0194] The first adjustment module 24 is used to compare the predicted microstructure state with the preset microstructure state, and adjust the laser process parameters based on the comparison result.

[0195] The second adjustment module 25 is used to re-predict and compare based on the adjusted laser process parameters until the deviation between the newly obtained microstructure state prediction value and the preset microstructure state meets the preset conditions.

[0196] The embodiments in this specification are described in a progressive manner, with each embodiment focusing on the related aspects.

[0197] For any differences between the embodiments, or for the same or similar parts between the embodiments, please refer to each other.

[0198] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for controlling the densification of inkjet-printed films, characterized in that, include: A prediction model is established to characterize the relationship between laser process parameters, intermediate physical state and microstructure state of thin film. The laser process parameters include linear energy density, which is obtained based on laser power and scanning speed. The intermediate physical state is the maximum melting depth formed in the thin film during laser sintering. The microstructure state is porosity. Determine the current laser process parameters based on the preset microstructure state; Based on the current laser process parameters and the prediction model, the corresponding microstructure state prediction value is obtained. The predicted microstructure state is compared with the preset microstructure state, and the laser process parameters are adjusted based on the comparison results. Based on the adjusted laser process parameters, prediction and comparison are performed again until the deviation between the newly obtained microstructure state prediction value and the preset microstructure state meets the preset condition. The prediction model includes a first prediction sub-model and a second prediction sub-model. The step of predicting the corresponding microstructure state based on the current laser process parameters and the prediction model includes: Based on the current laser process parameters and the first prediction sub-model, a prediction of the corresponding intermediate physical state is obtained. Based on the predicted intermediate physical state, the current laser process parameters, and the second prediction sub-model, the corresponding microstructure state prediction value is obtained.

2. The method according to claim 1, characterized in that, The step of comparing the predicted microstructure state with the preset microstructure state and adjusting the laser process parameters based on the comparison result includes: The prediction error is obtained by comparing the predicted microstructure state with the preset microstructure state. The laser power and / or the scanning speed are adjusted according to the magnitude and direction of the prediction error to adjust the linear energy density.

3. The method according to claim 1, characterized in that, The deviation between the re-obtained predicted microstructure state and the preset microstructure state satisfies a preset condition, including: The absolute value of the deviation between the re-obtained microstructure state prediction value and the preset microstructure state is less than the preset threshold.

4. The method according to claim 1, characterized in that, The establishment of the prediction model includes: At least one set of intermediate physical state calibration measured values ​​are obtained, which are obtained by laser sintering experiments on thin films on an experimental platform using at least one set of laser process parameters. Input the at least one set of laser process parameters into the numerical simulation model to obtain the corresponding intermediate physical state calibration simulation values; The error between the measured calibration value and the calibration simulation value is compared. If the error exceeds the preset error threshold, the parameters in the numerical simulation model are adjusted, and the calibration simulation value is recalculated based on the adjusted numerical simulation model for comparison, until the error is less than the preset error threshold, and the calibrated numerical simulation model is obtained. Using the calibrated numerical simulation model, the intermediate physical states under multiple different laser process parameters are simulated and calculated to obtain multiple intermediate physical state simulation values. The coefficients of the first prediction sub-model are obtained by fitting multiple laser process parameters and multiple intermediate physical state simulation values. Obtain measured values ​​of the microstructure state of thin films after laser treatment under multiple different laser process parameters; The coefficients of the second prediction sub-model are obtained by fitting multiple laser process parameters, multiple intermediate physical state simulation values, and multiple microstructure state calibration measured values.

5. The method according to claim 4, characterized in that, The method further includes: Offline measured values ​​of the microstructure state of the laser-treated thin film were obtained. Based on the laser process parameters corresponding to the offline measured values ​​of the microstructure state, the corresponding intermediate physical state reference values ​​are obtained using numerical simulation or the first prediction sub-model. Based on the offline measured values ​​of the microstructure state, the corresponding laser process parameters, and the corresponding intermediate physical state reference values, the coefficients of the second prediction sub-model are updated to obtain the optimized second prediction sub-model.

6. A densification control system for inkjet-printed films, characterized in that, include: A module is established to build a prediction model, which is used to characterize the relationship between laser process parameters, intermediate physical state and microstructure state of thin film. The laser process parameters include linear energy density, which is obtained based on laser power and scanning speed. The intermediate physical state is the maximum melting depth formed in the thin film during laser sintering. The microstructure state is porosity. The determination module is used to determine the current laser process parameters based on the preset microstructure state; The prediction module is used to make predictions based on the current laser process parameters and the prediction model to obtain the corresponding microstructure state prediction values. The first adjustment module is used to compare the predicted microstructure state with the preset microstructure state, and adjust the laser process parameters based on the comparison result. The second adjustment module is used to re-predict and compare based on the adjusted laser process parameters until the deviation between the newly obtained microstructure state prediction value and the preset microstructure state meets the preset conditions. The prediction model includes a first prediction sub-model and a second prediction sub-model. When the prediction module performs the prediction based on the current laser process parameters and the prediction model to obtain the corresponding microstructure state prediction value, it is specifically used for: Based on the current laser process parameters and the first prediction sub-model, a prediction of the corresponding intermediate physical state is obtained. Based on the predicted intermediate physical state, the current laser process parameters, and the second prediction sub-model, the corresponding microstructure state prediction value is obtained.

Images