Evaporation source control method and system for a vacuum evaporation machine

Abstract

The application discloses a kind of evaporation source control method and system for vacuum evaporation machine, the method includes pre-melting, power pre-adjustment, feedback delay and closed-loop regulation step, after pre-melting, before shutter opens, heating power is adjusted from pre-melting heating power to the estimated steady state power corresponding to target film deposition rate;Shutter is opened, and the estimated steady state power is maintained, only to the measured film deposition rate is monitored without closed-loop regulation, and adaptive soft delay criterion determined based on fixed hard delay time and rate change rate information generates closed-loop cut-in instruction;In response to the instruction, first, adjust heating power with pure proportional control, when measured film deposition rate enters tolerance band or the running length of pure proportional control reaches the preset maximum running length, activate integral item to PI or PID control, and according to the principle of continuous switching instantaneous power given value, initial value is assigned to integral item.By the application, closed-loop cut-in without overshoot, saturation, jump is realized, and the starting film layer thickness uniformity is improved.

Description

Technical Field

[0001] This invention belongs to the field of automatic control technology for vacuum coating equipment, and specifically relates to an evaporation source control method and system for a vacuum evaporation machine. Background Technology

[0002] Vacuum evaporation is a core process that uses heating or electron beam bombardment to vaporize and deposit coating materials onto a substrate surface to form a thin film under high vacuum conditions. It is widely used in high-end manufacturing fields such as display panels, optical lenses, semiconductor devices, solar cells, and flexible electronics. The stability of the coating rate, especially in the initial stage after the mask is opened, directly determines the uniformity of the initial film thickness and batch consistency. Currently, vacuum evaporation equipment generally uses a PID controller to perform closed-loop regulation of the evaporation source heating power, and sets a feedback delay period after the mask is opened, after which closed-loop control is initiated.

[0003] However, existing technologies have the following technical drawbacks in the initial transient evaporation phase after the baffle is opened: The use of excessively high pre-melting heating power during the feedback delay period, which is significantly higher than the steady-state power corresponding to the target coating rate, leads to severe overshoot of the coating rate during the delay period. The delay period typically uses a fixed delay, but the transient duration varies greatly depending on the coating material and operating conditions: if the delay is too short, the cut-in occurs while the rate is still fluctuating wildly, and direct closed-loop intervention introduces control overshoot; if the delay is too long, coating efficiency is lost, and the thermal state of the molten pool drifts during the extended wait. When switching to PI / PID control with an integral term at the end of the delay period, the integral term accumulates from the starting point of a large deviation, resulting in integral saturation and triggering a secondary overshoot after the cut-in, requiring a long time to converge to steady state. Clearly, the power starting point mismatch increases the deviation at the cut-in time, which in turn exacerbates integral saturation, collectively leading to large initial film thickness deviations and poor uniformity. Existing technologies can only provide partial remedies for single phenomena, such as extending the fixed delay, integral limiting, or integral preloading, but none can simultaneously address all three issues.

[0004] Furthermore, some existing solutions attempt to adjust the heating power to the level corresponding to the target evaporation rate before coating begins, based on the pre-collected correspondence between heating parameters and evaporation rates. They then wait for the evaporation rate fluctuations to fall within the target range before switching to conventional control. While this approach shortens the rate stabilization time to some extent, its rate closed-loop operation is already in place before the baffle is opened, making it unable to handle the transient disturbances introduced by the rebalancing of the molten pool surface and the initial establishment of the vapor flow after the baffle is opened. Moreover, its waiting criterion is based solely on whether the evaporation rate falls within the fluctuation range, without considering the shortest transient time that must be experienced after the baffle is opened, or the initial state of the integral term at the moment of entry. Therefore, it still generates a secondary overshoot after entry.

[0005] Based on this, this application provides a technical solution for controlling the evaporation source of a vacuum evaporation machine, in order to solve the technical problems in the prior art where the power starting point, timing of entry, and integral state of the closed-loop cut-in are mismatched at the same time, resulting in rate overshoot and integral saturation overshoot, and poor uniformity of the initial film thickness. Summary of the Invention

[0006] To address the technical problems in existing technologies where the power start-up, timing of the cut-in, and integral state of the closed-loop cut-in are mismatched during the initial transient evaporation phase after the shield is opened, leading to rate overshoot, integral saturation overshoot, and poor uniformity of the initial film thickness, this application provides the following technical solution: A method for controlling the evaporation source in a vacuum evaporation deposition machine includes the following steps: Pre-melting step: Close the shielding plate and perform pre-melting stabilization treatment on the coating material to obtain the target coating rate and the pre-melting heating power at the end of the pre-melting stage; Power pre-adjustment step: During the time period before the shield is opened, obtain the estimated steady-state power corresponding to the target coating rate, and adjust the heating power from the pre-melting heating power to the estimated steady-state power. Feedback delay step: After opening the shield, maintain the heating power in open loop at the estimated steady-state power, obtain the measured coating rate information and calculate the rate of change information, wherein the measured coating rate information is not used for closed-loop adjustment of the heating power; use a fixed hard delay time as a pre-gating condition, and after the fixed hard delay time is reached, use the rate of change information being continuously lower than a preset threshold as an adaptive soft delay criterion that the violent transient has ended; when the fixed hard delay time is reached and the adaptive soft delay criterion is satisfied, generate a closed-loop cut-in command. Closed-loop adjustment steps: In response to the closed-loop cut-in command, a pure proportional control command is first generated to adjust the heating power. When the measured coating rate enters the tolerance band of the target coating rate, or the running time of the pure proportional control reaches the preset maximum running time, a PI or PID control command containing an integral term is generated to adjust the heating power in a closed loop. When the integral term is activated, the initial value of the integral term is assigned to a value that keeps the power setpoint continuous at the moment of switching.

[0007] Furthermore, the estimated steady-state power is obtained by establishing a mapping relationship between the target coating rate and steady-state power through regression analysis of the steady-state power records of historical coating processes, and then querying or interpolating based on the current target coating rate; or by matching and searching the historical record library according to the working condition label, and directly calling the steady-state power record of the previous coating under the same working condition.

[0008] Furthermore, the heating power is adjusted from the pre-melting heating power to the estimated steady-state power through step-by-step or ramp-like adjustments.

[0009] Furthermore, the adaptive soft delay criterion and the closed-loop cut-in command are determined in the following manner: The fixed hard delay time is a preset fixed value, which is counted from the moment the cover is opened. Before the fixed hard delay time is reached, even if the adaptive soft delay criterion is met, the closed-loop cut-in command will not be generated. After the fixed hard delay time is reached, when the rate of change of the rate of change of N consecutive sampling periods is lower than the preset rate of change threshold, it is determined that the adaptive soft delay criterion is met, and the closed-loop cut-in command is generated.

[0010] Furthermore, the determination of whether the measured coating rate enters the tolerance band of the target coating rate includes a jitter criterion: the measured coating rate is determined to be within the tolerance band if it is within the range for M consecutive sampling periods, and the counting is restarted if any sampling point leaves the tolerance band during the period; the tolerance band is the range of ±5% to ±15% of the target coating rate, and the half-width of the tolerance band is not less than 3 times the standard deviation of the film thickness gauge measurement noise and less than the bandwidth corresponding to the upper limit of the allowable fluctuation of the coating rate.

[0011] Furthermore, the preset maximum running time is set according to the coating material type and the rate range to which the target coating rate belongs; when the running time of pure proportional control reaches the preset maximum running time, regardless of whether the measured coating rate enters the tolerance band, the integral term is forcibly activated and switched to PI or PID control.

[0012] Furthermore, when generating PI or PID control commands containing integral terms, the control model is selected according to the coating material type and the rate range to which the target coating rate belongs: when the target coating rate is greater than or equal to 5.0 Å / s, a PI control command is generated; when the target coating rate is less than or equal to 2.0 Å / s, a PID control command is generated and differential vibration suppression is enabled.

[0013] This application also provides an evaporation source control system for a vacuum evaporation deposition machine, specifically including: A coating rate setting module is used to obtain the target coating rate and the tolerance band of the target coating rate; The real-time monitoring module is used to collect measured coating rate information in real time; The deviation calculation and control module is used to obtain the target coating rate and the measured coating rate, calculate the deviation, and generate power adjustment commands. This includes generating a pure proportional control command before closed-loop switching, generating a PI or PID control command containing an integral term after activating the integral term, and assigning the initial value of the integral term to a value that keeps the power setpoint continuous at the moment of switching when activating the integral term. The control model of the PI or PID control command is selected according to the coating material type and the rate range to which the target coating rate belongs. A power output module is used to acquire the power adjustment command and generate a power output signal adapted to the evaporation source; The feedback delay module is used to adjust the heating power from the pre-melting heating power at the end of the pre-melting stage to the estimated steady-state power corresponding to the target coating rate before the shield is opened. After the shield is opened, it acquires the measured coating rate information, generates monitoring data and rate change information, and maintains the estimated steady-state power. Based on the adaptive soft delay criterion determined by the fixed hard delay time and the rate change information, it generates a closed-loop cut-in command and controls the operation to first run in pure proportional control. After the measured coating rate enters the tolerance band or the pure proportional control operation reaches the preset maximum running time, it activates the integral term to switch to PI or PID control.

[0014] Compared with the prior art, the present invention has the following beneficial effects: By pre-adjusting the power to the estimated steady-state power before the shield opens, the starting point of the power during the delay period is close to the steady-state value corresponding to the target rate. This limits the rate overshoot during the delay period to the pre-adjustment accuracy range from the source, solving the rate overshoot caused by power starting point mismatch and achieving "prevention" rather than "recovery". The fixed hard delay ensures that the most severe transient will pass, and the adaptive soft delay judges whether the most severe transient introduced by the shield opening has ended based on the actual rate of change. The timing of the switch is automatically matched with the duration of transients of different coating materials and different working conditions: if the rate of change slows down early, the switch is made earlier to improve efficiency, and if the rate of change slows down late, the switch is made later to ensure safety, solving the timing mismatch problem. The two-stage switching of proportional and integral terms ensures that the integral term accumulates from a small deviation after the deviation has converged, thus eliminating integral saturation at the moment of switch in principle. Combined with the non-intrusive assignment of the integral term, the power setpoint is continuous and there is no power jump at the moment of switching, solving the integral state mismatch problem. The above three elements constitute a complete synergistic chain: power pre-adjustment ensures bounded deviation at the cut-in time, enabling the pure proportional segment to converge quickly into the tolerance band; the integral term starting with small deviations will not saturate; and perturbation-free assignment ensures smooth switching. Together, they achieve closed-loop cut-in without overshoot, saturation, or jumps. Comparative experiments show that, compared to the scheme of fixed delay plus direct-cut-in PI control, this invention reduces the rate overshoot during the delay period from 44% to 16%, eliminates secondary overshoot, and shortens the settling time from approximately 8 seconds to approximately 3.5 seconds.

[0015] After integral activation, the control model for the closed-loop segment is automatically selected according to the coating material type and the rate range of the target coating rate: high-rate metal films (≥5.0 Å / s) use PI control, utilizing the rapid, oscillating-free convergence characteristics of proportional and integral coefficients to quickly stabilize within the tolerance band; low-rate precision films (≤2.0 Å / s) use PID control, with the derivative term suppressing rate oscillations within the small tolerance band. This differentiated selection, along with the differentiated tuning of the tolerance band width, maximum running time, and proportional coefficient according to the coating material and rate range, belongs to the same system, ensuring that the closed-loop convergence characteristics after integral activation match the coating material characteristics and tolerance band width, and working in conjunction with the two-stage switching to balance response speed and steady-state accuracy.

[0016] Comparative experiments show that the synergistic effect of the three methods mentioned above is not a simple summation of their respective effects: Under the same operating conditions, the scheme using a fixed delay with the pre-melting heating power used during the delay period, and directly switching to PI control after the delay, resulted in a rate overshoot of 44% during the delay period, followed by a second overshoot after switching, and only entering the steady-state range of 5.0±0.1 Å / s after approximately 8 seconds; while the scheme of the present invention resulted in a rate overshoot of only 16% during the delay period, with no second overshoot throughout the process, and entering the same steady-state range in approximately 3.5 seconds. It is evident that the present invention simultaneously eliminates rate overshoot and integral saturation second overshoot, and reduces the settling time by more than half. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a block diagram of the overall system structure of the present invention; Figure 2 This is an overall flowchart of the coating rate control of the present invention; Figure 3 This is a schematic diagram of the three-stage timing of the initial transient suppression of the present invention; Figure 4 This is a schematic diagram of the dual-segment criterion for hard delay and soft delay in this invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. 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.

[0019] like Figure 1 As shown, the evaporation source control system of the present invention for a vacuum evaporation coating machine consists of a coating rate setting module, a real-time monitoring module, a deviation calculation and control module, a power output module, and a feedback delay module working together.

[0020] The coating rate setting module allows setting the target coating rate, tolerance band, and allowable upper limit of coating rate fluctuation according to the coating material type and film function. The allowable upper limit of coating rate fluctuation is the maximum deviation of the measured coating rate relative to the target coating rate allowed by the process (e.g., ±20%). Exceeding this upper limit is considered an abnormal rate. Control parameters support storage, retrieval, and batch export, such as proportional coefficient, integral time, fixed hard delay time, N value, rate of change threshold, tolerance band, maximum running time, etc. It has a built-in process parameter library and supports one-click retrieval of parameters for metal films, optical films, and semiconductor films.

[0021] Real-time monitoring module: Employs a quartz crystal film thickness gauge to acquire instantaneous coating rate, cumulative film thickness, and frequency signals in real time. Built-in digital filtering suppresses interference from vacuum, electron gun, and vibration. It communicates with the evaporation source power supply (such as an electron gun power supply or thermal resistance power supply) to read the actual output power in real time, used to verify the deviation between the pre-adjusted actual output power and the estimated steady-state power. During the feedback delay step, the real-time monitoring module continuously acquires measured coating rate information, but the acquired rate information is only used to generate coating rate monitoring data and calculate the rate of change, not for closed-loop power adjustment—that is, the real-time monitoring module "only observes and does not act" during this stage: data enters the display path and criterion calculation path, but not the power feedback path, and the heating power remains at the estimated steady-state power.

[0022] Deviation Calculation and Control Module: Real-time calculation of deviation = target coating rate The measured coating rate was determined. No power adjustment command was output during the feedback delay step. In the first stage of the two-stage input (pure proportional control), the power adjustment was calculated only based on the current deviation using the proportional coefficient, without enabling the integral term. In the second stage (PI or PID control), all control terms of the selected control model were fully enabled. The control model supports automatic selection based on the coating material type and the target coating rate range: high coating rate metal films (≥5.0 Å / s) use PI control for fast response and no oscillation; low coating rate precision films (≤2.0 Å / s) use PID control for high accuracy and differential vibration suppression; PI control is used by default for rates between 2 and 5 Å / s. This selection, along with the differentiated tuning of tolerance bandwidth, maximum runtime T_pmax, and proportional coefficient Kp based on coating material and rate range, belongs to the same system, ensuring that the closed-loop convergence characteristics after integral activation match the coating material characteristics and tolerance bandwidth.

[0023] The above four parameters are all linked together using the same set of index keys, namely "coating material type + rate range to which the target coating rate belongs", rather than being set independently. For the same coating material in the same rate range, the tolerance band half-width δ, the maximum running time T_pmax, the proportional coefficient Kp, and the control model are all matched to a set of values, so that the integral activation threshold determined by the tolerance band width, the convergence speed of the pure proportional segment, the magnitude of the deviation at the integral activation time, and the convergence characteristics of the closed loop after activation are coordinated with each other.

[0024] Power output module: Generates a power output signal adapted to the evaporation source, compatible with electron gun and thermal resistance evaporation, with a response time ≤10ms.

[0025] Feedback Delay Module: The core module of this invention, executing the initial transient suppression process of the power pre-adjustment step, feedback delay step, and closed-loop adjustment step, including power pre-adjustment before the shield opens, a two-stage delay criterion after the shield opens, and a two-stage switching from proportional to integral during closed-loop entry. Specifically, from the time the shield opens until the closed-loop entry command is generated, the feedback delay module maintains the heating power at the estimated steady-state power in the open loop and limits the measured coating rate information to the monitoring and criterion calculation paths, without sending it to the power feedback path. That is, the rate closed loop is only put into operation by the deviation calculation and control module after the closed-loop entry command is generated.

[0026] like Figure 2 As shown, the overall control flow of this invention includes five stages: pre-melting and stabilizing the pool, power pre-adjustment, feedback delay monitoring, dual-segment criterion entry, and two-segment closed-loop switching. After activating the integral term, PI or PID control is used for continuous closed-loop adjustment until the coating is completed. Specifically: Phase 1, Pre-melting and Stabilizing the Molten Pool, corresponds to the following pre-melting steps: the shielding plate is closed, and the film material is fully heated and melted to form a uniform and stable molten pool, ensuring a stable coating rate during the formal coating process. When the molten pool reaches a stable state, a pre-melting completion signal is generated.

[0027] Phase 2 power pre-adjustment corresponds to the following power pre-adjustment steps: Before the shield is opened: During the time period between the end of the pre-melting stabilization phase and before the shield is opened, the feedback delay module obtains the pre-melting heating power at the end of the pre-melting phase and the estimated steady-state power corresponding to the target coating rate. The heating power is then adjusted from the pre-melting heating power to the estimated steady-state power in a step-like or ramp-like manner. The step-like or ramp adjustment itself is a controlled gradual process, with the single step size constrained by the adjustment method.

[0028] Among them, the first method for obtaining the estimated steady-state power (regression analysis method) establishes and uses the mapping relationship between the target coating rate and the steady-state power according to the following steps: Data Acquisition: During each coating process, once the closed-loop adjustment reaches a steady state (e.g., the deviation between the measured coating rate and the target coating rate is within ±2% for 10 consecutive sampling cycles), the "target coating rate - steady-state heating power" data pair is recorded at a fixed sampling period (e.g., 0.5s). The arithmetic mean of the heating power during this period is taken as the steady-state power record for this coating, and stored in the historical record database along with the coating material type, evaporation source type and crucible / evaporation boat model, distance from evaporation source to substrate, and position of the film thickness gauge crystal control plate, etc. Data cleaning: Remove coating records that have triggered alarms or abnormal fluctuations to ensure sample validity; Model establishment: For at least K groups (e.g., K≥10) of valid records under the same coating material type, with the target coating rate as the independent variable and steady-state power as the dependent variable, the least squares method is used to fit a univariate linear regression model P=aR+b (or a second-order polynomial model P=aR²+bR+c). When the goodness of fit R²≥0.95, the model is deemed usable. Online usage: Before the current coating begins, the target coating rate is substituted into the regression model to obtain the estimated steady-state power; if the target rate is between two historical records, linear interpolation can also be used directly.

[0029] Method 2 (Direct Retrieval Method): Search the historical records database by matching the operating condition tags. A condition is considered "same operating condition" only if all of the following criteria are identical or fall within their respective allowable deviations: Same type and purity level of plating material; same type of evaporation source (electron gun / thermal resistance evaporation) and material and capacity of crucible or evaporation boat; same distance from evaporation source to substrate (allowable deviation ±2mm); same installation position and model of the film thickness gauge crystal control plate; target coating rate deviation within ±10%; and the chamber's base vacuum level is at the same order of magnitude (e.g., not greater than 5×10⁻⁶). -4 Pa).

[0030] When a record of the same operating condition is matched, the steady-state power of that record is directly used as the estimated steady-state power; if no match is found, it reverts to method one; if method one is also unavailable (e.g., when using a new plating material for the first time, or when there is no valid historical record), then 70%~80% of the pre-melting heating power is used as a conservative initial value for the estimated steady-state power, which is then converged to steady state by subsequent closed-loop adjustment. The above three acquisition paths mutually support each other and degrade sequentially, and none of them require the pre-establishment of a complete power-rate database covering all operating conditions, resulting in low implementation cost: priority is given to directly calling the power based on the operating condition label; if no match is found, it reverts to regression analysis; if the regression model is unavailable or the goodness of fit is not up to standard, it reverts to the conservative initial value.

[0031] Through this pre-adjustment step, the heating power is at the starting point close to the steady-state value required for the target rate when the shield is opened, reducing rate overshoot during the delay period from the source. At the same time, since the heating power when the shield is opened has been pre-adjusted to the estimated steady-state power, the deviation between the measured coating rate and the target coating rate at the subsequent closed-loop cut-in time is constrained within the range determined by the pre-adjustment accuracy of the estimated steady-state power, that is, a bounded small deviation, rather than the large deviation when using the pre-melting heating power.

[0032] Phase 3 feedback delay monitoring, corresponding to the feedback delay step, after the shield is opened: the real-time monitoring module continuously collects the measured coating rate information. The feedback delay module processes the measured coating rate information and generates two types of outputs: one is coating rate monitoring data, which allows operators to monitor the rate status during the delay period and display it on the interface; the other is the rate of change information, which serves as the input basis for the adaptive soft delay criterion. In this phase, the measured coating rate information is not used for closed-loop power adjustment, and the heating power is maintained near the estimated steady-state power after the pre-adjustment in Phase 2. Because the starting power is close to the correct value, the actual rate will not overshoot significantly during the delay period. In other words, the heating power is maintained in an open-loop manner during the feedback delay phase: the heating power is neither calculated from the rate deviation nor changes with the measured coating rate, but is maintained constant at the estimated steady-state power determined in Phase 2 until the closed-loop cut-in command is generated; the rate closed loop is only put into operation after the closed-loop cut-in command is generated.

[0033] Phase 4 Dual-segment criterion entry, corresponding to the latter part of the feedback delay step: The feedback delay stage uses a dual-segment criterion of hard delay and soft delay to determine the timing of the closed-loop entry.

[0034] Fixed hard delay time: Timing begins from the moment the shield opens, ensuring the most severe transient phase after the shield opens has passed, such as the rebalancing of the molten pool surface and the drastic rate fluctuations during the initial stage of vapor flow establishment, preventing premature loop closure. Before the fixed hard delay time arrives, although the real-time monitoring module continuously calculates the rate of change information, the soft delay criterion's judgment result is ineffective, and the loop closure command is forcibly blocked by the hard delay gating signal to prevent premature loop closure during the most severe transient period after the shield opens. After the fixed hard delay time arrives, the soft delay criterion gating is released. If the rate of change for N consecutive sampling periods is lower than the preset rate of change threshold, the soft delay is deemed satisfied, and a loop closure command is generated together with the hard delay condition. The fixed hard delay time is taken as the historical statistical duration of the period of drastic rate fluctuations after the shield opens for the coating material, plus a margin (e.g., 2-3 seconds).

[0035] Adaptive soft delay: After the fixed hard delay time arrives, based on the rate of change information generated in stage 3, the rate of change |ΔRate / Δt| for each sampling period is judged in real time. When the rate of change for N consecutive sampling periods is lower than the preset rate of change threshold, it is determined that the violent transient has ended and the soft delay criterion is met. N is recommended to be 3~10; the rate of change threshold is more than 3 times the standard deviation of the noise of the rate of change in the steady state to ensure that the noise does not misjudge "stable" and is not too large to miss the real fluctuation.

[0036] The closed-loop entry condition is that the fixed hard delay time arrives and the soft delay criterion is satisfied. For example... Figure 4 As shown, the logic of the dual-segment criterion is as follows: No intervention occurs until the fixed hard delay time is reached (hard delay as a fallback to prevent premature intervention); no intervention occurs until the fixed hard delay time is reached but the rate of change is not stable (soft delay as a waiting condition to prevent late intervention); a closed-loop intervention command is generated when both conditions are met simultaneously. The duration of transients varies greatly under different plating materials and operating conditions, and a fixed delay cannot adequately address both. The hard delay, with its fixed value, ensures that the transient will pass under the worst-case scenario. The soft delay uses an adaptive criterion to determine whether a severe transient has ended based on the actual rate of change. If the rate of change slows down early, intervention is initiated earlier to improve efficiency; if the rate of change slows down late, intervention is delayed to ensure safety.

[0037] Therefore, compared with the scheme that uses "whether the rate falls within a certain numerical range and lasts for a certain period of time" as the criterion, this two-stage criterion is based on the rate of change of the rate rather than the absolute value of the rate. Thus, it can make an entry judgment when the absolute value of the rate has not yet stabilized near the target value, but its trend of change has become stable. At the same time, it uses a fixed hard delay time as a pre-threshold to avoid premature judgment before the most intense transient has ended.

[0038] It is important to note that during the feedback delay phase, because the heating power is maintained constant at the estimated steady-state power, the measured coating rate will not spontaneously stabilize near the target coating rate. The physical meaning of the rate of change stabilizing during this phase is that the most severe transient disturbance introduced by the opening of the baffle (reconstruction of thermal equilibrium on the molten pool surface, turbulence in the initial stage of the steam flow field, etc.) has passed, and the system has entered a relatively gentle 'slow change' phase. At this point, switching to closed-loop regulation is safe and efficient—subsequent pure proportional control is sufficient to pull the rate from the current value into the tolerance zone within a few seconds. Therefore, the soft delay criterion in this scheme is not to determine that 'the rate has accurately reached the target value,' but rather to determine that 'the most severe transient has ended,' a positioning that perfectly matches the physical reality of constant power open-loop monitoring.

[0039] Phase 5 Two-Stage Closed-Loop Switching, corresponding to the closed-loop control steps: After the closed-loop cut-in command is generated, it does not directly switch to PI or PID control containing integral terms, but switches in two stages: The first stage – pure proportional control: After switching in, it initially operates with pure proportional control. P control only calculates the power adjustment based on the current deviation and the proportional coefficient, without accumulating historical deviations. Naturally, it will not produce integral saturation, quickly pulling the measured speed close to the target speed.

[0040] The second section – Activating the integral term: The integral term is activated and switched to PI or PID control when any of the following conditions are met: (a) the measured coating rate enters the tolerance band of the target coating rate; (b) the pure proportional control runtime reaches the maximum allowable runtime T_pmax. The control model used after activation is selected based on the coating material type and the rate range to which the target coating rate belongs: PI control is used for rates greater than or equal to 5.0 Å / s, and PID control is used for rates less than or equal to 2.0 Å / s with derivative vibration suppression enabled.

[0041] Regarding the maximum permissible operating time T_pmax: To avoid the situation where the measured coating rate remains outside the tolerance band for a long time due to the inherent steady-state error of pure proportional control, and the integral activation condition cannot be met, T_pmax is set to a value based on the coating material type and rate range (recommended: 3~5s for high coating rate metal films, 5~10s for low coating rate precision films). When T_pmax is reached, the integral term is forcibly activated regardless of whether the rate has entered the tolerance band. Since the power has been pre-adjusted to the estimated steady-state power before integral activation and the deviation is bounded, the absolute value of the deviation during forced activation is still constrained by the tolerance band width and pre-adjustment accuracy, and will not cause integral saturation, thus eliminating timing deadlock during closed-loop switching from a mechanism perspective. Similarly, when the measured coating rate normally enters the tolerance band and the integral term is activated, the deviation within this tolerance band is also a bounded small deviation: its lower bound is determined by the pre-adjustment accuracy of the estimated steady-state power and the convergence capability of pure proportional control, and its upper bound is the tolerance band half-width δ. Therefore, regardless of the conditions under which the integral term is activated, the initial accumulated deviation is a bounded small deviation, and there is no precondition for integral saturation. This is precisely what distinguishes this invention from conventional integral separation control: conventional integral separation only stipulates that "integration is engaged when the deviation is small" and does not change the magnitude of the deviation itself at the cut-in moment. However, this invention, through power pre-adjustment before the baffle is opened, makes the deviation at the cut-in moment itself a bounded small deviation.

[0042] Regarding the determination of the tolerance band: The half-width δ of the tolerance band is determined and verified according to the following principles: ① δ ≥ 3σ, where σ is the standard deviation of the measurement noise of the film thickness gauge at the current rate setting (which can be obtained from the sampling statistics of the uniform speed stable section), ensuring that the noise does not cause repeated oscillations due to integral activation; ② δ is less than the bandwidth corresponding to the upper limit of the allowable fluctuation of the coating rate, ensuring that the tolerance band reaches the upper limit of the allowable fluctuation of the coating rate before the coating rate; ③ Verify according to the uniformity index of the initial film thickness. If the process requires the average rate deviation of the initial section to not exceed ±5%, then take δ = 5%. Recommended values: δ = 5% for high coating rate metal films, δ = 10% for medium and low rate films, and δ = 15% for coating materials with high measurement noise or easy sputtering. That is, the tolerance band width of the same coating material can be set differently according to the rate setting and noise level.

[0043] The method for determining the standard deviation σ of the film thickness gauge noise is as follows: under the condition that the evaporation source is closed or the coating rate is stable and there is no external disturbance, at least 100 measured coating rate data points are continuously collected with the same sampling period as normal coating, and the sample standard deviation is calculated as σ; this value is re-measured and updated after each equipment maintenance or replacement of the crystal control wafer.

[0044] Regarding the tuning of the proportional coefficient Kp in the pure proportional segment: Tuning should be based on the coating material type and rate range. Recommended ranges: High coating rate metal films (≥5.0 Å / s) Kp=0.6~1.0; Low coating rate precision films (≤2.0 Å / s) Kp=0.3~0.6; For the intermediate rate range, refer to the nearest value from both. The initial value can be estimated using Kp≈ΔP_max / (δ·R_set) (ΔP_max is the maximum allowable power adjustment in a single control cycle, δ is the tolerance band half-width, and R_set is the target coating rate), and then fine-tuned through on-site trial coating. The steady-state PI / PID parameters should use the existing tuning values ​​for this coating material in the process parameter library and should not be readjusted due to the two-stage switching.

[0045] Regarding the perturbation-free assignment of the integral term: At the activation time t0 of the integral term, let the current deviation be e(t0) and the current power setpoint be P(t0). For the positional algorithm, let the initial value of the integral term be I(t0) = P(t0). Kp·e(t0) ensures that the controller output remains continuous and the power setpoint does not jump during the switching instant; for incremental algorithms, the integral increment of the first control cycle after activation is set to zero. Combined with conventional limitations on the single power adjustment, this ensures that the power jump during the two-stage switching instant is bounded.

[0046] This two-stage switching strategy completely bypasses complex calculations such as integral preloading and integral tracking, and there is no need to worry about large deviations in the starting point at the end of the integral term inheritance delay period—because no integral term is running during the transition period. It is simple to implement and eliminates integral saturation at the moment of switching in principle.

[0047] like Figure 3As shown, the initial transient suppression process described above is divided into three stages in terms of timing: the power pre-adjustment stage corresponds to before the shielding is opened, during which the heating power decreases from the pre-melting heating power in a step or ramp manner to the estimated steady-state power; the feedback delay stage corresponds to after the shielding is opened until the dual-segment criterion is met, during which the heating power is maintained at the estimated steady-state power and operates in an open-loop manner, with the measured coating rate increasing from zero. During this stage, only monitoring is performed without closed-loop control, and the rate information does not enter the power feedback path; the closed-loop entry stage corresponds to after the dual-segment criterion is met, during which pure proportional control is first executed, and after the measured coating rate enters the tolerance band of the target coating rate (or reaches the maximum running time), it switches to PI / PID control with an integral term, and the measured coating rate gradually approaches and stabilizes at the target coating rate. After activating the integral term, PI or PID control is used for continuous closed-loop adjustment until the coating is completed.

[0048] The following are examples of practical application scenarios provided in this application: Example 1: High-rate aluminum film deposition; Target coating rate: 5.0 Å / s; Control mode: PI control; Gain=0.8; Time-C=0.5; Pre-melting pool stabilization: With the shield closed, the aluminum material is fully melted to form a stable molten pool. Power pre-adjustment: Pre-melting heating power is 8.5 kW, estimated steady-state power is 6.2 kW, and the estimated steady-state power is taken from the previous steady-state power record of aluminum film coating under the same conditions. Before the shield is opened, the power is adjusted from 8.5 kW to 6.2 kW within 1 second using a ramp method. Feedback delay: After the shield is opened, only monitoring is performed without closed-loop control, and the heating power is maintained at 6.2 kW. Fixed hard delay time is 2 s; Soft delay criterion: Sampling period is 100 ms, N=5, and the rate of change threshold is 0.1 Å / s / sampling period. Two-stage cut-in: First, P control is used for operation (T_pmax=4s). After the measured rate enters the tolerance band of 5.0±0.25Å / s (±5%) and remains within the band for three consecutive sampling cycles, the integral term is activated by non-perturbation assignment to switch to PI control. Results: In this embodiment, no significant overshoot occurred during the delay period in a single operation. There was no overshoot or power jump during the cut-in process. The steady-state error was ≤±0.1Å / s, and the film thickness uniformity was ≤3%. (The statistical average overshoot of 5 repeated experiments was approximately 16%, see Example 3).

[0049] Example 2: Deposition of silicon oxide optical films at low deposition rates; Target coating rate: 1.0 Å / s; Control mode: PID control, Gain=0.5, Time-C=0.3, Differentiation=0.2. Pre-melting and stabilizing the pool: With the shield closed, the silicon oxide is fully melted. Power pre-adjustment: Pre-melting heating power: 3.2kW; Estimated steady-state power: 2.1kW. The estimated steady-state power is obtained by regression analysis of historical silicon oxide film coating steady-state power records. Before the shield is opened, the power is adjusted from 3.2kW to 2.1kW in three steps. Feedback delay: After the shield is opened, monitoring is only performed without closed-loop control; the heating power is maintained at 2.1kW; Fixed hard delay time: 3s; Soft delay criterion: Sampling period 200ms, N=5, rate of change threshold 0.02Å / s / sampling period. Two-stage cut-off: First, P control is used for operation (T_pmax=8s). After the measured speed enters the 1.0±0.05Å / s (±5%) tolerance band and remains within the band for three consecutive sampling cycles, the integral term is activated by non-disturbance assignment, switching to PID control. Result: Steady-state accuracy ±0.02Å / s, no oscillation, smooth cut-off process without overshoot.

[0050] Example 3: Comparative verification; To verify the effectiveness of the three-stage strategy for initial transient suppression of the present invention, under the same conditions of vacuum evaporation machine, evaporation source, quartz crystal film thickness gauge, charge amount and background vacuum, a control group and an experimental group (the present invention) were set up for aluminum film coating process with a target coating rate of 5.0 Å / s. Each group was subjected to 5 repeated coatings, and the arithmetic mean of each index was taken for comparison.

[0051] The control group used a fixed delay of 3s, maintained the pre-melting heating power of 8.5kW during the delay period, and directly switched to PI control after the delay (PI parameters were the same as the experimental group): the average peak rate during the delay period reached 7.2 Å / s, with an overshoot of about 44%; after switching to PI control, the integral term accumulated from the starting point of the large deviation, resulting in a second overshoot peak of 6.8 Å / s; the average time from the opening of the shield to the rate stabilizing within the range of 5.0±0.1 Å / s was about 8s; and the initial film thickness uniformity deviation was large.

[0052] The experimental group used this invention: before the shield was opened, the power was pre-adjusted to the estimated steady-state power of 6.2kW, and the average peak rate during the delay period was only 5.8Å / s, with an overshoot of about 16%; after fixing the hard delay by 2s, the soft delay criterion was met at 2.5s (the rate of change was below the threshold of 0.1Å / s / s / sampling period for 5 consecutive sampling periods), and pure proportional control was switched in; after running pure proportional control for 1.2s, the rate entered the ±5% tolerance band and remained there for 3 consecutive sampling periods, and the integral term was activated by the perturbation-free assignment to switch to PI control; there was no secondary overshoot throughout the process, and the average time from the opening of the shield to stabilization within the range of 5.0±0.1Å / s was about 3.7s; the initial film thickness uniformity was significantly better than that of the control group.

[0053] It should be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0054] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for controlling the evaporation source in a vacuum evaporation deposition machine, characterized in that, Includes the following steps: Pre-melting step: Close the shielding plate and perform pre-melting stabilization treatment on the coating material to obtain the target coating rate and the pre-melting heating power at the end of the pre-melting stage; Power pre-adjustment step: During the time period before the shield is opened, obtain the estimated steady-state power corresponding to the target coating rate, and adjust the heating power from the pre-melting heating power to the estimated steady-state power. Feedback delay step: After opening the shield, maintain the heating power in open loop at the estimated steady-state power, obtain the measured coating rate information and calculate the rate of change information, wherein the measured coating rate information is not used for closed-loop adjustment of the heating power; use a fixed hard delay time as a pre-gating condition, and after the fixed hard delay time is reached, use the rate of change information being continuously lower than a preset threshold as an adaptive soft delay criterion that the violent transient has ended; when the fixed hard delay time is reached and the adaptive soft delay criterion is satisfied, generate a closed-loop cut-in command. Closed-loop adjustment steps: In response to the closed-loop cut-in command, a pure proportional control command is first generated to adjust the heating power. When the measured coating rate enters the tolerance band of the target coating rate, or the running time of the pure proportional control reaches the preset maximum running time, a PI or PID control command containing an integral term is generated to adjust the heating power in a closed loop. When the integral term is activated, the initial value of the integral term is assigned to a value that keeps the power setpoint continuous at the moment of switching.

2. The method according to claim 1, characterized in that, The estimated steady-state power is obtained by establishing a mapping relationship between the target coating rate and steady-state power through regression analysis of the steady-state power records of historical coating processes, and then querying or interpolating based on the current target coating rate; or by matching and searching the historical record library according to the working condition label, and directly calling the steady-state power record of the previous coating under the same working condition.

3. The method according to claim 1, characterized in that, The heating power is adjusted from the pre-melting heating power to the estimated steady-state power through step-by-step or ramp-like adjustments.

4. The method according to claim 1, characterized in that, The adaptive soft delay criterion and the closed-loop cut-in command are determined in the following manner: The fixed hard delay time is a preset fixed value, which is counted from the moment the cover is opened. Before the fixed hard delay time is reached, even if the adaptive soft delay criterion is met, the closed-loop cut-in command will not be generated. After the fixed hard delay time is reached, when the rate of change of the rate of change of N consecutive sampling periods is lower than the preset rate of change threshold, it is determined that the adaptive soft delay criterion is met, and the closed-loop cut-in command is generated.

5. The method according to claim 1, characterized in that, The determination of whether the measured coating rate enters the tolerance band of the target coating rate includes the anti-jitter criterion: the measured coating rate is determined to be within the tolerance band if the measured coating rate is within the tolerance band for M consecutive sampling periods, and the counting is restarted if any sampling point leaves the tolerance band during the period; the tolerance band is the range of ±5% to ±15% of the target coating rate, and the half-width of the tolerance band is not less than 3 times the standard deviation of the film thickness gauge measurement noise and less than the bandwidth corresponding to the upper limit of the allowable fluctuation of the coating rate.

6. The method according to claim 1, characterized in that, The preset maximum running time is set according to the coating material type and the rate range to which the target coating rate belongs; when the running time of pure proportional control reaches the preset maximum running time, regardless of whether the measured coating rate enters the tolerance band, the integral term is forcibly activated and switched to PI or PID control.

7. The method according to claim 1, characterized in that, When generating PI or PID control commands containing integral terms, the control model is selected according to the coating material type and the rate range to which the target coating rate belongs: when the target coating rate is greater than or equal to 5.0 Å / s, a PI control command is generated; when the target coating rate is less than or equal to 2.0 Å / s, a PID control command is generated and differential vibration suppression is enabled.

8. An evaporation source control system for a vacuum evaporation deposition machine, characterized in that, include: A coating rate setting module is used to obtain the target coating rate and the tolerance band of the target coating rate; The real-time monitoring module is used to collect measured coating rate information in real time; The deviation calculation and control module is used to obtain the target coating rate and the measured coating rate, calculate the deviation, and generate power adjustment commands. This includes generating a pure proportional control command before closed-loop switching, generating a PI or PID control command containing an integral term after activating the integral term, and assigning the initial value of the integral term to a value that keeps the power setpoint continuous at the moment of switching when activating the integral term. The control model of the PI or PID control command is selected according to the coating material type and the rate range to which the target coating rate belongs. A power output module is used to acquire the power adjustment command and generate a power output signal adapted to the evaporation source; The feedback delay module is used to adjust the heating power from the pre-melting heating power at the end of the pre-melting stage to the estimated steady-state power corresponding to the target coating rate before the shield is opened. After the shield is opened, it acquires the measured coating rate information, generates monitoring data and rate change information, and maintains the estimated steady-state power. Based on the adaptive soft delay criterion determined by the fixed hard delay time and the rate change information, it generates a closed-loop cut-in command and controls the operation to first run in pure proportional control. After the measured coating rate enters the tolerance band or the pure proportional control operation reaches the preset maximum running time, it activates the integral term to switch to PI or PID control. The adaptive soft delay criterion uses the rate change information continuously being lower than a preset threshold as the basis for judging that the violent transient has ended.

Images