Method for recovering and on-line purity monitoring of organic solvents in optical module packaging

Abstract

The application discloses a kind of organic solvent rectification recovery and on-line purity monitoring method for optical module packaging, belong to the cross technical field of optoelectronic device manufacturing and industrial waste solvent resource recovery, the whole process technical system from waste solvent pretreatment to closed loop reuse is constructed, high-purity purification is realized using multi-tank complex rectification and membrane pervaporation coupling process, high-precision purity detection is realized through four-source sensing array combined with DNN deep neural network model, and adaptive closed loop optimization of rectification process is realized by combining MPC model predictive control.The purity of the solvent recovered by the application is stable at 99.8% or above, the recovery rate is more than 95%, the energy consumption is reduced by more than 40%, and the application is suitable for the continuous production requirements of 100G-1.6T full series optical module production line, and has significant economic and environmental benefits.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary technical fields of optoelectronic device manufacturing, industrial organic waste solvent resource recovery and process intelligent control. Specifically, it relates to a method for the distillation recovery and online purity monitoring of organic solvents used in optical module packaging. It is applicable to the closed-loop recovery, real-time quality monitoring, intelligent adaptive control and graded safe reuse of organic solvents used in the entire process of packaging high-speed optical modules of 100G, 400G, 800G, 1.6T and higher specifications, including chip precision cleaning, fiber end face treatment, optical adhesive dilution and removal, packaging cavity degreasing, and microlens bonding pretreatment. It can directly meet the continuous industrial operation requirements of large-scale optical module production lines. Background Technology

[0002] With the explosive growth of 5G / 6G communications, hyperscale data centers, and AI computing infrastructure, high-speed optical modules are rapidly evolving towards higher integration and reliability. Organic solvents, as core auxiliary materials for optical module packaging, are widely used in critical processes such as precision chip cleaning, optical adhesive dilution, and flux residue removal. Different rate-level optical modules have progressively stringent quality requirements: 100G / 400G low-power products require a main component purity ≥99.5% and a moisture content ≤50ppm; 400G / 800G medium-power products require a purity ≥99.8% and a moisture content ≤20ppm; and 800G / 1.6T ultra-high-power products require a purity ≥99.9% and a moisture content ≤10ppm. Even trace impurities can lead to decreased optical coupling efficiency and batch device failures. A single large-scale optical module production line can consume tens of tons of organic solvents annually, with 70%-80% ultimately discharged as waste solvent. Efficient recycling technologies are urgently needed to achieve resource recycling and cost reduction.

[0003] Current technologies for the recovery and monitoring of such waste solvents in the industry still suffer from four major systemic defects, failing to meet the demands for green, high-yield, and large-scale production of high-speed optical modules:

[0004] Firstly, the separation processes lack specificity, and there is an irreconcilable contradiction between separation accuracy and energy consumption control. Existing technologies mostly adopt conventional single-tower distillation, simple distillation, or single membrane separation processes, without being customized for the characteristics of multi-component mixing, easy formation of azeotropic systems, and diverse trace impurities in optical module waste solvents. The recovery purity fluctuates greatly, making it difficult to consistently meet the industry requirement of over 99.8%. The energy consumption per unit of solvent recovery is as high as 1.2-1.5 kWh / kg, and the solvent recovery rate is generally below 90%, resulting in resource waste and high hazardous waste disposal costs.

[0005] Secondly, purity monitoring technology is severely lagging behind, making real-time and accurate quality control impossible. Existing technologies heavily rely on offline gas chromatography, Karl Fischer titration, and other laboratory testing methods, with single testing cycles ranging from 30 minutes to 2 hours. This significantly lags behind the distillation process, easily leading to the mass production of substandard products and even causing a 3%-7% decrease in the yield of optical module packaging. Existing online monitoring solutions mostly use single sensors, which are susceptible to interference from temperature and impurities, resulting in insufficient detection accuracy. Furthermore, they fail to address industry challenges such as spectral model drift and probe contamination, making long-term stable operation impossible.

[0006] Third, monitoring and control are completely disconnected, failing to form a closed-loop intelligent management and control system. Existing distillation systems mostly adopt a static operation mode with a fixed reflux ratio and constant heating power, which cannot dynamically adjust operating parameters according to fluctuations in feed composition and changes in product purity. This results in strong control lag, and the conservative strategy of over-distillation is commonly used, leading to a significant increase in energy consumption. At the same time, a solvent quality classification and recycling matching mechanism adapted to the optical module packaging process has not been established, making it impossible to achieve optimal resource allocation and easily causing packaging defects or ineffective waste of high-quality solvents.

[0007] Fourth, the technical solutions lack completeness and cannot meet the needs of continuous industrial production. Existing recycling solutions mostly focus on the single step of distillation and separation, lacking a complete system for pretreatment, harmless treatment of waste liquid, waste heat recovery, equipment self-maintenance and safety protection. They cannot effectively solve industrial pain points such as solid impurities in optical module waste solvent clogging equipment and VOCs leakage, and are difficult to adapt to the 24-hour uninterrupted production needs of optical module mass production lines.

[0008] With increasingly stringent national environmental protection policies and continuously tightening requirements for VOCs control and hazardous waste disposal, the demand for cost reduction, efficiency improvement, and yield enhancement in the large-scale production of high-speed optical modules is becoming increasingly urgent. Developing a dedicated method for optical module packaging that combines high separation accuracy, low operating energy consumption, real-time accurate monitoring, intelligent closed-loop control, and full-process safety and environmental protection for organic solvent distillation recovery and online purity monitoring has become a core technical problem that the industry urgently needs to solve. Summary of the Invention

[0009] The purpose of this invention is to overcome the aforementioned deficiencies of the prior art and provide a method for the distillation recovery and online purity monitoring of organic solvents used in optical module packaging. This method, targeting the unique properties of waste solvents from optical module packaging, constructs a complete technical system encompassing "pretreatment homogenization - multi-tank composite distillation - membrane coupling deep purification - multi-source fusion online monitoring - intelligent purity inversion - adaptive closed-loop control - graded safe reuse - environmentally friendly and energy-saving support." This addresses the core pain points of existing technologies, such as low separation accuracy, high energy consumption, monitoring lag, control disconnect, mismatched reuse, and poor industrial adaptability. It enables efficient and high-purity recovery of various types of organic solvents used in optical module packaging, ensuring stable solvent quality that meets the requirements of a full range of optical module packaging. Simultaneously, it significantly reduces raw material consumption and hazardous waste emissions, improves packaging yield, and possesses both outstanding technological innovation and industrial practicality.

[0010] To achieve the above-mentioned objectives, the present invention adopts the following technical solution: a method for the distillation recovery and online purity monitoring of organic solvents used in optical module packaging, comprising the following steps:

[0011] S1 Waste Solvent Classification Collection and Pretreatment Homogenization: Waste organic solvents generated in the optical module packaging process are classified and collected according to solvent type. After passing through two-stage precision filtration, molecular sieve adsorption dehydration, composite adsorption impurity removal, and vacuum degassing, they are sent to the homogenization feed tank to complete temperature and component homogenization, remove solid impurities, trace moisture, metal ions and organic impurities from the waste solvent, and obtain a qualified feed liquid with stable physical properties.

[0012] S2 Multi-Tank Composite Distillation and Separation: After preheating the qualified feed liquid using the system's waste heat, it is sent to a customized multi-tank composite distillation column. Using a segmented distillation process, the temperature, pressure, and reflux ratio parameters of each distillation section are precisely controlled to achieve precise separation of light component impurities, target solvent, and high-boiling-point heavy components. An equilibrium coefficient is introduced to control the endpoint of the total reflux operation, resulting in a pre-purified solvent.

[0013] S3 membrane pervaporation coupled with deep purification: The pre-purified solvent is fed into the polyimide pervaporation membrane module to further remove trace amounts of water and organic impurities that cannot be separated by distillation, resulting in high-purity recovered solvent. At the same time, the purification effect is verified in real time through online monitoring.

[0014] S4 Multi-Source Fusion Online Purity Monitoring: Multiple monitoring points are set up along the material flow direction. Each point is equipped with a four-source sensor array consisting of an ultraviolet-near-infrared broadband spectral probe, a high-precision online refractometer, a dual-frequency microwave dielectric constant sensor, and a high-frequency thermal conductivity trace moisture meter. The array simultaneously collects the full-dimensional physical property parameters of the solvent to be tested, realizing real-time, all-round acquisition of the purity status of the entire process.

[0015] S5 Multi-Source Data Fusion and Intelligent Purity Inversion: After time synchronization, temperature compensation and preprocessing of the collected multi-source data, a spectral-physical property joint feature vector is constructed and input into a pre-trained deep neural network (DNN) purity inversion model. The model outputs solvent comprehensive purity index, main component content, water content, UV absorption impurity concentration and key impurity mole fraction in real time, realizing second-level, high-precision quantitative purity analysis.

[0016] S6 Adaptive Dynamic Distillation Closed-Loop Control: With real-time purity data as the core input, it adopts a model predictive control (MPC) architecture and combines a preset purity grading standard for optical modules to dynamically adjust the reflux ratio, heating power, outlet valve position and feed rate of the distillation system. At the same time, it realizes purity trend early warning and pre-adjustment through statistical process control, achieving adaptive optimization and closed-loop control of the distillation process.

[0017] S7 Solvent Quality Grading and Production Line Closed-Loop Reuse: Based on real-time purity test results, recycled solvents are classified into grades. Different grades of solvents correspond to different rate-level optical module packaging processes. By connecting with the production line MES system, a full lifecycle traceability system for recycled solvents from reuse to packaging is established to achieve safe and efficient closed-loop reuse.

[0018] The S8 system features self-maintenance, waste liquid treatment, and waste heat recovery: It achieves long-term stable operation of the system through dual-optical-path self-cleaning and sensor health assessment, ensures environmental compliance by harmlessly treating distillation waste liquid, and realizes energy recycling through a waste heat cascade recovery system, thereby reducing the overall energy consumption of the system.

[0019] Furthermore, step S1 specifically includes the following sub-steps:

[0020] S11 Classification Collection: Using pipe color markings and RFID electronic tags, waste solvents generated in each process of optical module packaging are collected separately into corresponding sealed waste liquid storage tanks according to the type of isopropanol, acetone, N-methylpyrrolidone (NMP), dimethyl carbonate (DMC), propylene glycol methyl ether acetate (PGMEA), and fluorinated solvents. This avoids cross-contamination between different solvents. The waste liquid storage tanks are equipped with liquid level monitoring and nitrogen sealing protection devices.

[0021] S12 Two-Stage Precision Filtration: Waste solvent in the waste liquid storage tank is sent to a precision filtration device, employing a two-stage filtration process: The first stage uses a polypropylene filter membrane with a pore size of 5-10μm, controlling the filtration pressure at 0.1-0.3MPa, the filtration temperature at 25-35℃, and the filtration speed at 5-10L / min, removing large solid impurities such as chip debris and encapsulation adhesive particles from the waste solvent; the second stage uses a polytetrafluoroethylene filter membrane with a pore size of 0.22-0.45μm, controlling the filtration pressure at 0.2-0.35MPa and the filtration speed at 4-6L / min, removing fine solid impurities and colloidal particles from the waste solvent; a differential pressure monitoring device is installed during the filtration process, automatically prompting the replacement of the filter membrane when the pressure difference across the membrane reaches 0.1MPa;

[0022] S13 Molecular Sieve Adsorption Dehydration: The filtered waste solvent is sent to a dehydration device, which adopts a closed-loop dehydration process of "3A / 4A molecular sieve adsorption + online dew point monitoring". The dehydration device is filled with 3A or 4A spherical molecular sieves, with a filling amount of 10%-15% of the waste solvent volume. The residence time of the waste solvent in the dehydration device is controlled at 30-60 minutes. Trace amounts of water in the waste solvent are removed by selective adsorption of the molecular sieves. An online dew point meter is installed at the outlet of the dehydration device with a detection accuracy of ±5ppm to monitor the water content of the solvent after dehydration in real time. If the water content exceeds 50ppm, the solvent is automatically returned to the dehydration device for rehydration until the standard is met. The saturated molecular sieves are recycled through a 200-250℃ thermal regeneration process, with a regeneration time of 2-3 hours.

[0023] S14 Composite Adsorption Depurification: The dehydrated waste solvent is fed into a depurification device, employing a process combining modified activated carbon adsorption and chemical depurification. The device is filled with modified activated carbon activated by nitric acid, with a specific surface area ≥1000m² / g. The residence time of the waste solvent in the device is controlled at 20-40 minutes. Organic impurities such as encapsulant degradation products and surfactants are removed through activated carbon adsorption. Simultaneously, 0.5%-1% of activated alumina depurifying agent by mass of the waste solvent is added to the device, which undergoes physical adsorption and complexation reactions with metal ions and polar organic impurities in the waste solvent, further removing impurities. After depurification, 0. 22μm precision filtration removes impurity-removing agent particles, yielding a qualified solvent. S15 Vacuum degassing and homogenization: The qualified solvent is fed into a vacuum degassing module, with operating pressure controlled at ≤5kPa, temperature at 30-40℃, and residence time at 10-15 minutes, to remove dissolved oxygen and low-boiling-point volatile impurities from the solvent. After degassing, the solvent is fed into a homogenizing feed tank with jacketed temperature control and variable frequency magnetic stirring, with stirring speed controlled at 80-120rpm and solvent temperature fluctuation ≤±1℃, ensuring uniform and stable feed components and properties to obtain a qualified feed liquid. After testing, the qualified feed liquid has an impurity content ≤0.1% and a moisture content ≤50ppm.

[0024] Furthermore, step S2 specifically includes the following sub-steps:

[0025] S21 Waste Heat Preheating: The qualified feed liquid is sent into the shell and tube preheater. The waste heat recovered by the distillation system is used to preheat the feed liquid to 10-15℃ below the boiling point of the target solvent. A temperature monitoring device is set up during the preheating process to control the preheating temperature fluctuation within ±1℃ in real time, so as to avoid the solvent from vaporizing in advance and affecting the distillation effect.

[0026] S22 Multi-Tank Composite Segmented Distillation: The preheated feed liquid is fed into a customized multi-tank composite distillation column, which consists of 3-5 distillation tanks connected in series. Each distillation tank corresponds to an independent rectification section, with a total of 20-30 theoretical plates. It adopts a floating valve tray structure, and the operating pressure inside the column is controlled at 0.08-0.12 MPa. The reboiler is heated by heat transfer oil, and the top of the column is condensed by circulating water. The specific segmented control is as follows:

[0027] (1) First rectification section: The temperature at the top of the column is controlled to be 5-10℃ lower than the boiling point of the target solvent. The main purpose is to separate and remove low-boiling-point impurities. After vaporization, the impurities are condensed by the top condenser and sent to the low-boiling-point impurity storage tank.

[0028] (2) Second to fourth rectification sections: corresponding to the boiling point range of different target solvents in the feed liquid, the top temperature, pressure and reflux ratio of each rectification section are individually controlled, the reflux ratio is controlled within the range of 1.5-4.0, and different target solvent components are separated in sequence according to the boiling point difference of each component;

[0029] (3) Fifth distillation section: The bottom temperature of the column is controlled to be 5-10℃ higher than the boiling point of high-boiling point impurities. It mainly enriches heavy components such as residual encapsulant and high-boiling point organic matter, and discharges them to the waste liquid treatment device periodically. The balance coefficient φ value is introduced into the distillation process to control the end point of the total reflux operation. The φ value is controlled at 0.8-0.9. While ensuring the separation effect, the operation time is shortened and the energy consumption is reduced.

[0030] S23 Preliminary purification solvent discharge: When online monitoring shows that the purity of the main component of the solvent discharged from the rectification section is ≥99.5%, the preliminary purified solvent is sent to the subsequent membrane pervaporation coupled purification process. Solvents that do not meet the standard are returned to the rectification tower for re-rectification.

[0031] Furthermore, step S3 specifically includes the following sub-steps:

[0032] S31 membrane module feed control: The pre-purified solvent is fed into the membrane pervaporation coupling device, and the feed temperature is controlled at 40-60℃, the feed pressure at 0.1-0.2MPa, and the residence time of the material in the membrane module is 10-20 minutes.

[0033] S32 Pervaporation Deep Purification: The membrane pervaporation device uses an organic solvent-resistant polyimide composite membrane with a membrane area of ​​5-10 m². The permeation side operating pressure is controlled at 0.01-0.03 MPa. Through the selective permeability of the membrane, trace amounts of water and organic impurities in the solvent are further removed, reducing the water content to below 10 ppm and the organic impurity content to below 0.05%.

[0034] S33 Purification Effect Verification: An online monitoring point is set at the membrane module outlet to detect the purity and impurity content of the purified solvent in real time. Solvents that meet the standards are sent to the finished product buffer unit, while solvents that do not meet the standards are returned to the membrane module inlet for re-purification or sent to the distillation column for re-distillation.

[0035] Furthermore, step S4 specifically includes the following sub-steps:

[0036] S41 Monitoring Point Layout: Along the material flow direction, 6-8 monitoring points are set up at the pretreatment outlet, the top discharge port of each distillation section, the membrane pervaporation unit outlet, and the finished product storage tank inlet to achieve full-process purity status coverage from feed to finished product.

[0037] S42 Closed-loop sampling: Each monitoring point is connected to a quartz glass detection flow cell via a bypass sampling pipeline. The flow cell has an optical path of 10mm and an internal volume of ≤0.5mL. The sampling flow rate is controlled at 5-50mL / min. The solvent after sampling is returned to the corresponding main pipeline to achieve closed-loop sampling and avoid solvent evaporation and external impurity contamination.

[0038] S43 Four-Source Sensor Array Deployment: One four-source sensor array is deployed for each flow cell at each monitoring point. The specific parameters are as follows:

[0039] (1) Ultraviolet-Near Infrared Broadband Spectrometer: The detection band covers the ultraviolet 200-400nm and the near infrared 900-1700nm. The ultraviolet spectral resolution is ≤1nm and the near infrared resolution is ≤2nm. The single full spectrum scan time is ≤1s. It has a built-in 30.0±0.1℃ constant temperature control module and adopts a dual optical path design of main measurement optical path and self-cleaning reference optical path.

[0040] (2) High-precision online refractometer: measurement range 1.3300-1.4500, measurement accuracy ±0.00005, temperature control accuracy ±0.05℃, prism surface coated with fluorinated silicon hydrophobic coating to reduce organic matter adhesion;

[0041] (3) Dual-frequency microwave dielectric constant sensor: working frequency 2.4GHz / 5.8GHz dual-frequency hopping, real output of the real part and imaginary part of the complex dielectric constant of the solvent to be tested, response time ≤1s;

[0042] (4) High-frequency thermal conductivity micro moisture meter: range 0-2000ppm, response time ≤3s, detection accuracy ±5ppm; all sensor signals are connected to the central processing unit through an explosion-proof isolation fence to realize synchronous data acquisition.

[0043] Furthermore, step S5 specifically includes the following sub-steps:

[0044] S51 Data Synchronization and Temperature Compensation: Employing the IEEE 1588PTP precision time protocol, microsecond-level timestamp alignment of data from the four-source sensor array is achieved; for temperature-sensitive refractive index and spectral data, a temperature compensation model is established to eliminate the interference of ambient temperature fluctuations on the detection results.

[0045] S52 Spectral Preprocessing and Feature Extraction: The raw near-infrared spectrum was sequentially subjected to Savitzky-Golay smoothing filtering (15-point window, second-order polynomial), multivariate scattering correction, and first-order derivative preprocessing to eliminate baseline drift and optical path difference; Principal component analysis (PCA) was used for dimensionality reduction, and the top 8 principal components with a cumulative contribution rate ≥98% were retained to obtain the spectral feature matrix.

[0046] S53 Joint Feature Vector Construction: The pre-processed refractive index, real / imaginary part of dielectric constant, and moisture content are Min-Max normalized to the [0,1] interval, and then concatenated with the spectral feature matrix to form a 12-dimensional joint feature vector;

[0047] S54DNN Purity Inversion Model Construction and Operation: A standard sample set with a concentration gradient covering 95%-99.999% was prepared for each commonly used organic solvent in optical module packaging. Multi-source sensor data of the standard samples were collected, and the DNN purity inversion model was trained using offline gas chromatography and Karl Fischer titration results as labels. The model structure is as follows: Input layer 12 nodes → First hidden layer 64 nodes (ReLU activation function) → Dropout layer (inactivation rate 0.2) → Second hidden layer 32 nodes (Tanh activation function) → Output layer 4 nodes, corresponding to the comprehensive purity index, principal component content, moisture content, and UV absorption impurity concentration, respectively. The model uses weighted mean square error as the loss function, assigning weight coefficients of 1.5-2.0 to moisture and UV absorption impurities. During actual operation, the joint feature vector is input into the trained DNN model corresponding to the solvent, and the purity detection results are output in real time. The model automatically performs a local calibration every 30 minutes using standard purity solvents, and automatically triggers incremental model updates when the prediction deviation > 2%.

[0048] S55 Early Warning Threshold Setting: Set a purity early warning threshold in the intelligent analysis module. When the content of the main component is lower than the corresponding grade standard, or the moisture content or impurity content exceeds the limit, an early warning signal will be issued immediately, and a closed-loop control mechanism will be activated at the same time.

[0049] Furthermore, step S6 specifically includes the following sub-steps:

[0050] S61 Purity Grading Standard Preset: Based on the packaging requirements of optical modules of different rate levels, three purity standards are preset and stored in the control system.

[0051] (1) Grade A high purity standard: main component content ≥99.9%, moisture content ≤10ppm, impurity content ≤0.03%, and Class I key process suitable for 800G / 1.6T ultra-high power optical modules;

[0052] (2) Grade B medium purity standard: main component content ≥99.8%, moisture content ≤20ppm, impurity content ≤0.05%, suitable for Class II conventional processes of 400G / 800G medium power optical modules;

[0053] (3) Grade C basic standard: main component content ≥99.5%, moisture content ≤50ppm, impurity content ≤0.1%, and Class III auxiliary process suitable for 100G / 400G low power optical modules;

[0054] S62MPC control model construction: Establish a discrete state-space model of the distillation process, with comprehensive purity index, moisture content, and reflux tank level as control variables, and reflux ratio, reboiler heating power, outlet valve position, and feed rate as manipulated variables. Set 18 steps in the prediction time domain and 6 steps in the control time domain, and perform rolling optimization solution every 10 seconds to achieve multi-variable collaborative optimization.

[0055] S63 graded closed-loop control execution: The control system compares real-time purity data with preset standards and executes the corresponding control strategy:

[0056] (1) When the test results meet the Class A standard, open the main outlet valve at the top of the column and automatically adjust the reflux ratio to 1.8-2.2 to reduce distillation energy consumption;

[0057] (2) When the test results meet the Class B standard, switch to the corresponding distillation section side-stream outlet valve, increase the reflux ratio to 2.5-3.2, increase the tower bottom heating power by 5%-8%, and enhance the separation effect;

[0058] (3) When the test results only meet the C-level standard, close the main collection valve, send the solvent into the auxiliary distillation section for further purification, and adjust the distillation parameters at the same time;

[0059] (4) When the test result is lower than the C-level standard, immediately close all sampling valves, return the solvent to the feed tank for reprocessing, and at the same time conduct a comprehensive inspection of the equipment operation status; for different types of purity abnormalities, perform precise parameter adjustments: if the main component content is too low and the impurity content is too high, increase the reflux ratio of the corresponding rectification section by 0.2-0.5, extend the solvent residence time by 10-20 minutes, and increase the tower temperature by 1-2℃; if the moisture content is too high, extend the molecular sieve dehydration residence time by 10-15 minutes and increase the membrane pervaporation device operating temperature by 5-10℃; S64 trend warning and pre-adjustment: introduce cumulative sum (CUSUM) statistical process control to identify the systematic deviation trend of purity data. When more than 6 consecutive sampling points are detected to show a unidirectional deviation, even if the purity is still within the qualified range, the parameter pre-adjustment is actively triggered to avoid substantial deviation in quality.

[0060] Furthermore, step S7 specifically includes the following sub-steps:

[0061] S71 Quality Grading and Diversion: Based on real-time purity test results, the recovered solvent is automatically diverted to the corresponding grade of stainless steel buffer tank. The inner wall of each buffer tank is polished to Ra≤0.25μm and equipped with a nitrogen-sealed micro-positive pressure protection system to control the pressure inside the tank at 0.02-0.05MPa to prevent the solvent from absorbing moisture and oxidizing.

[0062] S72 Intelligent Process Matching: Through the OPCUA protocol, it interfaces with the MES system of the optical module packaging production line to achieve intelligent matching of solvent grade and packaging process: Grade A solvent, after being filtered through a 0.05μm terminal filter, is supplied to key processes such as optical chip cleaning, fiber end face activation, optical adhesive dilution, and microlens bonding; Grade B solvent is supplied to routine processes such as encapsulation adhesive dissolution, packaging mold cleaning, and cavity pre-cleaning; Grade C solvent is supplied to auxiliary processes such as fixture degreasing and non-optical surface wiping.

[0063] S73 Full Lifecycle Traceability: Each batch of recycled solvent generates a unique traceability code, recording the incoming batch, all parameters of the distillation process, the purity trajectory of the entire process, and the final inspection report. The data is synchronously uploaded to the production line's digital twin platform, enabling full traceability of the solvent from reuse to packaging. When an abnormality occurs in packaging, the solvent quality data can be traced back to support process optimization.

[0064] Furthermore, step S8 specifically includes the following sub-steps:

[0065] The S81 system features self-maintenance and calibration: Every 2 hours of operation, the online monitoring probe automatically initiates 40kHz ultrasonic cleaning and 0.3MPa nitrogen purging to remove surface contaminants; a built-in sensor health assessment module automatically switches to redundant algorithm mode and issues a maintenance warning when the signal-to-noise ratio of a channel falls below a threshold; regular maintenance and performance calibration are performed on the filtration unit, adsorption column, distillation column, and membrane modules; a comprehensive calibration of the entire system is conducted every 3 months to ensure that the purity fluctuation of the recovered solvent is controlled within ±0.05%.

[0066] S82 waste liquid harmless treatment: The high-boiling-point waste liquid generated at the bottom of the distillation column is first pretreated with a neutralizing agent to reduce toxicity, and then sent to the incinerator to decompose organic impurities at a high temperature of 800-1000℃. The flue gas generated by incineration is discharged in compliance with standards after the harmful gases are removed by the activated carbon adsorption device. The incineration residue is solidified and then entrusted to a qualified hazardous waste disposal agency for treatment.

[0067] S83 Waste Heat Cascade Recovery: Waste heat generated in the distillation column bottom and top condenser is recovered through plate heat exchangers, with a waste heat recovery efficiency of ≥70%. The recovered waste heat is used sequentially for feed liquid preheating, molecular sieve thermal regeneration, and workshop heating, realizing the cascade utilization of energy and reducing the overall energy consumption of the system.

[0068] Compared with the prior art, the present invention has the following outstanding substantive features and significant beneficial effects:

[0069] 1. Significantly improved separation accuracy and recovery rate, and significantly reduced energy consumption. This invention adopts a two-stage purification process of "multi-tank composite segmented distillation + membrane pervaporation coupling," which is customized for the characteristics of waste solvents from optical modules. The purity of the main components of the recovered solvent is stably maintained at 99.8%-99.9%, and the moisture content can be reduced to below 10ppm, fully meeting the packaging requirements of 1.6T optical modules. The solvent recovery rate is increased to over 95%, which is more than 15 percentage points higher than existing technologies. Combined with a waste heat cascade recovery system, the energy consumption per unit of solvent recovery is reduced to below 0.75kWh / kg, which is 40%-50% lower than conventional distillation processes, resulting in significant energy savings.

[0070] 2. A qualitative breakthrough in monitoring accuracy and response speed, completely solving the pain point of detection lag. This invention constructs a four-source integrated monitoring system of "ultraviolet-near-infrared spectroscopy + refractive index + dielectric constant + trace moisture meter", combined with a DNN deep neural network intelligent inversion model, compressing the detection cycle from hours to less than 5 seconds, with a main component detection accuracy of ±0.05% and a moisture detection deviation of ±2ppm, which is an order of magnitude improvement in accuracy compared to single sensor solutions; through dual-optical-path self-cleaning, automatic calibration and multi-model parallel architecture, it can achieve more than 720 hours of continuous and stable operation, fully adapting to the needs of continuous industrial production.

[0071] 3. Intelligent closed-loop control upgrades from "passive correction" to "active prevention." This invention deeply integrates MPC model predictive control and statistical process control, achieving second-level linkage between purity data and distillation operation parameters. The standard deviation of recovered solvent purity fluctuation is controlled within 0.03%, a reduction of more than 60% compared to manual control mode. Through trend early warning and pre-adjustment mechanisms, the yield of defective products is reduced by more than 90%, avoiding optical module packaging defects caused by solvent purity issues. The optical coupling offset exceedance rate is reduced from 0.82% to below 0.21%, significantly improving packaging yield.

[0072] 4. The graded recycling system achieves optimal resource allocation, combining economic and environmental benefits. This invention establishes a solvent quality grading standard and intelligent matching mechanism adapted to the entire series of optical module packaging processes. This enables the precise supply of high-purity solvents to key processes and the rational use of medium-purity solvents in auxiliary processes. The amount of new solvent purchased is reduced by 40%-55%, saving more than 700,000 yuan in costs per year for a single 800G optical module production line, with an investment payback period of less than 8 months. The complete waste liquid harmless treatment system can reduce VOC emissions by more than 90%, fully complying with environmental protection policy requirements.

[0073] 5. The closed-loop design throughout the entire process is adapted to continuous industrial production, exhibiting exceptional versatility and stability. This invention constructs a complete system from waste solvent collection to reuse. The pretreatment stage effectively removes impurities specific to waste solvents from optical modules, preventing equipment blockage and reducing maintenance costs by more than 30% compared to existing technologies. A robust safety protection and calibration mechanism ensures continuous system operation for over 120 days without major failures. Through switchable side-line sampling outlets and a multi-model parallel architecture, it can adapt to the recovery needs of various types of solvents commonly used in optical modules with a single click, covering the entire range of 100G-1.6T optical module packaging scenarios. Its industrial versatility and engineering promotion value are extremely high. Attached Figure Description

[0074] Figure 1 This is a flowchart illustrating the overall process flow of the method for organic solvent distillation recovery and online purity monitoring in optical module packaging according to the present invention.

[0075] Figure 2 This is a schematic diagram of the multi-tank composite distillation and membrane coupling purification system of the present invention;

[0076] Figure 3 This is a schematic diagram of the architecture of the multi-source fusion online monitoring and intelligent purity inversion system in this invention;

[0077] Figure 4 This is a logic flowchart of the adaptive closed-loop control in this invention. Detailed Implementation

[0078] The present invention will be further described in detail below with reference to specific embodiments. These embodiments are implemented based on the technical solution of the present invention, and provide detailed implementation methods, specific operating parameters, and effect verification data. However, the scope of protection of the present invention is not limited to the following embodiments.

[0079] This embodiment focuses on the industrial-scale implementation of waste organic solvent distillation recovery and online purity monitoring on the mass production line of an 800G high-speed optical module from a major domestic optical module manufacturer. The production line has an annual capacity of 280,000 800G optical modules. The main organic solvents used in the packaging process are isopropanol (IPA) and N-methylpyrrolidone (NMP). Isopropanol is used for precision cleaning of the chip and fiber end faces, while NMP is used for dissolving the encapsulating adhesive and cleaning the mold. The production line generates approximately 120L of mixed waste solvent daily, containing about 60% isopropanol, about 35% NMP, and the remaining 5% consisting of encapsulating adhesive residue, chip debris, metal ions, moisture, and other impurities, with an initial moisture content of approximately 120ppm. The production line requires that the recovered isopropanol and NMP have a purity of ≥99.8%, a moisture content of ≤20ppm, and an impurity content of ≤0.05%, allowing for direct reuse in the packaging process. Simultaneously, the system must be able to operate stably for 24 hours continuously, with a unit solvent recovery energy consumption of ≤0.8kWh / kg.

[0080] I. System and Equipment Configuration

[0081] The system used in this embodiment is built entirely according to the technical solution of this invention, and the core equipment configuration is as follows:

[0082] 1. Pretreatment Unit: Waste liquid classification collection tank (316L stainless steel, 1000L capacity, with nitrogen sealing and liquid level monitoring), two-stage precision filtration device (first stage 8μm polypropylene filter element, second stage 0.22μm polytetrafluoroethylene filter element, with differential pressure monitoring), molecular sieve dehydration device (500L capacity, filled with 4A molecular sieve, filling amount 12% of waste solvent volume, with online dew point meter), composite adsorption and impurity removal device (800L capacity, filled with nitric acid activated modified activated carbon, with impurity removal agent filling port), vacuum degassing module (ultimate vacuum 3kPa, with temperature control), homogenization feed tank (500L capacity, with jacketed temperature control and variable frequency magnetic stirring);

[0083] 2. Distillation and purification unit: shell and tube preheater (heat exchange area 8m²), customized multi-tank composite distillation column (4 distillation tanks in series, corresponding to 5 distillation sections, a total of 25 trays, floating valve tray structure, with heat pump auxiliary heating and automatic reflux ratio control system), polyimide membrane pervaporation device (membrane area 8m², modular design, with automatic temperature and pressure control);

[0084] 3. Online monitoring and control unit: 8 monitoring points, each equipped with a four-source sensor array (UV-NIR broadband spectral probe, high-precision refractometer, dual-frequency dielectric constant sensor, trace moisture meter), industrial control computer (equipped with DNN purity inversion model and MPC adaptive control algorithm), and PLC automatic control system (with explosion-proof isolation barrier, and interface with the production line MES system).

[0085] 4. Environmental protection and auxiliary units: waste liquid incineration treatment device (with activated carbon flue gas adsorption), plate waste heat recovery device (heat exchange efficiency ≥70%), three-stage solvent buffer tank (Class A / Class B / Class C, with nitrogen sealing and terminal filtration), and safety protection system (explosion-proof, leakage detection and emergency handling device).

[0086] II. Specific Implementation Steps

[0087] Step 1: Waste solvent pretreatment and homogenization

[0088] The 500L of mixed waste solvent generated by this production line is sent to a waste liquid collection tank and treated by a two-stage precision filtration device: the first stage filtration is controlled at a pressure of 0.2MPa, a temperature of 30℃, and a filtration speed of 8L / min to remove large particulate impurities; the second stage filtration is controlled at a pressure of 0.25MPa and a filtration speed of 6L / min to remove colloids and fine particles; the maximum pressure difference during the filtration process is 0.08MPa, which does not reach the replacement threshold, so there is no need to replace the filter element;

[0089] The filtered waste solvent is sent to a molecular sieve dehydration device, with a residence time of 45 minutes. The moisture content is monitored in real time by an online dew point meter. The initial moisture content is 120 ppm, which is reduced to 42 ppm after dehydration, meeting the requirement of ≤50 ppm. No secondary dehydration is required.

[0090] The dehydrated waste solvent is sent to a composite adsorption and impurity removal device, with a residence time controlled at 30 minutes. At the same time, 0.8% of the waste solvent mass of activated alumina impurity removal agent is added to remove organic impurities and metal ions. After impurity removal, the solvent is filtered through a 0.22μm precision filter to remove impurity removal agent particles, resulting in a qualified solvent.

[0091] The qualified solvent after impurity removal is fed into the vacuum degassing module, where the operating pressure is controlled at 4.5 kPa, the temperature at 35℃, and the residence time at 12 minutes to remove dissolved oxygen and low-boiling-point impurities. It is then fed into the homogenizing feed tank, where the temperature is controlled at 40±0.5℃ and the stirring speed at 100 rpm to ensure uniform feeding. After testing, the qualified feed liquid has an impurity content of 0.08% and a moisture content of 40 ppm, which fully meets the feeding requirements.

[0092] Step 2: Multi-tank compound distillation separation

[0093] The qualified feed liquid is sent into a shell-and-tube preheater and preheated to 75°C (below the boiling point of isopropanol 7.4°C and below the boiling point of NMP 127°C) using the waste heat of distillation. The preheating temperature fluctuation is controlled within ±1°C and then sent into a multi-tank compound distillation column.

[0094] The distillation column is controlled at an internal pressure of 0.1 MPa. The reboiler is heated by heat transfer oil, and the cooling temperature of the top condenser is controlled at 20°C. An equilibrium coefficient of φ=0.85 is introduced to control the total reflux endpoint. The parameters of each distillation section are precisely controlled as follows:

[0095] First rectification section: The top temperature of the column is 72℃. Low-boiling-point impurities are separated and condensed before being sent to the impurity storage tank. A total of 2.1L of low-boiling-point impurities are collected.

[0096] Second rectification section: Top temperature 80-82.4℃ (boiling point of isopropanol), pressure 0.1MPa, initial reflux ratio controlled at 1.8, separation of isopropanol components;

[0097] Third rectification section: top temperature 195-202℃ (NMP boiling point), pressure 0.1MPa, initial reflux ratio controlled at 2.2, separating NMP components;

[0098] Fourth rectification section: The top temperature is 207℃. High-boiling-point impurities are separated and concentrated at the bottom of the column and discharged periodically, generating a total of 5.8L of bottom waste liquid.

[0099] Online monitoring showed that the purity of isopropanol from the second distillation section was 99.62%, and the purity of NMP from the third distillation section was 99.58%, both meeting the initial purification requirements. The NMP was then sent to a membrane pervaporation coupling device for further purification.

[0100] Step 3: Membrane pervaporation coupled with deep purification

[0101] The pre-purified isopropanol and NMP were fed into a membrane pervaporation unit, with the operating temperature controlled at 50°C, the permeate side pressure at 0.02 MPa, and the material residence time at 15 minutes. After deep purification, the purity of isopropanol was increased to 99.86% and the moisture content was 18 ppm; the purity of NMP was increased to 99.83% and the moisture content was 16 ppm, both meeting the preset Grade B standard and meeting the requirements for reuse in the production line.

[0102] Steps 4-5: Online monitoring and intelligent purity inversion

[0103] Data was simultaneously acquired by a four-source sensor array at eight monitoring points. After microsecond-level time synchronization and temperature compensation, a joint feature vector was constructed and input into a pre-trained IPA / NMP dual-solvent DNN inversion model, outputting purity data in real time with a detection cycle of 4.8 seconds per instance. The model was calibrated with standard samples, achieving a principal component detection accuracy of ±0.04% and a moisture detection deviation of ±2.1 ppm, operating stably throughout the process. During implementation, an early warning occurred in the second distillation stage, indicating an isopropanol purity of 99.75% and an impurity content of 0.06%. The intelligent analysis module immediately issued a warning signal and simultaneously activated the closed-loop control mechanism.

[0104] Step 6: Adaptive Closed-Loop Control

[0105] The control system is preset with the following target standards for this production line: main component ≥99.8%, moisture ≤20ppm, and impurities ≤0.05%. In response to the abnormal isopropanol purity in the second rectification section, the system automatically determined that the main component purity was too low and the impurity content was too high, and immediately implemented parameter adjustments: increasing the reflux ratio of the second rectification section from 1.8 to 2.0, extending the solvent residence time by 15 minutes, and increasing the column top temperature from 81℃ to 81.5℃. Ten minutes after the parameter adjustments, online monitoring showed that the isopropanol purity had increased to 99.83%, and the impurity content had decreased to 0.04%, returning to the acceptable range. The system maintained stable operation with the current parameters. Throughout the entire process, the system, through CUSUM statistical process control, proactively identified two purity decline trends and pre-adjusted parameters, resulting in no unqualified products being produced and stable operation.

[0106] Step 7: Quality Grading and Closed-Loop Reuse

[0107] After deep purification, both isopropanol and NMP met or exceeded Grade B standards in online testing, with over 90% of batches reaching Grade A standards. These were automatically diverted to corresponding grade buffer tanks. Grade A solvents, after being filtered through a 0.05μm terminal filter, were supplied to critical processes on the production line, such as precision chip cleaning and fiber optic end-face treatment. Grade B solvents were supplied to routine processes such as encapsulation adhesive dissolution and mold cleaning. The system generates a unique traceability code for each batch of solvents, and all process data is synchronously uploaded to the production line's MES system, achieving full lifecycle traceability. After the recycled solvents were reused on the production line, a packaging test was conducted on 1000 800G optical modules, achieving a packaging yield of 99.2%, essentially consistent with the 99.3% yield achieved using brand-new electronic-grade solvents. No packaging defects due to solvent purity were observed.

[0108] Step 8: Environmental Protection, Energy Conservation, and System Maintenance

[0109] The 5.8L waste liquid generated at the bottom of the distillation column, after neutralization and pretreatment with sodium hydroxide solution, is sent to a 900℃ incinerator for incineration. The flue gas is then purified by activated carbon adsorption to meet emission standards, complying with the "Integrated Emission Standard for Air Pollutants". The waste heat generated by the distillation system is recovered through a waste heat recovery device and used for feed liquid preheating and molecular sieve regeneration. The waste heat recovery efficiency is 72%, and the energy consumption per unit solvent recovery of the entire system is 0.75kWh / kg, lower than the preset requirement of 0.8kWh / kg, representing a 45% reduction compared to the original conventional distillation process on this production line. The system has been running continuously for 30 days, with stable equipment operation, no significant attenuation in sensor accuracy, and no replacement thresholds reached for the filter membrane, molecular sieve, and activated carbon. Maintenance costs are 32% lower than the original process.

[0110] III. Implementation Results Verification

[0111] This embodiment operated continuously for 30 days, processing a total of 3600L of waste solvent, recovering 2120L of qualified isopropanol and 1360L of NMP, with a total solvent recovery rate of 96.7%, far exceeding the industry average. The purity of the recovered solvent remained stable above 99.8%, with a moisture content ≤20ppm, fully meeting the reuse requirements of the 800G optical module packaging process. Calculations show that after the system was implemented, the production line's new solvent procurement volume decreased by 52%, saving approximately 850,000 RMB annually in procurement and hazardous waste disposal costs, with an investment payback period of approximately 7.2 months. Annual VOC emissions were reduced by approximately 2.8 tons, and annual standard coal savings were approximately 14.4 tons, demonstrating significant environmental and economic benefits. Simultaneously, the system operated continuously without major malfunctions, with timely online monitoring response and precise closed-loop control, fully adapting to the industrial requirements of 24-hour continuous operation of optical module mass production lines, verifying the practicality, stability, and advancement of the method of this invention.

[0112] The above embodiments are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any variations, substitutions, or parameter optimizations that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the protection scope of the present invention.

Claims

1. A method for the distillation recovery and online purity monitoring of organic solvents used in optical module packaging, characterized in that: Includes the following steps: S1 Waste Solvent Classification and Pretreatment Homogenization: Waste organic solvents generated in the optical module packaging process are classified and collected according to solvent type. After passing through two-stage precision filtration, molecular sieve adsorption dehydration, composite adsorption impurity removal, and vacuum degassing, they are sent to the homogenization feed tank to complete temperature and component homogenization, and obtain a qualified feed liquid with stable physical properties. S2 Multi-Tank Composite Distillation Separation: After preheating the qualified feed liquid using the system's waste heat, it is sent to a multi-tank composite distillation column. The segmented distillation process is adopted. By adjusting the temperature, pressure, and reflux ratio parameters of each distillation section, the light component impurities, target solvent, and high-boiling-point heavy components are precisely separated. The balance coefficient is introduced to control the endpoint of the total reflux operation, and the solvent is preliminarily purified. S3 membrane pervaporation coupled with deep purification: The pre-purified solvent is fed into the polyimide pervaporation membrane module to further remove trace amounts of water and organic impurities that cannot be separated by distillation, resulting in a high-purity recovered solvent. S4 Multi-Source Fusion Online Purity Monitoring: Multiple monitoring points are set up along the material flow direction. Each point is equipped with a four-source sensor array consisting of an ultraviolet-near-infrared broadband spectral probe, a high-precision online refractometer, a dual-frequency microwave dielectric constant sensor, and a high-frequency thermal conductivity trace moisture meter to simultaneously collect the full-dimensional physical property parameters of the solvent to be tested. S5 Multi-Source Data Fusion and Intelligent Purity Inversion: After time synchronization, temperature compensation and preprocessing of the collected multi-source data, a joint feature vector of spectrum and physical property is constructed, which is input into a pre-trained deep neural network (DNN) purity inversion model, and the quantitative parameters related to the purity of the solvent are output in real time. S6 Adaptive Dynamic Distillation Closed-Loop Control: Using real-time purity data as the core input, it adopts a Model Predictive Control (MPC) architecture and combines a preset purity grading standard for optical modules to dynamically adjust the operating parameters of the distillation system, thereby achieving adaptive optimization and closed-loop control of the distillation process. S7 Solvent Quality Grading and Production Line Closed-Loop Reuse: Based on real-time purity test results, recycled solvents are classified into grades. Different grades of solvents correspond to different rate-level optical module packaging processes. By connecting with the production line MES system, a full life cycle traceability system is established to achieve closed-loop reuse. The S8 system features self-maintenance, waste liquid treatment, and waste heat recovery: It ensures long-term stable operation of the system through dual-optical-path self-cleaning and sensor health assessment, performs harmless treatment of distillation waste liquid, and realizes energy recycling through a waste heat cascade recovery system.

2. The method for distillation recovery and online purity monitoring of organic solvents for optical module packaging according to claim 1, characterized in that: In step S1, the two-stage precision filtration uses a polypropylene filter membrane with a pore size of 5-10 μm and a polytetrafluoroethylene filter membrane with a pore size of 0.22-0.45 μm, respectively; the molecular sieve adsorption dehydration uses 3A / 4A molecular sieves with a filling amount of 10%-15% of the waste solvent volume, and an online dew point meter is set at the outlet to achieve closed-loop dehydration control; the composite adsorption depurification uses nitric acid-activated modified activated carbon combined with activated alumina as a depurifying agent, and after depurification, the depurifying agent particles are removed by a 0.22 μm precision filter; the vacuum degassing operation pressure is ≤5 kPa, and the homogenized feed tank controls the solvent temperature fluctuation to ≤±1℃.

3. The method for distillation recovery and online purity monitoring of organic solvents for optical module packaging according to claim 2, characterized in that: In step S2, the multi-tank composite distillation column consists of 3-5 distillation tanks connected in series, corresponding to independent distillation sections, with a total theoretical number of 20-30 plates and an operating pressure of 0.08-0.12 MPa. During the segmented distillation process, the first distillation section separates low-boiling-point impurities, the middle distillation section separates the target solvent, and the final distillation section enriches high-boiling-point heavy components. The equilibrium coefficient φ at the end of the total reflux operation is controlled at 0.8-0.

9. When the purity of the main component of the discharged solvent is ≥99.5%, it is sent to the subsequent deep purification process.

4. The method for distillation recovery and online purity monitoring of organic solvents for optical module packaging according to claim 3, characterized in that: In step S3, the polyimide pervaporation membrane module uses an organic solvent-resistant polyimide composite membrane. The feed temperature is controlled at 40-60℃, the feed pressure at 0.1-0.2MPa, the permeate side operating pressure at 0.01-0.03MPa, and the material residence time in the membrane module is 10-20 minutes. After purification, the solvent moisture content is reduced to below 10ppm, and the organic impurity content is reduced to below 0.05%.

5. The method for distillation recovery and online purity monitoring of organic solvents for optical module packaging according to claim 4, characterized in that: In step S4, 6-8 monitoring points are set along the material flow direction at the pretreatment outlet, the top discharge port of each distillation section, the outlet of the membrane pervaporation device, and the inlet of the finished product storage tank. Each monitoring point is connected to a closed-loop detection flow cell through a bypass sampling pipeline. A four-source sensor array is deployed corresponding to the flow cell. The ultraviolet-near-infrared broadband spectral probe adopts a dual-path design with a main measurement optical path and a self-cleaning reference optical path. All sensor signals are connected to the central processing unit through an explosion-proof isolation fence to achieve synchronous acquisition.

6. The method for distillation recovery and online purity monitoring of organic solvents for optical module packaging according to claim 5, characterized in that: In step S5, the IEEE 1588PTP precision time protocol is used to achieve microsecond-level timestamp alignment of multi-source data; after smoothing filtering, multivariate scattering correction, derivative preprocessing, and principal component analysis dimensionality reduction of the spectral data, it is concatenated with the standardized refractive index, dielectric constant, and moisture content data to form a joint feature vector; the DNN purity inversion model is pre-trained on a standard sample set prepared with commonly used organic solvents for optical modules, and outputs the comprehensive purity index, principal component content, moisture content, and impurity concentration parameters in real time. The model automatically completes a local calibration every 30 minutes.

7. The method for distillation recovery and online purity monitoring of organic solvents for optical module packaging according to claim 6, characterized in that: In step S6, the preset purity grading standard for optical modules includes three levels: Grade A high purity standard with main component content ≥99.9% and moisture content ≤10ppm, suitable for key processes of 800G / 1.6T optical modules; Grade B medium purity standard with main component content ≥99.8% and moisture content ≤20ppm, suitable for conventional processes of 400G / 800G optical modules; and Grade C basic standard with main component content ≥99.5% and moisture content ≤50ppm, suitable for auxiliary processes of 100G / 400G optical modules.

8. The method for distillation recovery and online purity monitoring of organic solvents for optical module packaging according to claim 7, characterized in that: In step S6, a discrete state-space model of the distillation process is established, with purity-related parameters and reflux tank level as control variables, and reflux ratio, heating power, outlet valve position, and feed rate as manipulated variables. Rolling optimization is performed with 18 steps in the prediction time domain and 6 steps in the control time domain. The control system executes the corresponding control strategy based on the matching results of real-time purity data and grading standards. At the same time, it uses cumulative sum CUSUM statistical process control to identify purity deviation trends and trigger parameter pre-adjustment.

9. The method for distillation recovery and online purity monitoring of organic solvents for optical module packaging according to claim 8, characterized in that: In step S7, different grades of recycled solvents are automatically diverted to corresponding stainless steel buffer tanks with polished inner walls and nitrogen-sealed micro-positive pressure protection; the solvent grade is intelligently matched with the packaging process by interfacing with the production line MES system through the OPCUA protocol; a unique traceability code is generated for each batch of recycled solvents, and the data of the entire process is synchronously uploaded to the production line digital twin platform to achieve full life cycle traceability.

10. The method for distillation recovery and online purity monitoring of organic solvents for optical module packaging according to claim 9, characterized in that: In step S8, the online monitoring probe automatically starts ultrasonic cleaning and nitrogen purging every 2 hours; the distillation waste liquid is sent to an 800-1000℃ incinerator for incineration and decomposition after neutralization and pretreatment, and the flue gas is treated to meet emission standards; the waste heat of the distillation system is recovered through a plate heat exchanger with a waste heat recovery efficiency of ≥70%, and the recovered waste heat is used for feed preheating, molecular sieve regeneration, and other processes.

Images