| 摘要: | 目前國內PCB廠商在檢測與清除PCB製程殘膠上，仍是採取分站點的方式執行，執行時間較長且耗費人力，且使用的光學檢測與雷射清潔機台皆是由國外進口，成本較高且不易針對需求進行調整。本研究在此背景下，持續精進本團隊所研發的自動化光學檢測與雷射清除複合機(AOIR)。該機台整合光學檢測、雷射清除、精密平台移動等技術，並將自動化流程分為粗定位與精定位兩大部分。粗定位使用線性相機進行大範圍PCB拍攝並應用EmguCv函式庫透過影像處理的方式來定位殘膠，精定位則是將被定位標註的殘膠依序移至同軸雷射成像系統進行高精度的輪廓辨識並規劃雷射掃描路徑予以雷射掃描清除。
本研究大幅改善雷射清除效果並探討雷射清除機制與優化機台運作效率。雷射清除部分，使用紅外線皮秒光纖雷射作為雷射源，為避免雷射掃描傷及銅面，選定峰值能量較低的脈衝雷射參數進行測試，測試結果顯示使用能量密度為1.048 J/cm2的雷射參數透過重複掃描3次的方式，可以完全清除高度25 μm的殘膠，並經雷射共軛焦顯微鏡量測表面輪廓與粗糙度確認銅面沒有受到損傷。在雷射清除機制方面，使用一維熱擴散方程式估算材料接收脈衝雷射能量的表面溫度上升值，得到經單發脈衝雷射照射後，殘膠表面溫度會上升至大幅超過殘膠的沸點，故可判斷殘膠的清除機制為汽化燒蝕。
運作效率部分，本研究研發一種更為泛用及較快的影像處理流程，透過線路遮罩的方式來指定檢測範圍避免誤判，並加強對於在銅線路邊緣的殘膠的檢測能力。優化程式的運算，大幅改善影像處理時間，粗定位由13.7秒縮短至4.8秒，精定位則由1.2秒縮短至0.17秒，且精定位的殘膠輪廓辨識成功率為92.7%。採用直接針對疑似殘膠進行3次重複掃描而不再重新辨識的策略，可將單點除膠時間由8秒縮短至3.3秒。
最後進行全自動化PCB檢測與清除一貫作業驗證測試，在測試前後皆使用雷射共軛焦與掃描式電子顯微鏡進行表面觀察與元素組成分析，確認真實殘膠可被徹底清除，且銅面結構沒有產生變化，達成將光學檢測與雷射清除整合至單一機台並自動化運行的目標，並與PCB廠商目前的作業時間進行比較，產出效能可提升接近10倍。因此，本研究所研發之該整合機台已具備商業化之應用價值。
;In the domestic PCB manufacturing industry, the detection and removal of photoresist (PR) residues generated during the PCB process currently are still carried out in a station-by-station approach, resulting in a longer processing time and higher costs. Moreover, the optical inspection machine and laser cleaning machine used are imported from foreign countries, which leads to higher costs and limited flexibility for specific requirements. This study continued to improve an automated optical inspection and repair (AOIR) machine developed by our research team. This machine integrated optical inspection, laser cleaning, and precision stage movement technologies. The automated process was divided into two main parts, namely coarse positioning and precise positioning. Coarse positioning involved capturing wide-range PCB images using a line scan camera and utilizing the EmguCv library for image processing to locate PR residues. Precise positioning involved sequentially transferring these labeled PR residues to a coaxial laser imaging system for high-precision contour identification and planning of laser scanning paths for effective removal via laser cleaning.
This study was focused on improving the effectiveness of laser removing and exploring the mechanism of laser cleaning, as well as optimizing the process efficiency of the machine. In the laser cleaning process, an infrared picosecond fiber laser was used as the laser source. To prevent laser scanning from damaging the copper surface, pulse laser parameters with lower peak energy were selected for testing. The test results demonstrated that using laser parameters with a fluence (energy density) of 1.048 J/cm2 and performing three repeated scans could completely remove intact PR residue with a height of 25 μm. Laser scanning confocal microscopy (LSCM) was utilized to measure the surface profile and roughness, confirming that the copper surface was not damaged after laser scanning. For the laser cleaning mechanism, the surface temperature rise of the material upon receiving pulsed laser energy was estimated using a one-dimensional heat diffusion equation. The surface temperature of the PR residue increased significantly beyond its boiling point after a single pulse irradiation, indicating the removal mechanism of the PR residue was vaporization ablation.
In terms of process efficiency, this study developed a more versatile and faster image processing procedure. It employed a circuit mask to specify the detection area, thus avoiding false detections, and enhancing the detection capability for PR residues at the edge of PCB circuit trace. The optimization of program computation significantly improved the image processing time, reducing the coarse positioning from 13.7 s to 4.8 s and the precise positioning from 1.2 s to 0.17 s. The successful rate of PR residue contour identification in precise positioning reached 92.7%. Furthermore, by adopting a strategy of performing three repeated scans directly on suspected PR residues without re-identification, the single residue cleaning time was reduced from 8 s to 3.3 s.
Finally, a test was conducted to assess the automated PCB inspection and cleaning procedure. LSCM and scanning electron microscopy (SEM) were employed before and after the test for surface observation and elemental composition analysis. The results confirmed that real PR residues could be completely removed without changing the copper surface structure. The goal of integrating optical inspection and laser cleaning into a single machine for automated operation was accomplished. A comparison was made with the current process time of a local PCB manufacturer, showing a nearly 10 times improvement in performance. Accordingly, such automated optical inspection and laser cleaning machine is readily available for commercialization. |
