Purpose

Lapping is a vital flattening process to improve the quality of processed semiconductor wafers such as single-crystal sapphire wafers. This study aims to optimise the lapping process of the fixed-abrasive lapping plate of sapphire wafers with good overall performance [i.e. high material removal rate (MRR), small surface roughness (Ra) of the wafers after lapping and small lapping plate wear ratio (η)].

Design/methodology/approach

The influence of process parameters such as lapping time, abrasive size, abrasive concentration, lapping pressure and lapping speed on MRR, Ra and η of lapping-processed sapphire wafers was studied, and the results were combined with experimental data to establish a regression model. The multi-evaluation index optimisation problem was transformed into a single-index optimisation problem via an entropy method and the grey relational analysis (GRA) to comprehensively evaluate the performance of each parameter.

Findings

The results revealed that lapping time, abrasive size, abrasive concentration, lapping pressure and lapping speed had different influence degrees on MRR, Ra and η. Among these parameters, lapping time, lapping speed and abrasive size had the most significant effects on MRR, Ra and η, and the established regression equations predicted the response values of MRR, Ra and η to be 99.56%, 99.51% and 93.88% and the relative errors between the predicted and actual measured values were <12%, respectively. With increased lapping time, MRR, Ra and η gradually decreased. With increased abrasive size, MRR increased nearly linearly, whereas Ra and η initially decreased but subsequently increased. With an increase in abrasive concentration, MRR, Ra and η initially increased but subsequently decreased. With increased lapping pressure, MRR and η increased nearly linearly and continuously, whereas Ra decreased nearly linearly and continuously. With increased lapping speed, Ra initially decreased sharply but subsequently increased gradually, whereas η initially increased sharply but subsequently decreased gradually; however, the change in MRR was not significant. Comparing the optimised results obtained via the analysis of influence law, the parameters optimised via the entropy method and GRA were used to obtain sapphire wafers lapping with an MRR of 4.26 µm/min, Ra of 0.141 µm and η of 25.08, and the lapping effect was significantly improved.

Originality/value

Therefore, GRA can provide new ideas for ultra-precision processing and process optimisation of semiconductor materials such as sapphire wafers.

中文翻译：

---

蓝宝石晶片固定磨具研磨盘的表面质量预测与研磨工艺优化

目的

研磨是提高加工半导体晶片（如单晶蓝宝石晶片）质量的重要压平工艺。本研究旨在优化蓝宝石晶片固定磨料研磨板的研磨工艺，具有良好的综合性能[即材料去除率（MRR）高，研磨后晶片的表面粗糙度（Ra）小，研磨板磨损率小（ η)]。

设计/方法/方法

研究了研磨时间、磨料尺寸、磨料浓度、研磨压力和研磨速度等工艺参数对研磨加工蓝宝石晶片MRR、Ra和η的影响，并结合实验数据建立回归模型。通过熵法和灰色关联分析（GRA）将多评价指标优化问题转化为单指标优化问题，综合评价各参数的性能。

发现

结果表明，研磨时间、磨料粒度、磨料浓度、研磨压力和研磨速度对MRR、Ra和η有不同程度的影响。在这些参数中，研磨时间、研磨速度和磨料尺寸对MRR、Ra和η的影响最显着，建立的回归方程预测MRR、Ra和η的响应值为99.56%、99.51%和93.88%，预测值和实际测量值之间的相对误差分别<12%。随着研磨时间的增加，MRR、Ra和η逐渐减小。随着磨料尺寸的增加，MRR 几乎呈线性增加，而 Ra 和 η 最初减少但随后增加。随着磨料浓度的增加，MRR、Ra 和 η 最初增加但随后减小。随着研磨压力的增加，MRR 和 η 几乎呈线性连续增加，而 Ra 几乎呈线性连续下降。随着研磨速度的增加，Ra先急剧下降后逐渐上升，η先急剧上升后逐渐下降；然而，MRR 的变化并不显着。对比影响规律分析得到的优化结果，利用熵法和GRA优化的参数，得到蓝宝石晶片的研磨，MRR为4.26 µm/min，Ra为0.141 µm，η为25.08，研磨效果得到显着改善。而 η 最初急剧增加，但随后逐渐减小；然而，MRR 的变化并不显着。对比影响规律分析得到的优化结果，利用熵法和GRA优化的参数，得到蓝宝石晶片的研磨，MRR为4.26 µm/min，Ra为0.141 µm，η为25.08，研磨效果得到显着改善。而 η 最初急剧增加，但随后逐渐减小；然而，MRR 的变化并不显着。对比影响规律分析得到的优化结果，利用熵法和GRA优化的参数，得到蓝宝石晶片的研磨，MRR为4.26 µm/min，Ra为0.141 µm，η为25.08，研磨效果得到显着改善。

原创性/价值

因此，GRA可以为蓝宝石晶圆等半导体材料的超精密加工和工艺优化提供新思路。