节点文献
金刚石线锯切割半导体陶瓷的机理与工艺研究
Research on Slicing Mechanism and Technology with Fixed Diamond Wire Saw for Semiconductor Ceramics
【作者】 赵礼刚;
【作者基本信息】 南京航空航天大学 , 机械制造及其自动化, 2010, 博士
【摘要】 随着半导体工业的飞速发展,单晶硅、砷化镓和碳化硅等半导体陶瓷材料在此领域得到广泛应用。现代微处理器和其他逻辑芯片的制备要求不断提高,一方面,为了增大芯片产量,降低单元制造成本,要求半导体材料切片的直径不断增大;另一方面,为了提高集成电路(Integrated Circuit,简称IC)的集成度,要求半导体材料切片的刻线宽度越来越细,这对单晶硅和砷化镓等半导体陶瓷材料的切片技术提出了新的要求。切片是制造IC的一道重要工序。为解决大直径晶体,如单晶硅和砷化镓等的切片问题,本文研究了往复式金刚石线锯切片技术。该技术具有切缝窄、锯切效率高、切片质量好、对环境污染小、能加工大直径工件和超硬材料等优点。通过对这一切片技术进行深入系统的理论分析和试验研究,揭示了加工工艺参数与加工质量之间的关系,对工艺参数进行了正交优化,获得了最佳工艺参数,为半导体陶瓷材料的高质量、高效切片提供了科学依据和技术支撑。本文完成的主要工作和取得的成果如下:(1)对金刚石线锯切割机理进行了研究建立了金刚石线锯切割模型,研究了金刚石线锯切割的材料去除过程,通过压痕实验研究了单晶硅在室温状态下硬度和裂纹的产生、扩展及特征,获得了脆塑转变的临界压头压入深度值d c,结合能量脆塑转变理论获得了金刚石线锯切割过程的脆塑转变的临界切深公式。(2)对金刚石线锯切割过程中锯丝振动、断裂进行了研究,分析了锯丝振动、断裂与工艺参数之间的关系,为确定试验工艺参数范围提供了判定依据建立了锯丝振动模型,研究了工艺参数(锯丝线速度,工件进给速度和锯丝张紧力)和锯丝振动之间的关系;分析了锯丝断裂的原因,研究了工艺参数与锯丝断裂之间的关系,获得了锯丝防断裂的条件;根据锯丝振动、断裂与工艺参数之间的关系,为确定试验工艺参数范围提供了判定依据。(3)对金刚石线锯切割单晶硅和砷化镓的工艺进行了研究以锯丝线速度、工件进给速度和锯丝张紧力为考虑因素,以砷化镓和单晶硅为切割材料,设计并进行了一系列单因素试验,研究了加工工艺参数对砷化镓和单晶硅切片表面质量的影响,获得了加工工艺参数对表面质量的影响规律;运用Taguchi法对加工工艺参数进行了正交优化,对正交优化的结果进行了分析处理,获得了最佳工艺参数;运用离差分析法,分析了加工工艺参数对目标质量特征参数的影响,获得了加工工艺参数对表面质量的影响程度,对加工工艺参数选择具有指导作用。(4)对使用不同切削液的加工表面质量进行了研究将加速剂引入金刚石线锯切片加工中,对比分析了使用普通切削液(水)与加速剂后切片的表面质量,试验证明,加速剂的引入大大提高了切片的表面质量;通过对不同配比的加速剂进行金刚石线锯切片试验研究,获得了加速剂溶液的最佳配比参数,为加速剂在金刚石线锯切片加工中的使用提供了技术支持;结合加速剂溶液的物理化学特性以及金刚石线锯切片实验结果,分析了加速剂的作用。(5)对金刚石线锯切割过程以及切片表面三维形貌进行了仿真研究对单颗磨粒的切割进行了仿真研究,模拟了单颗磨粒的切割过程,分析了切割表面的形貌以及裂纹和损伤。通过研究锯丝磨粒的分布,建立了锯丝模型,对切片表面三维形貌进行了仿真研究,获得了不同加工工艺参数下表面粗糙度的仿真值以及加工参数对表面粗糙度的影响规律,其规律与试验所获得规律一致。(6)对金刚石线锯切割表面损伤层进行了研究采用X射线衍射仪以及逐层腐蚀法对砷化镓和单晶硅切片表面进行损伤层检测,获得了不同加工工艺参数下的表面损伤层厚度,研究了加工工艺参数对切片的表面损伤层厚度的影响,并获得了加工工艺参数对切片表面损伤层厚度的影响规律。将已测得的不同加工工艺参数下的表面损伤层厚度为样本,对神经网络进行训练学习,建立了BP神经网络的预测模型,实现了对表面损伤层厚度的预测。
【Abstract】 With the development of semiconductor industry, semiconductor ceramic material such as single silicon, gallium arsenide and silicon carbide, etc, is extensive used in this field. On one hand, for increasing the yield of chips and decreasing the cost of every unit, the chips’diameter increases continuously; on the other hand, the reticule width becomes thinner to increase the integration level of Integrated Circuit, which meet the requirements of modern chips and other logic chips. These new situations put forward new requirements to the manufacturing technology of single silicon. The slicing is an important process for manufacturing Integrated Circuit. This paper researched the slicing technology of fixed abrasive wire, which has the advantages including narrow cutting kerfs, high cutting efficiency, good slicing quality, little environmental pollution, processing ability for large diameter work piece and super hard material. The relationship between processing parameters and processing quality was revealed by deep and systematic experimental research and theoretical analysis for this slicing technology, the measures to obtain high process quality was put forward, which provided scientific basis and technical support to semiconductor ceramic material’s high quality and efficient slicing. The main work is as follows:(1)Slicing mechanism of fixed diamond wire saw was studied.The hardness of single silicon and the cracks’ generation, extension and characteristics at normal temperature was studied by Vickers hardness tester. The slicing mechanism was researched binding experiment results.(2)The relationships between the vibration, fracture of the wire saw and the processparameters were researched, and the process parameters’scope was defined. The effect factors of wire sawing process were analyzed, the wire saw’s vibration model was built and the relationship between process parameters, including linear velocity of wire saw, the work piece feed rate and the tension force, and vibration of wire saw was studied. The fracture of wire saw, such as formation cause, the relationship with process parameters and the conditions to form was researched. All of the above initially defined the process parameters’ scope for the experimental study.(3)The process for slicing single silicon and gallium arsenide by fixed diamond wire saw was studied.A single factor experiment for slicing single silicon and gallium arsenide was designed and carried on, which used linear velocity of wire saw, the work piece feed rate and the tension force as the factors. The effects of the process parameters to the surface quality of single silicon and gallium arsenide was researched based on this experiment. An orthogonal test was designed by Taguchi method, through which an orthogonal optimum for process parameters was made. Influence of characteristic parameters of surface quality and process parameters was analyzed, based on which the best process parameters’ scope was defined.(4)The working mechanism of cutting fluid was revealed by analyzing and contrasting different cutting fluid.The accelerator was introduced into the cutting process. The cutting surface quality under different situations such as the water and the accelerator was compared. The evidence showed that the accelerator can improve the surface quality greatly than the water. The ratio of the accelerator was researched, and the best ratio was got.(5)The simulation study of the slicing process and 3-D surface topography was carried on.Simulation study of single abrasive slicing single silicon was made, which simulated the slicing process of single abrasive, chip shape and crack damages on the surface of single silicon was analyzed. The wire saw model was built though the study of saw abrasive’s distribution; the simulation study of 3D topography of the slicing surface was carried on, from which the simulation value of surface roughness could be obtained.(6)The damage layer of slicing was researched.The damage layer of slicing surface for the single silicon and gallium arsenide was detected by X-ray diffractometer, by which thickness of surface damage layer under different process parameters was obtained and the influence between process parameters and the thickness of surface damage layer was studied; the thickness of surface damage layer was predicted by neuron algorithm.
【Key words】 Fixed abrasive; Reciprocating diamond wire saw; Silicon; GaAs; Surface roughness; 3-D surface topography; Damage layer; Neural net;