Chunhui Chung Research Fellow Department of Mechanical Engineering, National Chiao Tung University, Hsinchu, Taiwan 30010 Chad S. Korach Assistant Professor Imin Kao 1 Professor e-mail: kao@mal.eng.sunysb.edu Department of Mechanical Engineering, Stony Brook University, Stony Brook, NY 11794-2300 Experimental Study and Modeling of Lapping Using Abrasive Grits with Mixed Sizes In this paper, the lapping process of wafer surfaces is studied with experiments and con- tact modeling of surface roughness. In order to improve the performance of the lapping processes, effects of mixed abrasive grits in the slurry of the free abrasive machining (FAM) process are studied using a single-sided wafer-lapping machine. Under the same slurry density, a parametric experimental study employing different mixing ratios of large and small abrasive grits and various normal loadings on the wafer surface applied through a jig is conducted. Observations and measurements of the total amount of mate- rial removed, material removal rate, surface roughness, and relative angular velocity are presented as a function of various mixing ratios and loadings and discussed in the paper. The experiments show that the 1:1 mixing ratio of abrasives removes more material than other mixing ratios under the same conditions, with a slightly higher surface roughness. Modeling of the mixed abrasive particle distributions correspondingly indicates that the roughness trend is due to the abrasive size distribution and the particle contact mechan- ics. The results of this study can provide a good reference to the FAM processes that practitioners use today by exploiting different abrasive mixing ratios in slurry and nor- mal loadings in the manufacturing processes. [DOI: 10.1115/1.4004137] 1 Introduction Wafers made of materials such as silicon, III-V and II-VI com- pounds, and optoelectronic materials, require a high-degree of sur- face quality in order to increase the yield in micro-electronics fabri- cation to produce integrated circuits and devices. Due to the reduction of feature size in micro-electronics fabrication, the requirements of the wafer surface qualities, such as the commonly defined site flatness, nanotopography, total thickness variation, and warp [1], become more and more stringent. To meet such require- ments, the wafer manufacturing processes of brittle semiconductor materials, including slicing, lapping, grinding, and polishing have been continually improved. Following Moore’s law, the interna- tional technology roadmap of semiconductors indicates that the 450 mm wafer will be in production in 2012 [2] to keep the trend of cost reduction. Many analyses and discussions have started to focus on the next generation wafer size [3–8]. With the agreement of Intel, Samsung Electronics, and Taiwanese Semiconductor Manu- facturing Company Limited (TSMC) at the 450 mm wafer manu- facturing transition [9], the next increase of wafer size is inevitable. With such increase, it is more difficult to achieve the requirements of wafer surface quality. Therefore, the advance of the machining processes such as wiresawing, lapping, and grinding is important. Lapping has been a standard surface finishing process for glass products and semiconductor wafers for a long time. Lapping, by virtue of using third-body free abrasive for removing materials from substrate surface, belongs to the category of the free abrasive machining process, which is the same as slurry wiresaw slicing [10–12]. Although most research attributes the brittle material re- moval of lapping to indentation cracking models [13–17], the actual mechanism is more complicated [18–20]. Aside from the mechanical properties of the workpiece and lapping plate, the dis- tribution of abrasives, dynamic indentation cracking, motion of the abrasive grits and the ductile-regime machining [21] also com- plicate the analysis of the lapping mechanism. Lapping and grinding are both postslicing wafer surface finish- ing processes. Because of their advantages and disadvantages, one or both of them are utilized in the manufacturing process [22,23]. It is not clear which one will be favorable or employed in the 450 mm wafer industry. However, lapping is capable of removing warp efficiently until the invention of simultaneous double-sided grinding [23,24]. In this paper, an experimental study and model- ing of the surface roughness generated from lapping with a mixed abrasive slurry provides information on the influence of abrasive distributions in lapping. Past research has emphasized the importance of abrasive size dis- tribution in modeling; however, few have studied the change of the distribution of abrasive grit sizes. Bhagavat, et al. [1] is probably the first and the only one to study such topic. Their results showed that the mixed abrasives (for example, mixing F-400 and F-600 SiC) have higher material removal rate than the single-sized abra- sives (for example, only F-400 SiC abrasives). However, their experiments discussed one mixing ratio of the abrasives, and the concentration of mixed abrasive slurries were different from that of the single-sized abrasives slurry. To study the influence of the change of abrasive distribution in lapping, experiments with differ- ent abrasive distributions and constant slurry concentration are necessary. In this study, experiments were conducted by mixing two dif- ferent sizes of SiC powders: F-400 and F-600. Five different mix- ing ratios of the abrasives were employed, with the ratio of the total mass of abrasives to the volume of carrier fluid (de-ionized (DI) water) being kept the same. The results show that the 50% mixing ratio (1:1) of the two different abrasives have the highest material removal rate (MRR), with a slightly higher surface roughness. In addition, the material removal rate is nearly propor- tional to the normal loading. The surface roughness, however, depends on the distribution of mixed abrasive grits but not the total loading. This is comparable to results presented in the litera- ture [14,15]. A model of surface roughness based on particle con- tact depth was utilized to compare the effects of different mixing ratios. The roughness and penetration depth in lapping have been modeled by the Hertzian formulation [25,26], as well as the statis- tical nature of the contact using probability functions [17,18,27]. The particle distribution nonuniformity (standard deviation) has 1 Corresponding author. Contributed by the Manufacturing Engineering Division of ASME for publication in the JOURNAL OF MANUFACTURING SCIENCE AND ENGINEERING. Manuscript received May 1, 2010; final manuscript received April 25, 2011; published online June 8, 2011. Assoc. Editor: Prof. Shreyes N. Melkote. Journal of Manufacturing Science and Engineering JUNE 2011, Vol. 133 / 031006-1 Copyright VC 2011 by ASME Downloaded From: http://manufacturingscience.asmedigitalcollection.asme.org/ on 04/24/2014 Terms of Use: http://asme.org/terms