Measurement of Work of Adhesion on Wafers for Direct Bonding
- PDF / 2,419,419 Bytes
- 6 Pages / 612 x 792 pts (letter) Page_size
- 16 Downloads / 277 Views
A12.4.1
MEASUREMENT OF WORK OF ADHESION ON WAFERS FOR DIRECT BONDING Kevin T. Turner and S. Mark Spearing Massachusetts Institute of Technology, Cambridge, MA 02139 USA ABSTRACT The displacement loaded double cantilever beam (DCB), often referred to as the bladeinsertion test or crack-opening method by the wafer bonding community, has become a common method for evaluating the work of adhesion of bonded wafer pairs. The test, while easy to perform, often yields results with large scatter and questionable accuracy. The mechanics of the specimen are investigated in detail in the current work. Expressions that demonstrate how wafer bow may lead to residual stresses that result in large errors in the calculated work of adhesion are developed. A three-dimensional finite element model is used to show that due to the circular wafer geometry and silicon anisotropy there is a large variation of the strain energy release rate across a straight crack front. The model is used to predict the actual crack front shape and shows good agreement with experiments. The results of the finite element simulations are compared to the traditional expression used for data reduction and implications of the model highlighted. INTRODUCTION Direct wafer bonding has emerged as an important technology in the manufacture of semiconductor substrates and advanced microsystems. Successful bonding requires wafers that are sufficiently flat and smooth as well as have a high work of adhesion [1]. Various efforts are underway to develop alternative cleaning and activation techniques to achieve a high work of adhesion. As part of this process development, a robust and accurate method is required to measure the work of adhesion. Traditionally, this has been accomplished through the use of a displacement loaded double cantilever beam (DCB) geometry as shown in Fig. 1. A razor blade of specified thickness is inserted at the bonded interface and the work of adhesion is calculated from the resulting crack length, which is typically measured from above, Fig. 1(b). The specimen was first introduced to the wafer bonding community by Maszara et al. [2] and has been widely adopted because of its ease of use and simple data reduction. While the specimen has seen widespread use, there have been persistent questions with regard to its accuracy [3,4,5]. In the current work, the mechanics of the specimen are first reviewed, then the effect of residual stresses that result from wafer bow are examined. Next, finite element modeling is used to assess the variation of the strain energy release rate across the crack front and to predict the shape of the curved crack front that results due to the circular wafer shape and silicon anisotropy. The finite element results are compared to experiments and discussed with regard to implications on using this test specimen. SPECIMEN MECHANICS The basic geometry of the specimen is shown in Fig. 1(a), it consists of two layers of thickness h1 and h2 with a blade of thickness δ inserted at the bonded interface. In examining the mechanics o
Data Loading...
