J Fail. Anal and Preven.(2008)8: 524-532 m Fig 8 FIB microstructure(a) Etching section. (b)and(c) Morphologies of the cracking Table 1 Conditions of thermal cycle testing on the bare board segregated to the surface along the different layers of the Range of cycle temperature, C 40125 vias probably nucleated as a set of three-dimensional islands of Cu s, which would be expected to grow until 13(up)+ 20(remaining) Cool-down cycle, (down)+20 (remaining) or grain-boundary intersections with the surface [15].This Times for thermal cycle is highly undesirable, and there are several structura transformations of Curs in the temperature interval of eDs is far above the standard. as is shown in the former interest. These transformations favor the fatal crack part. Thus, it can be assumed that this embrittlement nucleation and propagation at grain boundaries under the fracture is largely affected by segregation of S at grain thermal stress generated during reflow soldering. Conse quently, the conclusion can be reached that the residue of s Although the structure formed by the segregated S at impurity along copper grain boundary has a significant grain boundaries is still unknown, the formation of Cu,s infiuence on the reliability of the blind vias, which is the ompounds can be expected, especially Cu2S. Similar predominant cause of the failure. The residue S primarily evidence was also found by Boulliard and Sotto [13] that results from the desmear process and the cleaning of the the structure of segregated sulfur on the face of Cu(100) blind vias. single crystals was 8 S atoms on 17 Cu atoms, with the s Moreover, through macroscopic observation, apparent atoms arranged in a p(2 x 2)lattice. Note that this cov- warpage can be noticed in type 2 PCB with a discrepancy erage approximates the stoichiometry of Cu2S phase of about 1 mm between the right and left sides of the which is highly stable [14]. It was also reported that the board. As mentioned previously, the type 2 PCB is only phase transition from the "high chalcocite"hexagonal 0.98 mm thick, the thickness of which is at least 10% phase to the complex"low chalcocite"phase occurs at thinner than that of type I PCB. It is both the improper 1035C for x= 2.000, but drops to 90oC for the slightly reduction of the thickness and asymmetric configuration of more sulfur-rich compound with x =1990 components on the board that lead to the excessive war- Hence, the mechanism of the cracking generation in the page and deformation of the board, which will produce a blind vias occurs through the following steps. The sulfur big tensile force or a bending stress around pads linked 2 SpringerEDS is far above the standard, as is shown in the former part. Thus, it can be assumed that this embrittlement fracture is largely affected by segregation of S at grain boundary. Although the structure formed by the segregated S at grain boundaries is still unknown, the formation of CuxS compounds can be expected, especially Cu2S. Similar evidence was also found by Boulliard and Sotto [13] that the structure of segregated sulfur on the face of Cu (100) single crystals was 8 S atoms on 17 Cu atoms, with the S atoms arranged in a p(2 9 2) lattice. Note that this cov￾erage approximates the stoichiometry of Cu2S phase, which is highly stable [14]. It was also reported that the phase transition from the ‘‘high chalcocite’’ hexagonal phase to the complex ‘‘low chalcocite’’ phase occurs at 103.5 C for x = 2.000, but drops to 90 C for the slightly more sulfur-rich compound with x = 1.990. Hence, the mechanism of the cracking generation in the blind vias occurs through the following steps. The sulfur segregated to the surface along the different layers of the vias probably nucleated as a set of three-dimensional islands of CuxS, which would be expected to grow until they coalesced. The nucleation sites may be at dislocation or grain-boundary intersections with the surface [15]. This is highly undesirable, and there are several structural transformations of CuxS in the temperature interval of interest. These transformations favor the fatal crack nucleation and propagation at grain boundaries under the thermal stress generated during reflow soldering. Conse￾quently, the conclusion can be reached that the residue of S impurity along copper grain boundary has a significant influence on the reliability of the blind vias, which is the predominant cause of the failure. The residue S primarily results from the desmear process and the cleaning of the blind vias. Moreover, through macroscopic observation, apparent warpage can be noticed in type 2 PCB with a discrepancy of about 1 mm between the right and left sides of the board. As mentioned previously, the type 2 PCB is only 0.98 mm thick, the thickness of which is at least 10% thinner than that of type 1 PCB. It is both the improper reduction of the thickness and asymmetric configuration of components on the board that lead to the excessive war￾page and deformation of the board, which will produce a big tensile force or a bending stress around pads linked Fig. 8 FIB microstructure. (a) Etching section. (b) and (c) Morphologies of the cracking Table 1 Conditions of thermal cycle testing on the bare board Range of cycle temperature, C -40–125 Heat-up cycle, min 13 (up) ? 20 (remaining) Cool-down cycle, min 17 (down) ? 20 (remaining) Times for thermal cycle 200 530 J Fail. Anal. and Preven. (2008) 8:524–532 123