(Second Best Student Paper Award) Blade Test in Atmospheric-Pressure Ar Gas to Characterize Bonded Interface Fabricated Using Atomic Diffusion Bonding

Abstract

Atomic diffusion bonding (ADB) of wafers using thin films is a promising process to achieve room-temperature wafer bonding [1]. For ADB processing, thin oxide [2] and nitride [3] films are useful for bonding, as are thin metal films [1]. For blade testing, a conventional method to examine the surface free energy at the bonded interface γ, the debonding length L is measured after a thin blade is inserted between the bonded wafers. The value of γ is calculated using L values. For a bonded interface such as Si/SiO2 or SiO2/SiO2, stress corrosion caused by oxygen and water gases in air is well-known to affect the L value as does the bonding strength [4]. For this study, a blade test was applied in Ar gas at atmospheric pressure for wafers bonded using ADB processing. Bonded interfaces of two kinds were examined: interfaces bonded using thin metal films and those bonded using oxide films. Using Maszara’s equation for convenience, we assessed the variation of γ values calculated from L values.Figure 1 presents values of γ for quartz glass wafers bonded using Ti(0.7 nm) film on each side as a function of measurement time t. Two series of data are shown: those measured in Ar and in air. The blade-insertion speed was fixed at 100 μm/s. As t increased, γ measured in air decreased remarkably from the initial value γ0(air) of 3.1 J/m2. The values of γ measured in Ar were higher than those in air. γ in Ar maintained the initial value of γ0(Ar) until t=400 s; it then started decreasing. In fact, γ0(Ar) did not increase concomitantly with increasing blade-insertion speed. Results revealed that the gradual reduction of γ in Ar resulted from oxidation of Ti films with impurities of O2 and H2O in Ar gas. The value obtained without oxidation effects is γ0(Ar).Figure 2 presents values of γ for quartz glass wafers bonded using oxide thin films as a function of t: (A) using Al2O3(2 nm) films and (B) using ZrO2(0.5 nm) films. In both figures, the values of γ measured in Ar are much higher than those measured in air. Actually, γ in Ar for wafers bonded using Al2O3(2 nm) films maintained the initial value γ0(Ar) until t=150 s; then it started decreasing. This suggests that γ0(Ar) corresponds to the value obtained without the effects of water-stress corrosion. However, γ in Ar for wafers bonded using ZrO2(0.5 nm) films did not maintain the initial value γ0(Ar): γ decreased gradually as t increased, which indicates that the water–stress corrosion effects on the values of γ were not suppressed even in Ar. T. Shimatsu and M. Uomoto, ECS Trans., 33 (4), 61 (2010).T. Shimatsu, H. Yoshida, M. Uomoto, T. Saito, T. Moriwaki, N. Kato, Y. Miyamoto, and K. Miyamoto, Proceedings of Seventh LTB-3D 2021. 51 (2021).M. Uomoto, H. Yoshida, T. Shimatsu, T. Saito, T. Moriwaki, N. Kato, Y. Miyamoto, and K. Miyamoto, Proceedings of Seventh LTB-3D 2021. 45 (2021).F. Fournel, L. Continni, C. Morales, J. Da Fonseca, H. Moriceau, F. Rieutord, A. Barthelemy, and I. Radu, J. Appl. Phys., 111, 104907 (2012). Figure 1