Structural characterization of ultra low-k carbon doped silicon dioxide films by X-ray Reflectometry and GISAXS

Full text: For scaling the future microelectronic devices, it was identified that one key factor was the substitution of both the high-k dielectrics employed as gates of the modern transistors [1] and the low -k dielectric used for interconnections in modern microelectronic circuits [2, 3]. One example of the advanced low-k dielectrics are carbon-doped silicon dioxide (SiCOH) in which voids was included to achieve k < 2.0. However, due to problems associated to materials infiltration into the pores during the CMOS process, which promotes the emergence of a current leakage through the dielectric. In order to overcome this difficulty, it is needing to find procedure which allows the partial sealing of the porous structure, but avoiding the loss of its dielectric properties. In this sense, we deposited 180 nm of SiCOH film upon silicon substrate by chemical vapor deposition (CVD) method and then they were submitted to plasma treatment composed by a mixture of Ar/N2 and Ar/H2 at 100 W and 200 W, respectively. The thickness and refractive index of them were estimated by ellipsometry, while XRR and GISAXS analysis were employed for computing the mass density, porosity, pore shape and size, and also the thickness as plasma treatment function. These experiments was carried out by using X-ray beam with wavelength = 0.1549 nm. For GISAXS, the X-ray incident angles were 0.15° and 0.25° that were chosen in order to maximize the scattering intensity. The results show that the total thickness and critical angle (refractive index) of the films are 176.8 nm and 0.120° (1.254), for pristine sample; 166.5 nm and 0.143° (1.263) for the film treated by Ar/N2, whereas for the one treated by Ar/H2 these values were 171.1 nm and 0.136° (1.258), respectively. The fitting procedure of both XRR and GISAX spectrum shows that plasma treatment produces pore sealing of the films superficial region, so that the total thickness is composed by two layers, being the superficial one thinner with higher mass density. For the sample treated by Ar/N2-plasma the entire thickness (mass density) was T = 6.6 nm + 159.9 nm ( = 0.99 g/cm³ + 0.75 g/cm³), while for the Ar/H2-plasma treated sample it was T = 34.2 nm + 137.0 nm ( = 0.97 g/cm³ + 0.73 g/cm³). The SiCOH porous structure can be modeled as composed by three noninteracting types of pores: two spheres with rougher interfaces (radius 1.01 to 1.28 nm) and one cylinder (radius 1.28-1.33 nm and length about 39 nm) which varies slightly after plasma exposition. The porosity was found to be equal to 47% for the pristine sample, while for the samples treated by Ar/N2- and Ar/H2-plasma, it was about 42 and 39%, respectively. In conclusion, pores of the superficial region of SiCOH films were successful sealed by plasma treatment forming a thin layer which is dependent on the plasma composition, being the Ar/N2- plasma more suitable for this task. In addition, the GISAXS analysis reveals that the pore structure can be modeled by two spheres and one cylinder with geometrical dimensions remain almost constant in spite of the plasma exposition. (author)