Table of Contents

Effectiveness of High Energy Ion Beam Techniques for the Characterization of Mesoporous Low Dielectric-Constant Materials S. Baskaran,(a,b) J. Liu, and S. Thevuthasan Supported by LTR, EMSL Operations. (a) EMSL User. (b) Environmental Health and Sciences Division. There is a growing interest in low dielectric materials, which can be used to decrease the effective capacitance between metal interconnections in semiconductor devices. Porous silica films are potential candidates for this purpose due to their low dielectric properties. Nanoporous silica films have been previously synthesized by several research groups using aerogel or xerogel (Smith et al. 1995; Jin et al. 1997) and surfactant-templating (Yang et al. 1996; Aksay et al. 1996; Bruinsma et al. 1997; Yu et al. 1997) processes. Typical pore size ranges from a few nanometers to a few tens of nanometers and the pores are either random or ordered in nature. We have synthesized "mesoporous" silica films using a spin-on sol-gel process (Bruinsma et al. 1997) and a detailed description of deposition processes along with various dehydroxylation treatment processes has been provided elsewhere (Baskaran et al. submitted). We have effectively used Rutherford backscattering spectrometry (RBS) and nuclear reaction analysis (NRA) to determine the porosity of these films. We have used NRA for the first time to quantify the total number of oxygen atoms in porous silica films. In this extended abstract, a brief summary of porosity measurements using the high-energy ion scattering techniques along with the effects of high-energy ion beams on the films is presented. A detailed discussion of this study is reported elsewhere (Thevuthasan et al. in press). The accelerator facility and the end stations are described in detail elsewhere (Thevuthasan et al. 1998). For the RBS and hydrogen NRA measurements, the sample was mounted on a molybdenum backing plate using Ta clips, and the conventional alumel-chromel thermocouple was attached to a corner of the sample surface with a Ta clip for temperature measurements. The backscattering spectrum was collected using a silicon surface barrier detector at a scattering angle of 150°. The primary energy of the ions was 2.04 MeV and the incident ion beam was directed normal to the sample surface. The 1H(19F,ag)16O resonant nuclear reaction was used to quantify the hydrogen (H) in the sample. The 16O(d,p1)17O nuclear reaction was used to quantify the O atoms in the mesoporous Si films. A 100-nm, thermally grown, dense SiO2 film on Si was used as a standard to determine the RBS thickness of these films. An aluminum coated Mylar window with known thickness was used as a cover to the detector to block the alpha particles and to reduce the backscattering deuterium yields. The physical thickness of the porous silica films was measured with an optical imaging profilometer (ZYGO New View 200). As an example for RBS, we show the aligned and random spectra from a 240-nm (RBS thickness) porous silica film in Figure 4.1. Figure 4.1. Aligned and random RBS spectra for an epitaxially grown SiO2 mesoporous film on silicon. He+ ions with 2.04 MeV were used for these measurements. The scattering angle was 150°. There is no significant difference between the aligned and random Si and O yields from the film. As is expected, there is virtually no channeling effect in the film since the atomic ordering is not present in the film. On the other hand, the aligned yield from the substrate Si appears to be much smaller compared to the random yield. A small energy window (DEb = 572-604 channels) near the surface region of the substrate was used to calculate the minimum yield (cmin) for Si in the substrate. For the substrate, cmin for Si is determined to be 38%. In general, the minimum yield for well-ordered Si substrate without any film is in the order of few percent and the present value of 38% appears to be significantly high. One possible explanation for the high minimum yield could be divergence of the ion beam due to the dechanneling effects in the film from the random ordering of pores. Although the minimum yield is relatively high for the substrate, the Si and O yields from the film can be determined with little better statistics from the aligned spectrum due to the low background compared to the random spectrum, especially in the case of oxygen atoms. Since the quantification of oxygen atoms can be effectively achieved by using the nuclear reaction as described below, we have used NRA to determine the O atomic density in the film and the RBS was used to calculate the Si atomic density in the film. Figure 4.2 shows a spectrum measured during the 16O(d,p1)17O nuclear reaction analysis. We have only shown the spectrum for the channel numbers between 400 and 1000 in this figure. The backscattered deuterium ions lose energy in the Al-coated Mylar window and a tail of deuterium ions which are not stopped in the foil appear in the lower channels, and those are not shown in this figure. The first (near 400 channel number) and second (near 700 channel number) peaks are due to the proton (p1 and p0, respectively) yield from the oxygen nuclear reaction and the third peak (near 925 channel number) is due to the proton yield from the carbon nuclear reaction. In our measurements, we used a thermally grown, 100-nm-thick, dense SiO2 film on a Si substrate as a standard to quantify the total number of oxygen atoms in the film. The equivalent thickness, which is noted as RBS thickness, can be determined from this data. The RBS thickness determined using O-NRA data is in good agreement with the RBS thickness determined from Si-RBS data. Since the films have pores, the RBS thickness is smaller than the physical thickness of the films. The physical thickness of the films was measured by the optical imaging profilometer in this study. By comparing the physical thickness to the RBS thickness, the porosity of the film can be determined. The porosity is defined as the ratio of the difference between the physical and RBS thickness and the physical thickness. Figure 4.2. The proton spectrum from the 16O(d,p1)17O nuclear reaction. The incident ions are 0.65 MeV d+. The scattering angle was 170°. An aluminum-coated Mylar foil was used to stop the scattered deuterons. The dielectric constant of these films is influenced by the porosity of the films and other polar groups, especially hydroxyl groups adsorbed in the films. As such, it is important to know whether the film has an affinity to attract hydroxyl groups mainly from water. We carried out a series of H measurements to demonstrate the water adsorption in the films and the need to develop some treatment processes that can control the water adsorption. We have used the 1H(19F,ag)16O nuclear reaction to quantify the H in the film which was used in oxygen NRA and RBS measurements. Figure 4.3 shows the H concentration profiles from the film as grown, from the film which was heated to about 570 K in vacuum for 25 minutes, and from the film which was heated to 570 K in 1.0x10-5 Torr of water for 25 minutes. In general, the energy of F beam can be converted into depth inside the film. However, since the pores are randomly ordered in the film, multiple surfaces, which are open to the vacuum, can be visible to the beam that travels through the walls of pores. Therefore, it is not meaningful to convert the energy scale into a depth scale. However, the total H concentration in the film can be determined by integrating the data from the surface to the SiO2/Si interface. The hydroxyl groups and the remnants of organics in the film contribute to the total H concentration. During these measurements, it was found that the background water contributed about 25% of an increase in water content in the film under these experimental conditions. These measurements suggested the need for the treatment processes that can minimize the water adsorption in the films. The dehydroxylation treatment processes have been developed and utilized in generating porous silica films with minimum water adsorption and dielectric constants in the range of 2.0 to 2.2. (Baskaran et al. submitted). However, further work is needed to completely eliminate water adsorption over a wide range of humidity. Figure 4.3. H concentration profiles determined using 1H(19F,a g)16O nuclear reaction. A 2"x4" BGO detector was used. High-energy ion scattering techniques have been effectively used to determine the porosity of mesoporous silica films grown on silicon. RBS and oxygen NRA were used to determine the Si and O atomic densities in the film, respectively. Hydrogen NRA has been used to profile H concentration in the films. Significant mechanical damage has been observed in the films due to the high-energy ion beams. Although the chemical state change of SiO2 is not significant, a tail was observed in the Si 2p core level spectrum indicating the reduction of some silica near the surface region. References Aksay, I. A., et al., Science 273, 892 (1996). Baskaran, S., J. Liu, K. Domansky, N. Kohler, X. Li, C. Coyle, G. E. Fryxell, S. Thevuthasan, and R. W. Williford, Appl. Phys. Lett. submitted. Bruinsma, P. J., N. J. Hess, J. R. Bontha, J. Liu, and S. Baskaran, MRS Symp. Proc. 445, 105-110 (1997). Jin, C., J. D. Luttmer, D. M. Smith, and T. A. Ramos, Mat. Res. Soc. Bull. 39 (1997). Lu, Y., et al., Nature 389, 364 (1997). Smith, D. M., J. Anderson, C. C. Cho, G. P. Johnston, S. P. Jeng, Mat. Res. Soc. Symp. Proc. 381, 261 (1995). Thevuthasan, S., C. H. F. Peden, M. H. Engelhard, D. R. Baer, G. S. Herman, W. Jiang, Y. Liang, and W. J. Weber, Nucl. Instr. Meth. A 420, 81 (1998). Thevuthasan, S., S. Baskaran, K. Domansky, J. Liu and M. Engelhard, Nucl. Instr. Meth. B, in press. Yang, H., N. Coombs, I. Sokolov, G. A. Ozin, Nature 381, 589 (1996). William R. Wiley Environmental Molecular Sciences Laboratory Feedback: webmaster@emsl.pnl.gov Revised: June 12, 2001 Security & Privacy PNNL-13147