Search

Cellulosic polymers were used as organic additives to reduce the wafer surface roughness during the polishing process. The physico-chemical affinity of the polymer for the wafer and the formation of a hydro-plane on the Si wafer surface were investigated to identify the role of the polymeric additives in the wafer-polishing process. Two different polymers, hydroxyethyl cellulose (HEC) and carboxymethyl cellulose (CMC), were employed as a surface modifier for the wafer. The wettability of the Si surface varied markedly between the different suspensions prepared with HEC and CMC due to different adsorptive behaviors. As a result, the suspension prepared with HEC reduced the haze level and micro-roughness of the wafer, enhancing overall polishing performance.

The dependencies of the non-Prestonian behavior of ceria slurry with anionic surfactant on the size and concentration of abrasive particles were investigated by performing chemical mechanical polishing (CMP) experiments using blanket wafers. We found that not only the abrasive size but also the abrasive concentration with surfactant addition influences the non-Prestonian behavior. Such behavior is clearly exhibited with small abrasive sizes and a higher concentrations of abrasives with surfactant addition, because the abrasive particles can locally contact the film surface more effectively with applied pressure. We introduce a factor to quantify these relations with the non-Prestonian behavior of a slurry. For ceria slurry, this non-Prestonian factor, βNP, was determined to be almost independent of the abrasive concentration for a larger size and a smaller weight conentration of abrasive particles, but it increased with the surfactant concentration for a smaller size and a higher concentration of abrasives with surfactant addition.

The reverse selectivity of nanosized ceria slurry for the damascene gate chemical mechanical planarization (CMP) process has been investigated. The relationship between the electrokinetic behaviors of abrasive ceria particles and the plasma-enhanced tetraethylorthosilicate (PETEOS), and CVD Si3N4 films in an aqueous medium were observed to determine the reverse selectivity performance in the CMP process. The surface potentials of abrasive and PETEOS/CVD Si3N4 films were found to have different surface charges in certain pH regions and these differences were found to be key parameters in the selective removal performance of damascene gate CMP process. The ceria slurry with polymeric additives showed reverse selectivity with a low PETEOS removal rate and a high CVD Si3N4 film removal rate.

By considering the physical characteristics of ceria particles, the factors affecting the chemical mechanical planarization (CMP) performance for shallow trench isolation (STI) were investigated. Ceria powders were synthesized by calcination at temperatures of 400, 600, 700, 800, and 900°C. The influence of the calcination temperature on the physical characteristics, such as the crystallinity, grain size, porosity, morphology and high-resolution lattice images of the ceria particles, were investigated, and a correlation between the physical characteristics and the electrokinetic behaviors of the ceria particles in aqueous suspending media was also investigated to identify the relationship with CMP performance. Grain size increased with calcination temperature, as did the removal rate of plasma-enhanced tetraethylorthosilicate (PETEOS) film with slurries including these particles. The results indicate that the grain size of ceria particles is one of the key parameters controlling the removal rate and uniformity of PETEOS film.

The nanotopography of the surface of silicon wafers has become an important issue in ULSI device manufacturing, as it affects the post–chemical mechanical polishing (post-CMP) uniformity of the thickness deviation of dielectric films. A spectral method is proposed to examine quantitatively how the nanotopography impacts the film thickness deviation during CMP. The nanotopography impact was investigated in terms of its dependence on the characteristics of consumables, such as the polishing pad hardness and the wafer manufacturing method. In addition, the effects of the surfactant and the abrasive size in ceria slurry on nanotopography impact were investigated. It was found that the magnitude of the post-CMP oxide thickness deviation due to nanotopography increased with the surfactant concentration in the case of smaller abrasives but was almost independent of the concentration in the case of larger abrasives. These results demonstrate that the nanotopography impact can be controlled by manipulating the slurry characteristics.

In a ceria slurry with an ionic surfactant, the grain size and particle size of the poly-crystalline abrasives were controlled independently by changing the calcination temperature and the mechanical milling time, respectively, during abrasive synthesis. A chemical mechanical polishing (CMP) experiment using the slurry showed that the oxide removal rate increased with both the grain size and the abrasive particle size, while the nitride removal rate was independent of both. On the other hand, examination of the nanotopography impact showed that the planarization efficiency increased with decreasing abrasive size but was independent of the grain size.

We investigated the nanotopography impact on the post-chemical mechanical polishing (post-CMP) oxide thickness deviation (OTD) and the non-Prestonian behavior of ceria slurry with a surfactant. Not only the surfactant but also the slurry abrasive size influenced the non-Prestonian behavior and the nanotopography impact. We created a one-dimensional numerical simulation of the nanotopography impact by taking account of the non-Prestonian behavior of the slurry, and good agreement with the experimental results was obtained. After examining the mechanism of oxide surface planarization during CMP, an index to relate the non-Prestonian behavior of the slurry to the nanotopography impact was developed.

We used a blank chemical mechanical polishing (CMP) test to examine the dependence of nanotopography impact on surfactant concentration in three ceria slurries differing in abrasive size. We found that the magnitude of film thickness variation after CMP increased with the surfactant concentration in case of the smaller abrasives but for the case of larger abrasive was almost independent of the surfactant concentration. This result was explained with a model based on the passivation layer of the surfactant adsorbed on the oxide film surfaces during polishing.

Through a chemical mechanical polishing (CMP) test using oxide and nitride blanket film wafers, we investigated the effect of adding an anionic surfactant to a ceria slurry, especially at very dilute concentrations (less than 0.2 wt%). The removal rate trend, which decreases with increasing surfactant concentration, was compared with the variation in the electrokinetic behavior of the ceria abrasives dispersed in the slurry. The removal rate transition did not coincide with the changes in the zeta potential or the secondary particle size distribution of the abrasives. This suggests that the electrostatic effect is not the dominant factor in the removal rate with ceria slurry. It is also found that abrasive size influences removal rate trend versus surfactant concentration, and that oxide removal follows Preston's law but nitride removal shows non-Prestonian behavior. To explain the result qualitatively, we have proposed a hydrodynamic model in which a particle travels through the viscous fluid and approaches the film surface.

The effects of the electrokinetic behavior of abrasive ceria particles suspended in an aqueous medium and the deposited plasma-enhanced tetraethylorthosilicate (PETEOS) and chemical vapor deposition (CVD) Si3N4 films on chemical mechanical planarization (CMP) for shallow trench isolation were investigated. The colloidal characteristics of ceria slurries, such as their stability and surface potential, in acidic, neutral, and alkaline suspensions were examined to determine the correlation between the colloidal properties of ceria slurry and CMP performance. The surface potentials of the ceria particles and the PETEOS and CVD Si3N4 films in an aqueous suspending medium were dependent on the pH of the suspending medium. The differences in surface charges of ceria particles and the PETEOS and CVD Si3N4 films have a profound effect on the removal rate and oxide-to-nitride selectivity of CMP performance.

Ceria powders were synthesized by two different methods, solid-state displacement reaction and wet chemical precipitation, and the influence of the physical characteristics of cerium oxide on the removal rate of plasma-enhanced tetraethylorthosilicate (PETEOS) and chemical vapor deposition (CVD) nitride films in chemical mechanical planarization (CMP) was investigated. The fundamental physicochemical property and electrokinetic behavior of ceria particles in aqueous suspending media were investigated to identify the correlation between the colloidal property of ceria and the CMP performance. The surface potentials of two different ceria particles are found to have different isoelectric point (pHiep) values and differences in physical properties of ceria particles such as porosity and density were found to be the key parameters in CMP of PETEOS films. Ceria powders synthesized by the solid-state displacement reaction method yielded a higher removal rate of PETEOS and higher selectivity than powders synthesized by the wet chemical precipitation method.

To reveal the mechanism behind the high oxide-to-nitride removal selectivity of ceria slurry in shallow trench isolation (STI) chemical mechanical polishing (CMP), we examined the effects of the abrasive morphology and the concentration of surfactant added to control the selectivity. A slurry with small abrasives showed a more drastic drop in the oxide removal rate as the surfactant concentration increased, compared to a slurry with large abrasives. For the nitride removal rate, however, both slurries showed the same trend. These results can be qualitatively explained from the movement of abrasives though the layer of surfactant adsorbed on the film surface.

Single-side-polished (SSP) wafers prepared using six different methods were examined following unpatterned oxide chemical mechanical polishing (CMP). We found a clear and positive correlation between the standard deviation of the nanotopography profile and that of the film thickness variation after unpatterned oxide CMP. Our power spectral density (PSD) analysis of the profiles quantitatively demonstrated how the nanotopography of the different wafers affected the film thickness variation. We also confirmed that the wafers prepared with an improved SSP technique showed small film thickness variation after unpatterned oxide CMP, indicating that the nanotopography had less impact on the film thickness.

The effect of surfactant in a ceria slurry on the impact of nanotopography on post-CMP oxide thickness deviation (OTD) was investigated in a blanket wafer study. The ceria slurry was prepared using the solid-state displacement reaction method and an anionic organic surfactant was added with the concentration from 0 to 0.8 wt%. Under the condition of a fixed removal depth, the magnitude of post-CMP OTD due to nanotopography increased with the surfactant concentration in the slurry, demonstrating that the impact of nanotopography can be controlled using the slurry characteristics.

The polishing pad dependency of nanotopography impact on oxide chemical mechanical polishing (CMP) was analyzed using a spectral method. Oxide films on wafers were polished using two types of polishing pads having different compressibility (soft pad and hard pad). The power spectral density (PSD) of the film thickness variations got closer to that of nanotopography after CMP. A transfer function was proposed as the ratio of PSD of the film thickness variation to that of nanotopography. The calculated transfer function quantitatively showed that the nanotopography impact on film thickness variation is apparent up to the longer wavelengths when using the hard pad.

We report a case of tumefactive demyelinating lesion (TDL) diagnosed using (1)H-magnetic resonance spectroscopy (¹H-MRS) and conventional magnetic resonance imaging (MRI). A 7-year-old girl was admitted to our hospital with complaints of sleepiness and clumsiness of the right limbs. Neurological examination showed somnolence, right-sided apraxia, and hemiparesis with enhanced tendon reflexes and Babinski sign. Conventional brain MRI revealed extensive hyperintensity in the subcortical white matter of the left frontal lobe in both T₂ weighted and fluid attenuated inversion recovery images. Gadolinium-enhanced T₁ weighted images showed a tumor-like lesion in this area with interrupted rim enhancement, termed open ring sign, and a periventricular lesion along the inferior horn of the right lateral ventricle and a juxtacortical lesion under the right motor cortex. In ¹H-MRS, both single voxel spectroscopy (SVS) and chemical shift imaging showed elevation of choline and reduction of N-acetylaspartate in the left frontal lobe lesion. Furthermore, SVS with a short echo time revealed elevated peaks for glutamate/glutamine complex in this lesion. These results suggested the demyelinating nature of this tumor-like lesion, in accordance with the concept of TDL. Based on this diagnosis, we treated the patient with three sets of methylprednisolone pulse therapy, which resulted in the reduction of TDL and neurological improvement. A follow-up study using MRI also demonstrated two more lesions in the corona radiata and internal capsule of the left hemisphere, supporting a diagnosis of multiple sclerosis based on the revised McDonald's criteria (2010). We concluded that ¹H-MRS may be beneficial in the differential diagnosis of TDL.

We have characterized subsurface damage profiles of hydrogen-ion implanted silicon wafers by using a non-contact UV/Millimeter-Wave Technique and Light Scattering Topography (LST). A subsurface damage profile that was less than one micrometer was controlled by chemical mechanical polishing after hydrogen-ion implantation. On the area with the subsurface damage, the Photoconductivity Amplitude (PCA) signals measured by the UV/Millimeter-Wave Technique drastically weakened and the haze values measured by LST increased. A clear correlation has been found between the peak depth of the subsurface damage and the haze value. The spectral analyses of the surface images obtained by Atomic Force Microscopy (AFM) were carried out in order to separate the influences of surface micro roughness and subsurface damage on the haze value. The contribution of subsurface damage to the haze value can be formulated as the convolution of the damage profile and the transparency function of the incident laser in silicon crystal.

The UV/mm-wave technique composed of ultraviolet photoexcitation and millimeter wave probe was examined with photoconductivity amplitude (PCA) to characterize the slight subsurface damage induced by implanting H2+ ion into the subsurface at sub micron depth of Si wafers. The identical samples were also characterized using pulse photoconductivity amplitude (PPCA) obtained by another technique which is specified by blue laser photoexcitation and microwave probe. PCA decreased with increase of ion dose, which coincided well with the result in PPCA. PPCA decreased with increase of implantation energy as 90 to 120 keV, but PCA increased at 120keV. Both PCA and PPCA well reflected the damage at sub micron depth. PCA reflected damage in shallower depth compared to PPCA.

We examined both the dispersant mixing time during ceria slurry synthesis and a method for reducing the quantity of agglomerated large particles, which influence the number of light point defects (LPDs) formed after chemical mechanical polishing (CMP). We quantified the dispersion stability of slurries with the abrasive particle size by examining with and without ultra-sonic treatment. Without the addition of dispersant before mechanical milling, the dispersion stability was worse than the case with dispersant addition before milling. A lower pH of cerium carbonate improved both the dispersion stability and reduced the number of LPDs formed on an oxide film after CMP.

We examined the dependence of a non-Prestonian behavior on surfactant concentrations in ceria slurries that contained abrasives of various sizes by performing chemical mechanical polishing (CMP) tests using blank wafers. We found that both the surfactant concentrations and the slurry abrasive sizes influence the non-Prestonian behavior. In this paper, we explain in detail a factor that can be used to quantify relation on the non-Prestonian behavior of a slurry, βNP. This non-Prestonian factor βNP was determined to be almost independent of the surfactant concentration in slurry A (large abrasive), but increased with the surfactant concentrations in slurries B (medium-sized abrasive) and C (small abrasive).

We examined the effect of the molecular weight of surfactants in ceria slurry during chemical mechanical polishing (CMP) for shallow trench isolation (STI). We found that for a surfactant with a higher molecular weight, the oxide removal rate decreased drastically as the surfactant concentration increased, but in the case of a lower molecular weight, it only slightly decreased. In addition, slurries whose surfactants had lower molecular weights maintained a higher nitride removal rate with increasing surfactant concentration. The results showed that the molecular weight and surfactant concentration have complex effects on the oxide removal rate and the oxide-to-nitride removal selectivity.

The purpose of this study is to investigate quantitatively how the height change of silicon wafer surfaces (nanotopography) impacts on the variation of removal rate during oxide chemical mechanical polishing (CMP). We have analyzed the spatial power spectral densities (PSD) of both nanotopography and thickness variation of oxide film after CMP. We defined a transfer function as the ratio of PSD for the oxide film thickness variation to that for nanotopography. The transfer function decreased with increase of the wavelength and this result quantitatively indicates that longer wavelength of nanotopography influence relatively less on the oxide thickness variation after CMP.