Myung-Goon Gil

Optimization of High Order Control including overlay, alignment and sampling

Overlay requirements for semiconductor devices are increasing faster than anticipated. Overlay becomes much harder to control with current methods and therefore novel techniques are needed. In this paper, we present our investigation methods for High Order Control, and the candidates for improvement. This paper will present the study for each components of high order control. High order correction is one component for high order control and several correction methods were compared for this study. High order alignment is another important component for higher order control instead of using conventional linear model for the alignment. Alignment and overlay measurement sampling decision becomes a more critical issue for sampling efficiency and accuracy. Optimal sampling for high order was studied for high order control. Using all these studies, various applications for optimal high order control have also been studied. This study will show the general approach for high order control with theory and actual experimental data.

Investigation of electron beam stabilization of 193-nm photoresists

193nm lithography is a promising candidate for the fabrication of microelectronic devices at the 130nm design rule and below. With smaller feature sizes, below 130nm, reduced resist thickness is essential because of the pattern collapse issues at high aspect ratios and the limited depth of focus with 193nm lithography tools. However, ArF resists have shown problems with etch selectivity, especially with the thin resist layers necessary. Additionally, pattern slimming during CD-SEM measurement, due to the nature of the resist chemistry, is an issue with feature stability after patterning. At present, many studies have been performed for improving the etch selectivity of resists and addressing line slimming issues. In this study, the electron beam stabilization process has been applied for improving the etch selectivity of resist patterns having an aspect ratio less than 3.0. The electron beam stabilization has been applied to two different ArF resist types; acrylate and cyclic-olefin- maleic-anhydride (COMA), which have been evaluated with respect to materials properties, etch selectivity, and line slimming performance as a function of electron beam dose and etch condition. Film shrinkage and the change in index of refraction were monitored as a function of stabilization condition. The chemical properties were characterized before and after electron beam stabilization using FTIR analysis. Blanket resist etch rate studies were performed as a function of stabilization condition for each resist type. Cross- sectional views of resist patterns after etch processing were also investigated to evaluate the improvement in etch resistance provided by the electron beam process. CD SEM measurements were performed to evaluate the impact of the stabilization process on the patterned features. The issue of line slimming has also been evaluated, with and without electron beam stabilization, for the different ArF resist materials considered. The results were compared with a KrF resist currently used in production. Based on the experimental results, the electron beam process provides a method for improving etch selectivity and reducing line slimming issues of ArF resists.

Reflow process stabilization by chemical characteristics and process conditions

With the shrunken device rule below 130nm, the patterning of smaller contact hole with enough process margin is required for mass production. Therefore, shrinking technology using thermal reflow process has been applied for smaller contact hole formation. In this paper, we have investigated the effects of chemical characteristics such as molecular weight, blocking ratio of resin, cross-linker amount and solvent type with its composition to reflow process of resist and found the optimized chemical composition for reflow process applicable condition. And several process conditions like resist coating thickness and multi-step thermal reflow method have been also evaluated to stabilize the pattern profile and improve CD uniformity after reflow process. From the experiment results, it was confirmed that the effect of crosslinker in resist to reflow properties such as reflow temperature and reflow rate were very critical and it controlled the pattern profile during reflow processing. And also, it showed stable CD uniformity and improved resist properties for top loss, film shrinkage and etch selectivity. The application of lower coating thickness of resist induced symmetric pattern profile even at edge with wider process margin. The introduction of two-step baking method for reflow process showed uniform CD value, also. It is believed that the application of resist containing crosslinker and optimized process conditions for smaller contact hole patterning is necessary for the mass production with a design rule below 130nm.

The study of defect control and patterning performance for top coating free resist process

In this study, the top coating (TC) resist and new top coating free (TCF) resist at the condition of immersion patterning process have been evaluated and compared in the respect of leaching rate, process margins and immersion related defects. The 55nm of dense L/S patterns are defined by using immersion process. The leaching rates of TCF resist are investigated for the different conditions of baking temperatures and leaching times. Their measurements have been done for the cation and anion dissociated from PAG under the conditions of unexposed and exposed areas. The cation leaching rate of new TCF resist is 1.4 × E-13 mol/cm2.sec at the unexposed area. It is lower than the spec suggested by ASML. From the leaching data, it is confirmed that the TCF resist can apply for the patterning without top coating material. The process margins of TCF resist are slightly wider than those of TC resist in the respect of depth of focus and energy latitude. And the top loss of TC resist and TCF resist are 20nm and 8nm, respectively after development process. The immersion related defects are also investigated for these resists. The defect map of TCF resist shows much fewer defects than those of TC resist. The immersion related defect of TCF resist is not founded among the patterns and the number of micro-bridge defect is fewer than 5. The defect level of TCF resist is more stabilized by optimizations of chemical structure of resist and process. From the experimental results, it is confirmed that the TCF resist is available to apply for device production of sub-55nm technology. Especially, the application of TCF resist can reduce the process step of top coating and also make the improvement of through-put with cost of ownership in process. Therefore, the TCF resist should be progressively studied and applied for device production in the respect of improvement of device property and process simplification.

New ArF resist introduction for process through-put enhancement

In this study, the four ArF resists having methacrylate structure have been evaluated to check the reliability of process through-put enhancement in track by applying the different baking time. The dense L/S and isolated patterns of T80nm and T66nm node device are investigated for these models. The chemical properties of applied resists are slightly different for each other in the respect of protecting ratio and molecular weight. The applied resist thickness are 2,000Å and 1,700Å for the patterning of T80nm and T66nm node device, respectively. The process margins of evaluated patterns for T80nm node device show the almost similar results for DOF and E/L for 90s, 60s and 45s baking time conditions. And the LER of these patterns also is not much changed by the different baking time conditions. Form the experimental results, it is confirmed that the less baking time below 60s for T80nm node pattern is available to apply for enhancing the through-put in litho process. The process margins and LER of evaluated patterns for T66nm node device are slightly affected by their chemical properties like molecular weight and protecting ratio of resin for the different baking time conditions. However, the more improved pattern profile and process margin can be obtained by optimizing the chemical properties in the certain experimental range. And also, it is found that the less baking time below 45s can be applied for the reliable patterning process of T80nm and T66nm node device through the crosssectional SEM views with the more optimizing the material compositions.

Implantation blocking characteristics study of organic BARC materials

In this study, the four different types of ArF and KrF OBARC have been evaluated to know the implantation blocking and gap fill performance for 80nm node device application. The boron implantation conditions of 11B and 49BF2 are processed and the minimum energy for implantation blocking of these OBARC are obtained by SIMS analysis. The minimum energy of ArF and KrF OBARC are about 13.0KeV and 15.0KeV, respectively. The chemical density of each OBARC is also calculated from the minimum blocking energy. Their values of ArF and KrF OBARC are about 0.8g/cm3 and 1.0g/cm3, respectively. The minimum energy trends among the tested materials show the almost similar results with those of chemical density as expected. Even though the OBARC are composed of the similar chemical structure, they induce the different chemical density because of their own molecular weight and other additional structure as like chromophore. Both of KrF and ArF OBARC show the good gap fill performance on 0.2μm size of via substrate and real topology pattern without void. It seems that the gap fill property is not much affected by the chemical structure or molecular weight of OBARC. It is thought that OBARC is an effective material for gap fill application than other resists, especially for deep topology patterns. In general, the etch rate of OBARC is slightly faster than that of ArF resist or similar with that in this experimental condition. The OBARC having high chemical density shows the slower etch rate and that of OBARC is inversely proportional to the chemical density of it. Therefore, it is confirmed that the OBARC is able to apply for implantation blocking purpose without gap fill void in real device below 80nm, since they have the good characteristics for gap fill, reflectivity control from substrate and implantation blocking property at a certain coating thickness.

Optimization of 248nm bottom anti-reflective coatings with thin film and high etch rate on real device

A frequent problem encountered by photoresists during the manufacturing of semiconductor device is that activating radiation is reflected back into the photoresist by the substrate. So, it is necessary that the light reflection is reduced from the substrate. One approach to reduce the light reflection is the use of bottom anti-reflective coating (BARC) applied to the substrate beneath the photoresist layer. The BARC technology has been utilized for a few years to minimize the reflectivity. As the chip size is reduced to sub 100nm, the photoresist thickness has to decrease with the aspect ratio being less than 3.0. Therefore, new Organic BARC is strongly required which has the minimum reflectivity with thinner BARC thickness and higher etch selectivity toward resists. Hynix Semiconductor Inc., Nissan Chemical Industries, Ltd., and Brewer Science, Inc. have developed the advanced Organic BARC for achieving the above purpose. As a result, the suitable high performance 248nm Organic BARCs, NCA series, were achieved. Using CF4 gas as etchant, the plasma etch rate of NCA series is about 1.4 times higher than that of conventional 248nm resists. NCA series can be minimizing the substrate reflectivity at below 45nm BARC thickness. NCA series show the excellent litho performance and coating property on real device.

Application of new thin BARC technology for KrF lithography at 80-nm node device

The new thin BARC has been developed for the application of small size patterning below 100nm by the optimized simulation and the evaluations on each substrate condition of silicon nitride and silicon oxide. The optical parameters of thin BARC of Exp225 are 1.81 and 0.58 for n and k values, respectively. They are obtained by the simulation for the lower reflectivity at the conditions of silicon nitride and silicon oxide. The optimized BARC thickness of Exp225 are 320Å and 460Å for silicon nitride and oxide substrate, respectively, at the condition of reflectivity. These thickness are much lower than those of commercial BARC of DUV44 for the same substrate conditions. The pattern profile and process margin are compared between the inorganic SiON and organic BARC. The dense L/S pattern profile of 100nm size on SiON shows the severe standing wave and undercutting. However, the pattern on Exp225 is much stable and gives wider depth of focus margin than that of SiON condition. The 85nm dense L/S pattern with feasible process margin is obtained by the application of Exp225 at the thickness of 320Å. The baking temperature is also investigated for the application of mass production. The most optimized baking temperature ranges of Exp225 are between 205°C and 225°C. From the experimental results, it is confirmed that the application of thin BARC is much effective for the small size patterning of 80nm node device. And it is thought that 80nm node device by KrF lithography is possible under the conditions of thin BARC, high contrast resist and high NA exposure tool.

Diffusion parameter analysis for chemical amplification resists as a function of resist process

The diffusion parameters of chemical amplification resists are evaluated and analyzed as the functions of illumination condition, resist type and resist thickness. The pattern linearity is also compared among the different types of resist and process conditions. From the experimental results, it is confirmed that the diffusion parameters of dense L/S type resist is sensitively influenced by illumination condition, but those of C/H type resist are less affected by exposure condition due to the limited resolution of resist. Generally, the diffusion parameters seems to be much affected by sigma condition rather than the numerical aperture condition and it does not much influenced by resist thickness within similar thickness range. The 120nm, 100nm and 90nm dense L/S patterns are obtained with high contrast KrF resist of R5767 having diffusion parameter below 0.3 at the conditions of 0.80NA (89/60 sigma) and 2500 Angstrom thickness. Under the validation of DAIM, the most important parameter is the diffusion length of acid. Since the image contrast is given by the cross product of aerial image contrast and the resist function, the lithography performance depends upon not only the aerial image but also the resist function. The resist function is related with pattern pitch and diffusion length. Therefore, the increased value of resist function is required by introducing of high contrast resist having smaller diffusion parameter to induce the smaller pattern formation as explained in this experiment results.

Improved characteristics of rainbow defects with novel wafer edge exposure technique

The demand for manufacturing integrated circuit with high circuit speed and high packing density requires reduced feature sizes in ULSI structures. As the device feature size shrinks below sub-130 nm it needs the tight control of defect reduction in lithography process. Especially, resist peeling at the wafer edge is one of the major sources for particle generation. The WEE process removes resist up to a width of a few mm from the wafer edge in order to prevent particle generation in succeeding process. The defect induced form wafer edge after WEE has given the critical damage to electrical properties and device yield. In this paper, we have applied novel WEE kit to reduce the rainbow bandwidth caused by WEE step in wafer wedge. The novel WEE kit consists of chrome slit and lens assembly to minimize the scattering of UV beam from the optic fiber in comparison with the conventional WEE kit. The change of rainbow bandwidth was also characterized by OM and SEM. With the novel WEE kit the bandwidth of rainbow is reduced to 5 micrometers , while the conventional WEE kit has been induced 20 micrometers of bandwidth on bare silicon wafer. In the case of patterned wafer, the bandwidth of rainbow is reduced to 60 micrometers for the novel WEE kit, while the conventional WEE kit has induced 230 micrometers of bandwidth. Therefore, ti is confirmed that the application of novel WEE kit has induced 230 micrometers of bandwidth. Therefore, it is confirmed that the application of novel WEE kit for patterned process makes less rainbow defect and finally increases the device yield for mass production.