No abstract

No abstract

The basic chemistry and lithographic characteristics of anhydride-containing, diazo-based NUV and DUV resists as well as silylation of top resist patterns with aqueous solutions of silicon-containing diamines in the Si-CARL bilayer process (CARL: Chemical Amplification of Resist Lines) were reported recently. This paper describes technical control of the Si-CARL process for g-line and DUV in a 6" pilot line using automatic equipment. Linewidth uniformity of top resist patterns is not affected by silylation and is found to be 0.045 im (3) for nominal 0.4 jim lines! spaces, the resolution limit of the 0.55 NA g-line stepper used. Both overexposure and linewidth increase due to silylation conditions can be used in the Si-CARL process for optimization of defocus latitudes. With the use of a 0.55 NA g-line stepper total defocus latitudes are 2.8 im for 0.6 im equal lines and spaces and > 3.2 xm for isolated 0.6 im spaces. In order to meet the requirement for sufficient throughput on KrF-excimerlaser steppers the sensitivity of DUV top resists is improved by chemical variations of resist polymers. The use of maleimide-containing resist polymers with improved alkaline solubility in diazo-inhibited top resists allows resolution of 0.25 im lines and spaces at 161 mJ!cm2 on a 0.37 NA KrF-excimerlaser stepper. Further considerable improvement ofDUV sensitivity to 1 1 mJ!cm2 was achieved using an acid-catalyzed top resist based on onium-salt and a terpolymer containing N-t-BOC-maleimide-units.

Alternating phase shifting masks have proven their capability to enhance the process window and to reduce the mask error enhancement factor effectively. The application of this mask type requires additional mask-properties compared to binary masks or halftone PSM. In this paper two of these mask-properties, the intensity and the phase balancing, are investigated experimentally for 4X and 5X masks at DUV and compared with simulations applying the T-Mask configuration of the SOLID-CM™ program. In a first part the experimentally determined balancing results are discussed. For the measurements two independent methods are compared: Balancing measurements with an AIMS-system (MSM100) and direct optical phase and transmission measurements using a MPM-248 system. The T-Mask as a 3D Maxwell solver allows the simulation of real 3D mask topography. We compare the results of simulations with measured AIMS data. All available mask data like depth of trenches, thickness and composition of chromium/Cr x O y layers, etc. are taken as input for the simulations. The comparison enables an assessment of the possibilities and limitations of 3D masksimulation. Based on 3D mask simulations CD-sensitivity of the different balancing methods was investigated also taking the influence of proximity into account. The simulations allow an assessment of the CD-sensitivity for four analyzed mask types for feature sizes below 150nm on the wafer.

By assessment of options for the fabrication of small contact holes in DRAM devices the method of focus drilling was identified and investigated to overcome the depth of focus limitations. By use of ArF-dry lithography a practical shrink of the target CD by 15nm can be achieved both with a focus offset double exposure (FODEX) and with a tilted stage approach. This was optimized in simulation and demonstrated by CD measurement on wafer, as well as by electrical measurement on integrated lots. Application of dual lambda focus drilling is limited by the chromatic magnification error of the lens. The increase of hole-to-hole CD variations due to a lower dose latitude and to increased MEEF was characterized. As improvement option the use of a high transmission attPSM was identified.