(Color online) Layout of the EUV MiMICS at CNSE. Carbon containing molecules were injected into the vacuum chamber to increase the contamination rate.
oval aperture was designed to cause the photoinduced chemical reaction of hydrocarbons on surface within only a selected area. As shown in the optical micrograph taken by a reticle SEM, the dark oval is the carbon contamination that covers a subfield on a patterned EUV mask.
Reticle SEM images show that the width of the designed 200 nm half-pitch dense lines increased by 23.5 nm after contamination.
(Color online) CD change (labeled as ) and the reflectivity loss were measured from selected fields on the mask, and results show that the reflectivity loss scales with the CD change, which means that there is carbon contamination on the sidewalls of the absorbers. The density of carbon was assumed to be and carbon thickness can then be calculated and shown below the reflectivity loss data. The calculated carbon thickness approximately matches the measured CD change, which indicates that there is about one-half as much carbon thickness on each sidewall as on the top surface of the multilayer and absorbers.
(Color online) Focus exposure matrix shown in (a) represents the printability at various doses and focuses. Trend lines of measured linewidth, or so called Bossung plots, are also shown in (b). With 5% dose steps, the exposure latitude across ±10% CD variation was 30% at best focus.
(Color online) Focus exposure matrix in (a) had worse printability than images printed at the clean region on the mask. From the Bossung plot shown in (b), the best dose to print a 40 nm target CD is , which is about 21% more than the best dose to print the clean region on the mask with 8 nm of carbon deposited. The exposure latitude across ±10% CD variation is 25% at best focus.
(Color online) Required dose change to print dense lines vs carbon thickness is plotted and compared to the measured reflectivity loss on the selected fields of the mask. Carbon thicknesses were determined using absorption data from the CXRO website.
(Color online) (a) 3D computer images of surface morphology were created based on surface scan using an AFM. Features shown are 225 nm lines and 675 nm spaces (1:3 lines/spaces ratio). rms roughness on a clean mask was 0.29 nm, which increased to 0.45 nm on a contaminated region on the mask. (b) FWHM of 225 nm lines and spaces features was measured on the mask. The sidewall of the absorber increased by 40.6 nm for the contaminated features.
(Color online) Aerial images recorded using the SEMATECH Berkeley AIT show 200 nm dense line features on the mask for the clean and contaminated areas. The boxes shown were selected for data analysis.
Intensity profile was recorded based on the location of imaging processing on the mask. Results show less contrast is recorded from the contaminated region on the mask.
(a) CD through focus showed the normalized intensity at different threshold levels from 0.1 to 0.9, spaced by 0.1. The CD change at best focus for clean region on the mask was increased from 218 to 250 nm after contamination. (b) Contrast through focus is calculated at a 200 nm 1:1 lines/spaces region on the mask. (c) LWR was shown above at 1:1 intensity on the mask. Equivalent sizes for 5× demagnification are also shown. (d) NILS was determined at the best focus for each field. The larger NILS, the better image quality of printed features, and the clean region had a 10% larger NILS than the clean region.
Two possible topographies: direct and conformal deposition. The film stack used in the simulation consisted of 40 bilayers of Mo/Si, 10 nm of as a buffer layer, and 70 nm of TaN as an absorber layer. The gray layer shown in the schematic view represents the deposited carbon layer.
Shadowing effects occurs on patterned masks when EUV radiation is perpendicular to the absorbers.
(Color online) Simulations of the dose required to print 40 nm target CD were performed for both the ASML ADT and the SEMATECH Berkeley MET. The difference in optical designs between these two tools led to a 7% difference in dose required to print these features.
(Color online) Simulations of dose required to print 40 nm target CD were performed for conformal and direct carbon topography, as well as shadowed and nonshadowed illumination conditions. The experimental results fall between the conformal and direct topographies for both the shadowed and nonshadowed illumination conditions.
(Color online) Plot of the dose required to print the target CD of 40 nm dense lines on the wafer. With increasing carbon thickness, a larger dose is required and the divergence of the dose curves for shadowed and nonshadowed cases increases.
Article metrics loading...
Full text loading...