|Investigating Extreme Ultraviolet Lithography Mask Defects|
Since the 1970s, the semiconductor industry has strived to shrink the cost and size of circuit patterns printed onto computer chips in accordance with Moore’s law, doubling the number of transistors on a computer’s central processing unit (CPU) every two years. The introduction of extreme ultraviolet (EUV) lithography, printing chips using 13-nm-wavelength light, opens the way to future generations of smaller, faster, and cheaper semiconductors. EUV lithography relies on specialized lenses made of curved mirrors with reflective coatings called multilayers to print patterns with high resolution. One special flat mirror called a mask is particularly sensitive to even the smallest imperfections. To better detect and characterize mask defects, scientists at Berkeley Lab worked with SEMATECH, an international semiconductor industry consortium, to create a unique Fresnel zone-plate microscope on Advanced Light Source Beamline 11.3.2 called the SEMATECH Berkeley Actinic Inspection Tool (AIT).
EUV light can be difficult to control because nearly all materials, even air, strongly absorb it. So EUV lithography relies on exotic, curved-mirror lenses in vacuum chambers to focus and control light. These mirrors start as atomically smooth surfaces onto which a resonant-reflective multilayer coating is applied. The multilayer coating gives EUV mirrors close to 70% reflectivity, but they require exquisite control to produce and, of course, are sensitive to blemishes.
In lithography, the complex process used to create computer chips, a six-inch glass plate called a mask carries one layer of a circuit pattern—the image of which is transferred onto a silicon wafer that becomes a computer chip. A single undetected defect can ruin the mask and the entire process. Currently, one of the major obstacles to the commercialization of EUV lithography is the lack of mask inspection and imaging tools that can detect defects before they are printed and replicated on millions of chips.
Since EUV mask defects can look very different when inspected with different wavelengths of light, SEMATECH and Berkeley Lab researchers have created the AIT, which is specifically designed to image different types of mask defects that cannot be completely characterized with any other standard inspection technique. The microscope serves the semiconductor community by providing advanced mask research and development, years ahead of the availability of commercial tools.
EUV masks are vulnerable to different types of defects, including ten-nanometer-scale particles that fall on the surface, potentially creating dark spots. Tiny bumps or pits on the mask can also become buried below the thin, multilayer reflective coating, causing a local ripple, or phase change. These "phase defects" are as small as 1 nm high and can be very difficult or impossible to detect with non-EUV techniques.
For the first time, using a full-field EUV mask (a mask that contains a real CPU pattern), a group of researchers from Berkeley Lab’s Center for X-Ray Optics (CXRO) has cross-compared the appearance of different types of mask defects recorded in several state-of-the-art tools: the AIT, a scanning electron microscope (SEM), and a commercial deep ultraviolet (DUV) mask inspection tool.
Researchers found that some defects appear to be transparent or thin when inspected with DUV or SEM, but can be completely opaque to EUV light. Conversely, others that appear to be severe defects can be largely transparent to EUV. Phase defects in particular are often subtle and hard to detect with non-EUV techniques, but they can produce a strong intensity-change signature when imaged with the AIT.
Sensitivity differences among the various inspection techniques are caused by the strong wavelength dependence of the defects’ optical properties, enhanced by the physics of multilayer reflective coatings. Currently, the only accurate way to predict the effects of a given defect is to inspect it with the same wavelength used in the lithography process (at-wavelength inspection). Based on the results from the AIT, the semiconductor industry has realized this important point and has recently begun significant investment in EUV mask imaging tools in preparation for high-volume manufacturing.
Research conducted by I. Mochi and K. Goldberg (Berkeley Lab), B. LaFontaine (Cymer), and A. Tchikoulaeva and C. Holfeld (Global Foundries).
Research funding: SEMATECH. Operation of the ALS is supported by the U.S. Department of Energy, Office of Basic Energy Sciences.
Publication about this research: I. Mochi, K.A. Goldberg, B. La Fontaine, A. Tchikoulaeva, and C. Holfeld, “Actinic imaging of native and programmed defects on a full-field mask,” Proc. SPIE 7636, 76361A (2010).
ALS Science Highlight #213