Radiant energy – Irradiation of objects or material – Irradiation of semiconductor devices
Reexamination Certificate
1999-10-14
2003-06-03
Berman, Jack (Department: 2881)
Radiant energy
Irradiation of objects or material
Irradiation of semiconductor devices
C250S492220, C250S492300, C430S005000, C430S030000, C430S296000
Reexamination Certificate
active
06573515
ABSTRACT:
FIELD OF THE INVENTION
This invention pertains to microlithography apparatus and methods as used, for example, in the manufacture of semiconductor integrated circuits and displays. More specifically, the invention pertains to such apparatus and methods that employ a charged particle beam (e.g., electron beam or ion beam) as an energy beam for performing projection-transfer of a pattern, defined by a segmented reticle, onto a sensitive substrate such as a semiconductor wafer. Yet more specifically, the invention pertains to such apparatus and methods exhibiting greater pattern-transfer accuracy whenever a segmented stencil reticle is used.
BACKGROUND OF THE INVENTION
The known prior art is summarized below in the context of electron-beam microlithographic systems as representative charged-particle-beam (CPB) microlithographic systems. Whereas electron-beam microlithography potentially is more accurate for performing pattern transfer than optical microlithography (including optical microlithography performed using ultraviolet light), conventional experience with electron-beam microlithography has been plagued by, among various problems, low “throughput” (number of wafers that can be exposed per unit time).
Various approaches have been investigated to increase throughput. One approach is “cell projection” which is conventionally used whenever the pattern comprises a small basic unit portion (measuring, e.g., (5 &mgr;m)
2
on the wafer) that is repeated a large number of times in the pattern, such as a pattern for a memory chip in which the unit portion is a single memory cell. An image of the single unit portion is transferred to the wafer per exposure dose (“shot”); hence, many shots are required to transfer all the unit portions in the pattern. The same unit portion can be defined in multiple regions on the reticle. Unfortunately, circuit patterns such as memory chips include portions that are not repeated, and transfer of such portions requires application of another technique such as “variable-shaped beam” lithographic writing. The need to use multiple techniques to achieve transfer of the entire pattern reduces throughput. In practice, the throughput achieved with cell projection is typically less than ten.
Another conventional approach (termed “full-field exposure”), in which a reticle defining an entire pattern is transferred in one shot to a corresponding die on the wafer, offers prospects of very high throughput. Unfortunately, however, the very large exposure field required necessitates using electron optics having a correspondingly extremely large field. Such large electron-optical systems are prohibitively costly and bulky. Also, in such large fields, the peripheral regions of the field as projected tend to exhibit large aberrations that have been impossible to date to adequately correct. Furthermore, a reticle for use with full-field exposure is extremely difficult to fabricate.
In response to the problems with the full-field exposure technique, the “divided-pattern projection-exposure” technique was proposed. In the divided-pattern technique, a reticle (mounted on a movable reticle stage) defines the entire pattern to be transferred to a corresponding die on the wafer (mounted on a movable wafer stage). Rather tan being exposed entirely in one shot, the pattern field as defined on the reticle is divided into multiple “exposure units” (e.g., “subfields”) that are individually and sequentially illuminated. Illumination is performed by an “illumination beam” passing through an “illumination-optical system” located upstream of the reticle. An image of the illuminated exposure unit passes (as a “patterned beam”) through a “projection-optical system” located between the reticle and the wafer. The projection-optical system has a field that is much smaller than the field of the entire pattern as defined on the reticle. The image that is projected by the projection-optical system onto a corresponding region of the wafer is “demagnified” or “reduced,” by which is meant that the image is smaller (usually by an integer “demagnification ratio” such as 1/4 or 1/5) than the corresponding exposure unit on the reticle.
Systems that perform divided-pattern projection-exposure achieve lower throughput than the full-field exposure technique but substantially higher throughput than the cell projection technique. For details on divided-pattern projection-exposure, see, e.g., U.S. Pat. No. 5,260,151, incorporated herein by reference, and Japan Kokai Published Patent Document No. Hei 8-186070.
In divided-pattern projection-exposure, two basic types of reticles, termed “stencil” and “membrane” reticles, are currently used. Stencil reticles are usually configured as “scattering-stencil” reticles in which pattern features are defined by corresponding voids (openings) extending through the thickness dimension of a silicon membrane approximately 1 to 5 &mgr;m thick. Charged particles in an illumination beam incident on an exposure unit of such a reticle pass through the voids without being scattered or absorbed by the reticle. In contrast, charged particles of the illumination beam incident on the membrane itself also pass through the membrane, but are scattered during such passage. To prevent such scattered particles from reaching the wafer, a “contrast aperture” is situated in the projection-optical system at or near the conjugate plane of the entrance pupil of the projection-optical system (which is also the Fourier plane of the reticle surface). Particles that are not scattered pass through an axial aperture defined by the contrast aperture, whereas scattered particles are blocked (absorbed) by the contrast aperture and thus prevented from propagating to the wafer. Particles of the beam passing through the axial aperture are not further scattered and form an image of the illuminated exposure unit of the reticle on the wafer.
In a stencil reticle, the feature-defining voids are termed “white” regions and surrounding membrane regions are termed “black’ regions. The white and black regions collectively define the pattern defined by the reticle. Certain features defined by a stencil reticle include so-called “island” regions that are defined by a black region surrounded by a white region. As readily can be surmised, an island (black) region cannot be situated within a surrounding white region in a stencil reticle because the island region would not have any physical support. Such a problem is referred to as the “stencil problem” or the “donut problem.”
To solve the donut problem, the exposure unit containing an island region is divided into two “complementary” exposure units in which the white region surrounding the island black region is divided in a manner providing (in each complementary exposure unit) physical support for the island black region. Each complementary exposure unit is individually exposed onto the same region on the wafer. Such double exposure on the same region of the wafer ideally results in the corresponding two images being in accurate registration with each other to form the complete island region. Unfortunately, the need to perform two exposures on at least some of the exposure units of the reticle correspondingly decreases throughput.
Another solution to the donut problem encountered with stencil reticles is to use instead a “scattering-membrane reticle” that is not subject to the donut problem. In a scattering-membrane reticle, a patterned layer of a high-scattering material is layered on a membrane made of a low-scattering material. The high-scattering material (e.g., chrome or tungsten approximately 10 to 200 nm thick) causes a high degree of scattering to particles of an illumination beam incident on an exposure unit of the reticle, even though such particles are transmitted by the membrane. The low-scattering material is typically a thin silicon membrane (approximately 100 nm thick) that transmits particles of an incident illumination beam while imparting relatively little scattering to the transmitted particles. Whenever a scattering-membrane reticle is used, hig
Berman Jack
Fernandez Kalimah
Klarquist & Sparkman, LLP
Nikon Corporation
LandOfFree
Charged-particle-beam projection-exposure apparatus and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Charged-particle-beam projection-exposure apparatus and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Charged-particle-beam projection-exposure apparatus and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3148643