Radiant energy – With charged particle beam deflection or focussing – Magnetic lens
Reexamination Certificate
2001-02-16
2003-09-30
Lee, John R. (Department: 2881)
Radiant energy
With charged particle beam deflection or focussing
Magnetic lens
C250S3960ML, C250S492200, C250S492220, C250S492230, C706S013000
Reexamination Certificate
active
06627900
ABSTRACT:
FIELD OF THE INVENTION
This invention pertains generally to arrays of components, such as an array of deflectors as used in a charged-particle-beam (CPB) optical system, that function cooperatively to achieve a desired performance of the system comprising the components. The invention also pertains to methods for exploring a “design space” to find an optimal combination of respective values of configurational and/or operational parameters of the components to achieve the desired performance. Such methods are especially useful in CPB microlithography systems that comprise one or more arrays of multiple deflectors. The deflectors must function cooperatively with each other to produce a desired imaging quality. For example, individual respective values of configurational and/or operational parameters for each deflector must be optimized in view of the parameters of the other deflectors to produce a combination of values of the parameters achieving the desired imaging quality.
BACKGROUND OF THE INVENTION
In view of the resolution limits of optical microlithography (microlithography performed using a light beam such as a beam of ultraviolet light), charged-particle-beam (CPB) microlithography has received considerable attention as a possible successor technology. The reasons are similar to the argument that electron microscopy achieves greater resolution than optical microscopy.
Examples of conventional CPB microlithography include (a) spot-beam exposure systems, (b) variable-shaped-beam exposure systems, and (c) block exposure systems. Each of these exposure systems offer prospects of much greater resolution than so-called “one-shot” optical microlithography systems, but are grossly inferior in terms of throughput. Specifically, in exposure systems (a) and (b), throughput is low because exposure is accomplished by tracing the pattern using a beam having an extremely small round or square spot diameter, respectively. It is immediately apparent that a complex pattern requires a large amount of time to “draw” line-by-line. Exposure system (c) was developed to achieve improved throughput over systems (a) and (b). In exposure system (c), throughput is improved because the pattern includes an array containing a large number (typically thousands) of individual repeat units, such as the memory cells on a memory chip. The repeat unit normally is very small, typically about 5 &mgr;m square on the substrate. The repeat unit is defined on a reticle and is exposed over and over again within a region on the substrate corresponding to the chip. As readily understood, considerable time is required to expose the array in each chip, which results in low throughput. Also, this technique is not used to transfer non-repeated portions of the chip pattern. Instead, the non-repeated portions usually are exposed by direct writing using a variable-shaped beam. This need to exploit multiple different techniques to expose each chip further compromises throughput. As a result, system (c) does not provide the throughput required for mass-production wafer fabrication.
A technique offering prospects of substantially greater throughput than the techniques summarized above involves exposing, in a single “shot,” a reticle defining the entire pattern to be transferred to a chip or defining a pattern for multiple chips. The reticle is exposed with “demagnification,” by which is meant that the reticle image is smaller (usually by an integer factor termed a “demagnification factor”) than the corresponding pattern as defined on the reticle. Whereas the throughput potentially achievable using this technique is at least as good as currently achievable using optical microlithography, this technique unfortunately has several serious problems. One problem is the current impossibility of fabricating a reticle configured to be exposed in a single shot of a charged particle beam. Another problem is the current impossibility of adequately correcting off-axis aberrations, especially in peripheral regions of the large image produced by the charged particle beam.
A more recently considered approach is termed “divided-reticle” pattern transfer, which involves dividing the pattern, as defined on the reticle, into multiple individual exposure units usually termed “subfields.” Each subfield is exposed individually onto a respective region on the wafer. The subfield images are transferred to the wafer so that, after exposing all the subfields, the subfield images are “stitched” together in a contiguous manner to form the entire chip pattern. Divided-reticle pattern transfer allows exposures to be made over an optically wide field with much better resolution and accuracy than could be obtained by exposing the entire reticle in one shot. Although divided-reticle pattern transfer does not yet achieve the same throughput as optical microlithography, the throughput nevertheless is much better than obtainable using the other CPB microlithography techniques summarized above.
Certain aspects of divided-reticle pattern transfer are shown in
FIGS. 5 and 6
.
FIG. 5
depicts a wafer on which multiple chips have been exposed. As exposed, each chip comprises multiple “stripes,” and each stripe comprises multiple subfields arranged in rows. This same divided arrangement of stripes and subfields is used to define the pattern on the reticle.
FIG. 6
depicts an actual exposure. For exposure, the reticle and wafer are mounted on respective stages (not shown but well understood in the art) configured to move the reticle and wafer horizontally (in the figure) as required for exposure. During exposure of a stripe (a portion of which is shown), the reticle stage and wafer stage both move along the longitudinal line of the respective stripes. Movements of the reticle and wafer are at constant respective velocities (but in opposite directions) in accordance with the demagnification ratio. Meanwhile, the charged particle beam incident on the reticle (the beam incident on the reticle is termed the “illumination beam” and passes through an “illumination-optical system” to the reticle) illuminates the subfields on the reticle row-by-row and subfield-by-subfield within each row (the rows extend perpendicularly to the movement directions of the reticle and wafer). As each subfield is illuminated in this manner, the portion of the illumination beam passing through the respective subfield (now termed the “patterned beam” or “imaging beam”) passes through a projection-optical system to the wafer.
During exposure of a stripe, to expose the rows and subfields within each row of the stripe in a sequential manner, the illumination beam is deflected at right angles to the movement direction of the reticle stage and the patterned beam is deflected at right angles to the movement direction of the wafer stage. After completing exposure of each row, the illumination beam is deflected in the opposite direction, as shown in
FIG. 6
, to expose the subfields in the next row of the stripe. This exposure technique reduces extraneous deflections of the beam and improves throughput.
Normally, on the reticle, each subfield is surrounded by “struts” configured as a lattice separating the subfields from one another and providing the reticle with considerable rigidity and mechanical strength. The struts also ensure that, in each shot, only the respective subfield is illuminated and exposed onto the wafer.
In a divided-reticle CPB projection-microlithography apparatus, throughput is improved by performing exposures using a relatively high beam current in the illumination-optical system and projection-optical system. To perform exposures with a high beam current, it usually is necessary to enlarge the area of the reticle being exposed per shot, and to accelerate the beam with a high beam-acceleration voltage, so as to reduce image blur due to the Coulomb effect.
Throughput also can be increased by performing exposures at the widest practical range of beam deflection. Enlarging the deflection field in this manner results in a corresponding increase in the width of the stripes, which decre
Klarquist & Sparkman, LLP
Lee John R.
Nikon Corporation
Souw Bernard
LandOfFree
Methods, based on a genetic algorithm, for configuring... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Methods, based on a genetic algorithm, for configuring..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Methods, based on a genetic algorithm, for configuring... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3009101