JP2009531855A - Optical system for generating high current density patterned charged particle beams - Google Patents

Abstract

A method and a charged particle optical apparatus for generating a shaped beam with a high current density.
The charged particle beam lithography apparatus and / or method includes a beam pattern limited aperture (212) for generating a high current density shaped beam (222) without the need for a plurality of beam shaping apertures, and a charged particle beam (222) on a wafer (221). And a blanking deflector for deflecting the charged particle beam 222 without requiring a crossover in the middle between the electron source 201 and the wafer 221.
[Selection] Figure 2A

Description

  The present invention relates to the field of charged particle optics, and more particularly to a method and system for generating a high current density shaped electron beam.

The use of electron beams to form semiconductor masks, reticles, and wafer writing patterns is an established technique. The various drawing strategies used can be characterized by several key parameters.
[Beam positioning strategy]

  There are two main approaches for positioning the electron beam for resist exposure during the writing process.

  (A) Raster scanning. The beam is moved on a fixed two-dimensional lattice pattern. This method has the advantage that the scanning electronics is usually simple, but the disadvantage is that the beam may spend a lot of time moving through areas that do not require exposure. Moreover, advanced gray scale and / or multiple pass scanning may be required to achieve very accurate pattern edge placement.

  (B) Vector scanning. The beam is moved directly in two dimensions to the area to be drawn. This method has the advantage of reducing time on areas that do not need to be exposed, but has the disadvantage that the deflection electronics are complex and expensive. By using the beam placement function on the 2D address grid which is much smaller than the beam size, accurate pattern edge placement is also facilitated.

Each approach is advantageous in certain circumstances, and the optimal choice depends on the critical dimension of the pattern, the pattern density (% of the area to be written), and the profile of the beam current distribution.
[Beam shape control]

  There are two well known approaches to shaping the electron beam used to expose the resist on the substrate.

(A) A Gaussian beam is characterized by a maximum current density (usually greater than 2000 A / cm ² ) which, in such a system, focuses the electron source image on the substrate surface, thereby increasing the high brightness of the electron source It is for use. The main drawback of the Gaussian beam is the length of the tail of the current extending far beyond the center beam diameter, and only 50% of the beam current at the substrate falls within the full width at half maximum of the two-dimensional Gaussian distribution.

(B) The shaped beam is usually formed by an electro-optic column having several intermediate shaped apertures, combined with an additional deflector and lens, to form an aperture (group) focused image on the substrate surface. Such systems typically have a lower beam current density (eg, 20-50 A / cm ² ) than a Gaussian beam. The advantage of such a system is a reduction in the tail of the current outside the desired beam shape, making the patterning less sensitive to process variations. Another advantage is that a large number of pixels can be written simultaneously, since the area of the variable shaped beam can be large compared to a single pixel.

  There is a need in the semiconductor industry to achieve maximum patterning throughput in both mask and reticle writing, and sometimes direct wafer writing. Any of the two approaches to beam positioning can be combined with either of the two approaches to beam shaping, but none of these four combinations can fully meet the needs of the semiconductor industry. As simple as possible to ensure the ability to pattern very small CDs with edge placement accuracy of less than CD / 8 and sufficient system reliability, long mean time between failures (MTBF), and short mean repair time (MTTR) Clearly, there is a need for an electronic drawing system with high throughput (drawing reticles with at least several wafers in an hour or less than an hour) in combination with a new electro-optic design.

  A charged particle optical device for generating a high current density shaped beam is disclosed. This device utilizes the charged particle optical column design commonly used to generate high current density Gaussian beams with the addition of a beam pattern limited aperture that can be customized for insertion at various positions of the column. An example of a charged particle optical column design utilizes two lenses, and the charged particle source emits a diffused beam of charged particles, and a first particle forms a substantially parallel charged particle beam. The second lens then focuses a nearly parallel charged particle beam onto the substrate surface with a general Gaussian current distribution with a high current density in the center and a long tail extending outward in all directions from the beam center. Let In this example, the beam pattern limited aperture can be positioned between two lenses. Based on the design requirements of the pattern to be drawn, the beam shape on the substrate is determined. The shape of the beam pattern limited aperture (PBDA) is created in the multi-step method disclosed herein. The PBDA shape must meet two requirements: (1) it should transmit most of the charged particles in the beam that enters the predetermined beam shape, and (2) the charge in the beam that deviates from the predetermined beam shape. The penetration of most of the particles should be shielded.

  A charged particle optical system utilizing the present invention may include a number of additional components as follows.

  Beam blanker—used to turn a beam on and off by deflecting the beam to a blanking aperture. In the embodiments of the invention described herein, the PBDA also functions as a blanking aperture.

  Deflector—used to move a beam over the substrate surface to pattern an area. In this embodiment, a double deflection main deflector moves the beam to the center of a 2 μm square subfield. Within each subfield, a subfield deflector composed of a single octupole deflects the beam.

  Moving lens—To minimize off-axis aberrations in the shaped beam, the effective optical axis of the second lens is displaced off-axis to match the deflection of the beam by the main field deflector.

  Astigmatism corrector—used to correct optical lens imperfections caused by mechanical defects or positioning errors in various elements.

  The design method of the beam pattern limited aperture is started by data related to the pattern to be drawn (IC dimensions and layout on the wafer, IC critical dimensions, alignment mark design, etc.), and this data is used for the charged particle beam column. In combination with the optical properties, an optimal shaped beam size is determined that allows pattern writing with maximum efficiency (ie, maximum throughput). The PBDA design is first created as an ideal shape and then modified to allow production. After a PBDA design proposal, tests were initially performed using the same procedure used to create the design, and usually using charged particle design software using electron beam tracing, The actual charged particles are simulated under the influence of the electric and magnetic fields formed by the pole pieces.

  The shaped beam produced by this column is characterized by an improved sharpness of the edge of the current profile compared to a Gaussian beam and a current distribution close to normal at the resist exposure (the latter for lithographic applications). Highly desirable). Advantages of the apparatus include the ability to generate a shaped beam without further complicating the shaping devices, deflectors, and lenses that are typically present in variable shaped beam barrels. In addition, a current density close to that of a Gaussian beam system is achieved, greatly reducing resist exposure time and increasing writing throughput in lithography applications.

  A method of designing a beam pattern limited aperture (PBDA) will be described. The heart of this method involves performing electron beam tracing to determine which electron beam of the charged particle beam contributes to the desired beam profile at the substrate at a number of locations on the substrate surface. A design process of a beam pattern limited aperture that transmits an electron beam contributing to a desired beam profile and shields an electron beam that deviates from the desired profile is performed. Furthermore, another innovative aspect of the present invention is described in the following paragraphs.

  Blanking system—The apparatus described herein employs a unique blanking system that does not require the use of an intermediate crossover between the electron source and the wafer. A double deflection blanker is used to project the effective blanking surface back to the virtual source position. This is advantageous because the lack of an intermediate crossover substantially reduces the space charge beam diffusion caused by electron-electron interactions. Another advantage of the double deflection blanker geometry is the ability to blank the beam over a very wide range of beam sizes, and in prior art designs, the (single) blanker is positioned at the crossover and blanked. It is impossible to achieve the wide range of beam sizes (less than 30 nm to more than 120 nm) that is possible with the present invention, because of this wide size range, This is because it is necessary to change the magnification of the lens barrel by moving the lens to various (spaced) positions. Another novel aspect of the blanker system is the use of a square beam trimming aperture above the blanker to reduce the beam size and beam shape to a square cross section. This has the advantage of shaping the beam only slightly larger than PBDA (which also serves as a blanking aperture) to maximize the achievable blanking speed. In addition, the square beam makes the deposition current on the wafer more uniform within the shaped beam by irradiating the entire PBDA opening uniformly when sweeping the PBDA.

  Main Deflector--The present invention provides a unique main deflector design that best suits the requirement that the off-axis deflection distance of the patterned beam is much longer in one direction (generally greater than 25 μm) than in the other direction (approximately 1 μm). adopt. In the deflector design, a large number of separation electrodes (22 in the embodiment of the present specification) are used, but only four types of drive signals are required. The arrangement of 22 deflector electrodes simulates the electric field generated by a set of parallel plates that is more uniform than is possible with prior art octupole designs. A more uniform electric field reduces the deflection aberrations induced in the beam and allows for sharper edge profiles in the patterned beam of the invention described herein. Prior art deflectors employ a symmetric octupole design that increases aberrations for the large deflections required here.

  Main lens design—To form a shaped high current density beam over a wide range of off-axis positions (at least 25 μm) on the wafer surface, the present invention ensures that the beam always appears on the optical axis of the main lens. A main lens structure is adopted in which the effective axis of the lens can be moved in synchronization with the deflection of the beam. The lens structure of the present invention utilizes two sets of octupole electrodes integrated into the lens structure to add a small lateral dipole field to the lens electric field generally along the axis. Such a dipole field allows the beam in the center of the lens to offset the axial field by more than 25 μm. Thus, the beam is always subject to a focusing effect that is close to that seen on the axis. All off-axis aberrations are essentially eliminated both in geometric aberrations (coma, astigmatism, field curvature, distortion) and chromatic aberrations (magnification change), because the sharpness of the patterned beam edge is increased This is an advantage. Prior art systems utilizing "moving lenses" required a much more complex electrode design than that utilized here.

  Control system-The control system for multi-lens optical instruments is common to all lens barrels and adjusts many optical elements that can be controlled with a single control, while other optical elements are Requires separate control.

  Pattern Data Path—The data path of the present invention utilizes a number of features required from the need to coordinate the pattern formation of a number of lens barrels. In order to maintain the pattern quality on the wafer, it is necessary to combine the drawing patterns of all the lens barrels. In addition, in order to maximize drawing efficiency, various patterned high current density beams may be generated (if necessary, the size varies for each column). Therefore, one lens barrel may draw a 30 nm feature, while another lens barrel may draw a bonding pad including a 2 μm square subfield using a 120 nm shaped beam.

  Proximity Effect Correction Method—In order to correct the proximity effect, the present invention employs a method of changing the beam dose for each subfield to maximize the degree of processing freedom during resist development. An iterative process is used to determine the section of the region to be drawn in each subfield, modify the dose in adjacent subfields, and correct backscattered molecules (BSE) that contribute to the total resist exposure.

  The present invention will be described in detail using examples in the field of electron beam lithography as specific examples. However, many other fields of use are conceivable, as outlined shortly thereafter.

  In a scanning electron microscope, in order to maximize the beam current density, the imaging time and / or the signal-to-noise ratio of the image are usually maximized by using an object close to a Gaussian beam. A disadvantage of using a Gaussian beam in a microscope is the long current tail extending from the center of the beam, which tends to reduce the achievable image contrast. The present invention can also be used in scanning electron microscopes to narrow the range of such current tails and increase image contrast. Similar considerations apply to many types of scanning electron beam imaging and analysis tools, such as scanning Auger electron microscopes, scanning electron microscopes, scanning transmission electron microscopes, and the like.

  The present invention also has potential applications in the field of semiconductor measurement and inspection. In such applications, close to a Gaussian beam is used to maximize the measurement and inspection throughput by minimizing the time required to measure or inspect features on the semiconductor wafer or mask and reticle. Eliminating long current tails with a Gaussian distribution improves imaging contrast in such systems. Conversely, if the contrast is kept constant, the present invention allows for rapid pixel data acquisition, which improves throughput.

  The beam pattern limited aperture of the present invention may also be used in other types of particle beam systems that utilize, for example, ions. One example is a focused ion beam system for maskless ion implantation. In such a system, the ion beam consists of the desired implanted ions (boron, arsenic, phosphorus, etc.) and the reduction of the irrelevant current tail reduces the implantation of ions outside the region where doping is required. Another example is a focused ion beam direct lithography tool that uses an ion beam to expose the resist, as in the electron beam direct lithography system. Irrelevant current reduction increases the processing freedom of resist development by increasing the contrast in the lithography process. Yet another example is a scanning secondary ion mass spectrometry (SIMS) system that irradiates a sample surface with a focused ion beam to induce the release of secondary ions that are characteristic of the chemical composition of the material. Because the ion tail of the primary ion beam is significantly reduced, secondary ions are generated almost entirely from the region of interest and are rarely generated outside this region, so irrelevant ion reduction can be achieved with SIMS images and mass spectra. Increase contrast and resolution.

  FIG. 1 shows a multi-step method for designing an electron optical column that utilizes a beam pattern limited aperture for use in generating a high current density shaped electron beam. In this example, a square beam is desired on the wafer, but a wide range of beam shapes can be realized by appropriate selection of the beam limited aperture pattern.

  In block 102, an initial value for the integrated circuit (IC) pattern to be drawn, including the critical dimension (CD) of the pattern, the XY dimension of the IC, the XY layout of the IC on the wafer, and other data as required. Define the data.

In block 103, the desired writing throughput (usually wafer / hour), resist sensitivity to the writing beam (usually μC / cm ² ), desired writing beam energy on the wafer, and drawing overhead (wafer transfer time, alignment). Initial data regarding system operating parameters, including time etc.) and other parameters as needed.

In block 104, the optimal patterned beam shape and size are determined along with the required beam current density from the pattern and drawing designations in blocks 102 and 103. For example, if a 45 nm pattern CD is specified in block 102, a 40 nm square beam profile may be appropriate. If the resist sensitivity is 5 μC / cm ² , a beam current density of 3000 A / cm ² may be required to achieve the desired writing throughput.

  Next, in block 106, an electro-optic design calculation is typically performed to determine the lens barrel design, including lens electrode aperture, thickness, position, and voltage, and the final patterned beam determined in block 104. The diameter of the circular beam larger than the size of the wafer is determined. For example, if a 40 nm square beam is desired, a circular beam with a diameter of √2 × 40 nm≈56 nm is required, and this beam diameter then causes the block 110 to produce a 40 nm × 40 nm square beam without rounding corners. You can get it. 2A through 3M show a typical electronic lens barrel design created with the aid of such a process. Alternatively, the beam pattern limited aperture may be designed to input parameters of an existing barrel and then start at block 108.

  Block 108 includes a series of electro-optic design calculations utilizing the lens barrel design created in block 106, where the electron beam (X, Y) position in the beam limiting aperture 212 (see FIG. 2A) is the wafer surface 221 (FIG. 2A)) and its end point (X, Y). In general, five sets of electron beams are generated at different (X, Y) positions on the wafer: 1) on-axis (ie, scan center), 2) ± 1/4 width of scan, and 3) of scan Used at ± 1/2 width (ie, both ends of the farthest off-axis scan). This data then indicates which electron beam in each set falls within the desired patterned high current density beam profile and which electron beam in each set deviates from the desired pattern individually (X, Y) on the wafer. Used to determine for each position. 4A-4C illustrate the separation of trajectory data into two groups inside and outside the desired pattern on the wafer surface 221. In such a set of electron beams, the exact same electron beam does not need to correspond to each of the five types of positions on the wafer, that is, a light beam that has passed a specific position in the beam limiting aperture has a beam on the axis. It should be noted that the inside of the desired beam profile may be reached when positioned and the outside of the desired beam profile may be reached when the beam is deflected by ± 1/2 width of the scan. In general, FIGS. 4A-4C show that the beam on the wafer is circular in all cases shown, and that there is a small variation in the position of the individual electron beams, which is a careful column in block 106. The design results, in particular, the design of the main field deflector that introduces minimal beam aberrations, and the use of a movable main lens that substantially eliminates off-axis aberrations throughout the main field scan (± 25 μm in this example). Is the result of

  At block 110, the next step of finding the intersection of the five sets of electron beams from block 108 is performed, which is within the desired patterned high current density beam at all five wafer locations. It corresponds to the contained electron beam. Typically, this set of electron beams is about 10-15% smaller than any of the first five sets of electron beams corresponding to each of the five individual wafer positions in block 108. This process is necessary because the electron beam 222 impinges on the beam pattern limiting aperture before it is deflected by the main field deflectors 213 and 214, so that exactly the same set of irradiation rays is applied to the wafer surface by the beam pattern limiting aperture 212. This is because it is transmitted to 221 and becomes all part of the beam 222 on the wafer surface 221.

  Block 112 uses the trajectory data from block 110 to transmit all the electron beams that contribute to the desired beam profile simultaneously at all five positions on the wafer (ie, across the scan), An ideal (i.e., physically unrealizable) aperture design is defined for the purpose of shielding all electron beams reaching outside the desired profile. FIG. 6A shows the resulting electron beam to be transmitted in the beam limiting aperture 212. FIG. 6B shows the resulting electron beam to be shielded in the beam limiting aperture 212.

  At block 114, a final change is made to the beam pattern limited aperture to allow a practical aperture 212 design as shown in FIGS. 7A-8A. The resulting beam pattern limited aperture must meet two requirements: 1) it should transmit most of the charged particles in the beam contained within the desired beam shape, and 2) from the desired beam shape. The transmission of most of the charged particles in the off-beam should be shielded.

  Next, at block 116, the accuracy of the optics and aperture design is tested by tracing a number of electron beams through the electron column of FIG. 2A using the aperture of FIG. 8A.

  Block 118 combines the multiple electron beams generated in block 116 (typically over 30,000) to obtain a beam current density profile as shown in FIGS. 10A-12.

  Finally, at block 120, a graph of the beam current profile can be generated and compared with the current profile of the corresponding Gaussian beam. The improved edge sharpness with the patterned beam generated by an electron column utilizing the present invention can be seen in FIGS. 13-15 compared to the Gaussian profile of FIG.

  FIG. 2A is a cross-sectional view of a typical electron column that can utilize the present invention to generate a high current density patterned electron beam. FIG. 2A is expanded along the Y axis so that the beam 222 and various electrodes can be clearly seen. This barrel design is representative of that created in block 106 of FIG. The illustrated components are: an electronic source chip 201, an extraction electrode 202, a first source lens electrode 203, a beam limiting aperture (BLA) 204, a second source lens electrode 205, a gun mounting plate 206, an upper alignment deflector / non- Point corrector 207, acceleration assembly 209, electron beam 222, lower alignment deflector 208, beam trimming aperture (BTA) 276, upper blanker 277, lower blanker 278, optical system mounting plate 210, beam limited aperture mounting section 211, beam Pattern limited aperture (PBDA) 212, upper main field deflector 213, lower main field deflector 214, subfield deflector / astigmatism corrector 215, focus 1 electrode assembly 216, focus 2 electrode assembly 217, field free tube 18 includes a detector assembly 219, the substrate 221 in the pattern formation by lithography by voltage contrast plate 220 and electron beam 222,. The combination of field free tube 218, detector assembly 219, and voltage contrast plate 220 is called detector optics and is used to image alignment marks on the substrate in the case of electron beam lithography. Please note that. The combination of the focus 1 electrode assembly 216, the focus 2 electrode assembly 217, and the field free tube 218 is called the main lens.

  Electrons are emitted from the source chip 201 under the influence of a high electric field induced by a voltage difference (usually 2500 to 3500 V) between the source chip 201 and the extraction electrode 202. Some of these electrons, near the axis of symmetry of the optical system, pass through the hole in electrode 202 and move toward first source lens electrode 203. The beam limiting aperture 204 is attached within the aperture of the electrode 203 so that only electrons within a small angle (usually about a half angle of about 2.0 °) can be sent to the lens barrel. Normally, a voltage of 430 to 640 V (as compared to the source chip 201 which is 0 V) is applied to both the electrode 203 and the beam limiting aperture 204, and this potential is approximately the same applied to the second focusing electrode 205. In combination with 510V, the beam 222 is focused to form a decelerating lens into a parallel beam that passes through the gun mounting plate 206. Upper alignment deflector / astigmatism corrector 207 and lower alignment deflector 208 are used to direct electron beam 222 through beam trimming aperture 276 parallel to the optical axis (Z-axis). The acceleration region 209 between the upper alignment deflector / astigmatism corrector 207 and the lower alignment deflector 208 raises the beam energy from 510 eV to 5000 eV. The beam 222 then passes through the upper blanker 277 and the lower blanker 278. Some of the electron beams in the electron beam 222 are stopped by the beam pattern limiting aperture 212 instructed by the beam limiting aperture mounting portion 211, and the other electron beams pass through the main field deflectors 213 and 214, the sub-beams. The field deflector / astigmatizer 215 is reached and then enters the main lens. The main lens focuses the beam 222 onto the substrate surface 221 (a detailed description of a similar electron column design is hereby incorporated by reference, US Pat. No. 6,734,428 B2). As described in).

  The illustrated barrel design is for illustrative purposes only, and the beam pattern limited aperture generated by the method of the present invention may be employed in many barrel designs known to those skilled in the art.

  FIG. 2B is a cross-sectional view of the bottom of a typical electron column that can employ the present invention to produce a high current density patterned electron beam. The illustrated components are a beam limiting aperture attachment 211, a beam pattern limiting aperture 212, an upper main field deflector 213, a lower main field deflector 214, a subfield deflector / astigmatism corrector 215, and a focus 1 electrode assembly 216. , Focus 1 support electrode 230, focus 1 octupole electrodes 231 to 238, focus 2 electrode assembly 217, focus 2 support electrode 240, focus 2 octupole electrodes 241 to 248, field free tube 218, detector assembly 219, voltage Contrast plate 220 and substrate 221 being patterned using lithography with electron beam 222 impinging on substrate surface 221 at location 250 are included.

  The electron beam shown in each figure is SIMION 3D, ver. Calculations were performed using 6.0 (a charged particle beam tracking program developed by Dvid Dahl, Idaho National Engineering and Environmental Research Laboratory).

  FIG. 3A is a pair of diagrams of electron beams that are separated from the source chip 201. The initial distribution of the electron beams is “layered”, that is, the electron beams are spread uniformly from the chip 201 without crossing each other. It has a distribution. FIG. 4A is a side cross-sectional view of the source chip 201, the extraction electrode 202, the first source lens electrode 203, the beam limiting aperture 204, and the beam 222, at 30 ° that appears from the source chip 201. The half angle of the open beam is shown. An axial section of the beam 222 is extracted at position 301. The optical axis is parallel to the Z axis 310 and orthogonal to the Y axis 320.

  The cross section in the axial direction of the beam 222 at the position 301 is shown in FIG. Both the X axis 319 and the Y axis 320 are orthogonal to the optical axis 310. It can be confirmed that each of the electron beam sections 302 is uniformly separated on the grid corresponding to the X axis 319 and the Y axis 320. The electron beam of the beam 222 retains this laminar flow property almost all the way to the substrate surface 221.

  The design described herein for the beam pattern limiting aperture 212 relies on the assumption that each electron beam represents a distinct amount of current. The calculation of this current is as follows.

I _S = radial solid angle used to illuminate the beam pattern limited aperture 212 (typically a half angle of 0.4 ° to 1.5 °, in this example 0.8 ° —typically, I _S is 100 μA / sr Source angle intensity at ˜500 μA / sr δ = angle increment between electron beam 302 along X axis 319 and Y axis 320 (generally 0.04 ° to 0.15 °, in this example δ = 0) .08 °)
ω = solid angle determined for each electron beam 302 with respect to δ = 0.08 ° = [δ (π / 180 °)] ² = 1.95 × 10 ⁻⁶ sr
I _{ray 302} = I _s ω (500 μA / sr) (1.95 × 10 ⁻⁶ sr) = 0.98 nA for each electron beam 302

This calculation makes an implicit assumption that the angular intensity is uniform over the range of radiation angles used to generate the square beam (including the electron beam 306 of FIG. 3G) that illuminates the beam pattern limiting aperture 212. Including. In the example of a Schottky field emitter, this assumption is valid because the angular intensity is generally very uniform in the central (ie, on-axis) portion of the angular emission distribution. This assumption is invalid electronic source, a method of designing a patterned beam-defining aperture described herein can be modified by putting the value of different I _Ray302 corresponding to the initial angle of each electron beam 302 at the source chip 201 into account.

  FIG. 3B is a pair of diagrams showing electron beams in the upper alignment deflector / astigmatism corrector 207. FIG. 4A is a side sectional view of the gun mounting plate 206, the upper alignment deflector / astigmatism corrector 207, and the beam 222.

  An axial section of the beam 222 at the position 303 is shown in FIG. Beam 222 is centered within upper alignment deflector / astigmatism 207. Eight octupole electrodes 260-267 of the upper alignment deflector / astigmatism corrector 207 are shown. A voltage may be applied to the eight electrodes 260-267 to generate a rotatable dipole field and deflect the beam 222. In addition, a voltage is applied to the eight electrodes 260 to 267 to generate a rotatable quadrupole field and perform astigmatism correction of the beam 222 in the upper barrel. The electron beam intercept 304 corresponds to electrons leaving the source chip 201 with average energy. Each of the electron beam segments 304 is uniformly separated on a grid substantially corresponding to the X-axis 319 and the Y-axis 320 while maintaining almost the same relative positions as the corresponding electron beam segments 302 of FIG. 3A. Can be confirmed.

  FIG. 3C is a pair of diagrams showing electron beams in the lower alignment deflector 208. FIG. 4A is a side sectional view of the lower alignment deflector 208, the beam trimming aperture 276, the upper blanker 277, and the beam 222. An axial section of the beam 222 at position 398 is shown in FIG. The beam 222 is centered within the lower alignment deflector 208. The eight octupole electrodes 268-275 of the lower alignment deflector / astigmatism corrector 208 are shown. A voltage may be applied to the eight electrodes 268-275 to generate a rotatable dipole field and deflect the beam 222. The electron beam section 399 corresponds to electrons leaving the source chip 201 with average energy. Each of the electron beam segments 399 is uniformly separated on a grid substantially corresponding to the X-axis 319 and the Y-axis 320 while maintaining substantially the same relative position as the corresponding electron beam segment 302 of FIG. 3A. Can be confirmed.

  FIG. 3D is a pair of diagrams showing the electron beam immediately above the beam trimming aperture 276, and shows that the initial layered distribution of angles shown in FIGS. 3A to 3B is substantially held at a position further advanced through the lens barrel. FIG. 5A shows the beam 222 immediately above the beam trimming aperture 276 in the case of a 30 nm beam on the wafer 221, and the voltage applied to the first source lens electrode 203 and the beam limiting aperture 204 is generally about 640V. . In this case, the external electron beam of the beam 222 corresponds to a maximum half angle of 2.0 ° transmitted by the beam limiting aperture 204, and is a central square (half angle ± 0 in the source chip 201 along the X axis 319 and the Y axis 320). Only the inner electron beam (corresponding to an angle within .45 °) is transmitted further into the barrel. The electron circle outside the central square is shielded by the beam trimming aperture 276. The beam trimming aperture is square to form a square beam cross-section at the beam pattern limiting aperture 212, which, as will be described in more detail below, is suitable for the dose in the patterned beam at the wafer surface 221. Necessary for control.

  FIG. 3D (b) shows the beam 222 just above the beam trimming aperture 276 for a 120 nm beam on the wafer 221, and the voltage applied to the first source lens electrode 203 and the beam limiting aperture 204 is generally about 430V. The external electron beam of the beam 222 corresponds to the maximum half angle of 2.0 ° transmitted by the beam limiting aperture 204, and the central square (half angle ± 1.5 ° in the source chip 201 along the X axis 319 and the Y axis 320). Only the inner electron beam (corresponding to the inner angle) is transmitted further beyond the lens barrel.

  FIG. 3E is a pair of views showing the lens barrel near and inside the beam blanker. FIG. 6A shows a lower alignment deflector 208, a beam trimming aperture 276, an upper blanker 277, a lower blanker 278, an optical system mounting plate 210, a beam limiting aperture mounting portion 211, and a beam pattern limiting aperture 212. FIG. An axial section of the beam 222 at a position 395 at the center of the upper blanker 277 is shown in FIG. The electron beam intercept 394 corresponds to electrons leaving the source chip 201 with average energy. Each of the electron beam segments 394 is uniformly separated on a grid substantially corresponding to the X axis 319 and the Y axis 320 while maintaining substantially the same relative position as the corresponding electron beam segment 302 in FIG. 3A. Can be confirmed. To blank the beam 222, a voltage typically in the range of ± 1.7V + 5000V is applied to the blanker plates 280 and 282. The 3.4 V difference between plates 280 and 282 produces a transverse electric field parallel to Y axis 320 that bends electron beam 222 away from optical (Z) axis 310 as shown in FIG. 3E (a). The plate 281 is always maintained at a common mode voltage of 5000 V to ensure a uniform electric field in the gap between the plates 280 and 282.

  FIG. 3F shows an axial cross section of beam 222 at position 393 at the center of lower blanker 278. The electron beam intercept 392 corresponds to electrons leaving the source chip 201 with average energy. Each of the electron beam segments 392 is uniformly separated on a grid substantially corresponding to the X axis 319 and the Y axis 320 while maintaining substantially the same relative position as the corresponding electron beam segment 302 of FIG. 3A. Can be confirmed. To blank the beam 222, a voltage typically in the range of ± 1.54V + 5000V is applied to the blanker plates 283 and 285. The 3.08V difference between plates 283 and 285 creates a transverse electric field parallel to the Y axis 320 and in the opposite direction from the electric field in the upper blanker 277, causing the electron beam 222 as shown in FIG. 3E (a). Bend in the direction of the optical (Z) axis 310. The plate 284 is always maintained at a common mode voltage of 5000V to ensure a uniform electric field in the gap between the plates 283 and 285. The combined deflection effect of the upper 277 and lower 278 blankers causes the beam 222 to be deflected off-axis and toward the beam pattern limited aperture 212 in such a way that the beam still appears to come from a virtual source on the optical (Z) axis. Thus, conjugate blanking is ensured even if there is no actual intersection in the lens barrel between the source chip 201 and the wafer surface 221. The benefit of avoiding intermediate crossings is that the large diameter of the beam 222 traveling through the barrel greatly reduces the spread of the Coulomb (space charge) beam, thereby increasing the sharpness of the beam edge at the wafer surface 221. is there.

  FIG. 3G shows a pair of beams 222 directly above the beam limiting aperture 212, showing that the initial laminar distribution of the angles shown in FIGS. 3A-3B is substantially held at a position further advanced through the lens barrel. The upper square beam 305 corresponds to the blanked beam shown in FIG. 3E (a) and is deflected to a position completely away from the opening of the beam pattern limiting aperture 212. There is no permeation to the surface 221. The electron beam 306 is shown in the lower (non-blanking) square beam. These electron beams that pass through the apertures of the beam pattern limiting aperture 212 are transmitted to the wafer surface 221 to form a high current density square beam at location 250 in FIG. 2B.

  The purpose of the square beam trimming aperture 276 can be seen in FIG. 3G, and the beam size in the beam pattern limited aperture 212 needs to be as small as possible in order to minimize the blanking time. In addition, the beam has a square cross-section so that each position in the opening of the PBDA 212 is illuminated for the same amount of time as the beam sweeps the beam pattern limited aperture (PBDA) 212 (see FIG. 22). There is a need. This is possible only when the beam cross section is square. For any possible beam size on the wafer surface 221, the square illumination beam in the PBDA 212 is of the external beam diameter (corresponding to a half angle of 2 ° in the chip 201) shown in FIGS. 3A and 3B. Even if the size changes greatly, it is almost the size shown in FIGS. 3A and 3B. Without the beam trimming aperture 276, the beam diameter that irradiates the PBDA 212 shows a similar wide size range, and the blanking time is very long compared to the small beam size on the wafer surface 221.

  FIG. 5B is a close-up view of the beam 222 cross section of the beam pattern limited aperture (PBDA) 212 when the wafer surface 221 is a 30 nm square beam. For large beam sizes of 40 nm to 120 nm on the wafer 221, the square beam at the PBDA 212 is slightly smaller than FIG. (B), so for a 30 nm to 120 nm square beam, the beam trimming aperture 276 is PBDA 212. The square beam at is maintained at an approximately optimal (minimum) size for maximum bandwidth blanking.

  FIG. 3H is a diagram of the electron beam 330 at the center of the upper main field deflector 213. The initial layered distribution of angles shown in FIGS. 3A to 3B is substantially maintained at a position further advanced through the lens barrel. Note that the beam cross section holds an image of the opening of the beam pattern limiting aperture 212.

  The purpose of the main field deflector with 213 and 214 is as follows.

  (1) Correct a small (about ± 2 μm) wafer stage position error in both the X axis 319 and the Y axis 320.

  (2) Correcting a small (about ± 2 μm) mechanical error in the lens barrel position in both the X axis 319 and the Y axis 320.

  (3) Position the beam at the center of a specific subfield to be drawn along only the X axis 319 (deflection to at least ± 25 μm).

  All these requirements are relatively low bandwidth, but having a large (eg, ± 25 μm) off-axis deflection along the X-axis 319 ensures that the main field deflector does not induce aberrations in the beam. Is required. Since the deflection of the Y-axis 320 is small, a deflector design optimized for large X deflection and small Y deflection is used for both the upper 213 and lower 214 main field deflectors as shown. In general, the voltage of the upper main field deflector 213 for the maximum X-axis 319 deflection is as follows (for a common mode voltage of 5000V):

  In this example, electrodes 4001 through 4009 and electrodes 4012 through 4020 are two parallel, with electrodes 4010, 4011, 4021, and 4022 serving to maintain a uniform X-direction electric field required for minimum deflection aberrations. Play the role of a plate. Note that a set of electrostatic equipotential lines 4025 are illustrated every 0.5V from 4997.5V to 5002.5V, and note the high degree of uniformity in the electric field that minimizes deflection aberrations. In the case of pure Y-axis 320 deflection, electrodes 4009 to 4012 and 4001, 4020 to 4022 serve as two parallel plates, together with electrodes 4002 to 4008 and 4012 to 4019 that serve to maintain a uniform Y-direction electric field. Fulfill. Since the main field deflectors 213 and 214 are optimized for very large deflections parallel to the X axis 319, the design of the deflectors 213 and 214 is symmetric and the high X axis required for ± 25 μm X deflection. In order to maintain electric field uniformity at an electric field strength of 319, it has a large width along the Y axis. For the required ± 1 μm Y deflection, the required electric field strength is much lower, allowing a reduction in the degree of electric field uniformity. Aspects of parallel plate deflector design are well known to those skilled in the art.

Figure (b) illustrates a possible method for connecting a voltage to each of the electrodes 4001 to 4022. The electrodes 4004 to 4006 and 4015 to 4017 are omitted for the sake of simplicity. Each pair of adjacent electrodes is connected by a resistor 4035 of value R. Such resistors form four linear voltage dividers between the drive signals V _{+ X + Y} 4031, V _{−X + Y} 4032, V _−XY 4033, and V _{+ XY} 4034 as shown. Therefore, only four drive signals are required to control the 22 electrodes 4001 to 4022. The value R of the resistor 4035 needs to be kept low enough so that the RC time delay between the four drive signals 4031 to 4034 and the internal electrodes 4004 to 4006 and 4015 to 4017 is minimized, However, the value of R must not be so low that excessive power loss (V ² / R, where V = deflection voltage) occurs. In order to minimize the capacitance C to the connection with the electrodes 4001 to 4022, it is desirable to install the resistor 4035 as close as possible to the lens barrel assembly, possibly in the vacuum enclosure, however, in the vacuum enclosure Because of the lack of convective cooling, heat dissipation is difficult, and as a result, it is necessary to avoid excessive RC delay and at the same time make R as large as possible to minimize power loss in resistor R4035.

  FIG. 3I is a diagram of the electron beam 331 at the center of the lower main field deflector 214. The initial laminar distribution of the angles shown in FIGS. 3A to 3B is substantially held at a position further advanced through the lens barrel together with the image of the opening of the PBDA 212. The differential voltages of the electrodes 4101 to 4122 of the lower main field deflector 214 (ignoring the common mode voltage 5000V) are always equal in magnitude and opposite in polarity to the corresponding electrodes 4001 to 4022 of the upper main field deflector 213. It becomes. A set of electrostatic equipotential lines 4125 is shown every 0.5V from 4997.5V to 5002.5V. The X axis 319 corresponds to a large deflection direction of the beam 222 that is at least ± 25 μm from the normal optical axis on the wafer surface 221. The moving direction of the stage is parallel to the Y-axis 320 (+ Y and −Y directions alternately for each subsequent wafer scan—see FIG. 27B). Normally, the lengths of the electrodes 4101 to 4122 along the Z axis 310 are the same, and are equal to the lengths of the electrodes 4001 to 4022 of the upper main field deflector 213. Since the lengths of the upper and lower main field deflectors 213 and 214 are equal, the applied electrode voltages are equal in magnitude and opposite in polarity, the combined result is that the beam 222 is deflected off the optical axis 310, It returns to a state parallel to the optical (Z) axis 310.

  FIG. 3J is an axial cross-sectional view of the electron beam 332 at the center of the subfield deflector / astigmatism 215, with the initial laminar distribution of angles shown in FIGS. 3A-3B, along with an image of the opening of the PBDA 212. A state in which the lens barrel is held at a further advanced position is shown. The subfield deflector / astigmatism corrector 215 combines two functions,

  (1) As a deflector (ie, by excitation of a rotatable dipole), the subfield deflector / astigmatism 215 scans the beam 222 to ± 1 μm in both the X direction 319 and the Y direction 320, Used to cover 2 μm square subfield,

  (2) As an astigmatism corrector (ie, by excitation of a rotatable quadrupole), the subfield deflector / astigmatism corrector 215 corrects astigmatism induced by other elements of the barrel.

  The beam 222 is off-center in the subfield deflector / astigmatizer 215 due to the combined beam deflection in the upper and lower main field deflectors 213 and 214. The electrodes of the subfield deflector / astigmatism corrector 215 are arranged in a conventional octupole configuration as shown. Usually, the lengths of the octupole electrodes 223 to 230 along the Z axis 310 are the same, and the deflection and astigmatism correction voltages are as follows (the common mode voltage 5000 V is not shown).

  For beam deflections smaller than ± 1 μm, the table voltages increase and decrease linearly. In compound XY deflection, as is well known to those skilled in the art, the middle two rows of voltages are added linearly with an appropriate scaling factor corresponding to the desired X and Y deflection. The astigmatism corrector voltage tends to change in proportion to the square of off-axis deflection and is added to the respective XY deflection voltages of each electrode 223-230.

  FIG. 3K is an axial sectional view of the electron beam 333 inside the focus 1 electrode assembly 216. The initial laminar distribution of the angles shown in FIGS. 3A to 3B is substantially held at a position further advanced through the lens barrel together with the image of the opening of the PBDA 212. The focus 1 electrode assembly 216 is part of a main lens assembly that includes electrode assemblies 216, 217 and a field free tube 218. The main lens assembly is used to focus the beam 222 onto the wafer surface 221. Beam 222 is shown off-center within the focus 1 electrode assembly 216 (in the + X direction) for combined beam deflection at the upper and lower main field deflectors 213 and 214. The focus 1 electrode assembly 216 includes (1) a support electrode 230 having a large cylindrical ID (see FIG. 2C), and (2) a very short electrode length parallel to the optical (Z) axis and an ID smaller than the support electrode 230. And electrodes 231 to 238 forming conventional electrostatic octupoles. The dipole electrostatic excitation of the octupoles 231-238 is varied in proportion to the excitation of the upper 213 and lower 214 main field deflectors to maintain the effective axis of the focus 1 electrode assembly 216 concentric with the beam 222. .

  A typical electrode excitation voltage for a +25 μm deflection in the X direction is as follows (describes a 5000 V common mode voltage).

  The individual values shown in the above table were theoretically determined by the procedure described in FIG.

  FIG. 3L is an axial cross-sectional view of the electron beam 334 inside the focal point 2 electrode assembly 217 (scale is larger than FIG. 3K). The initial laminar distribution of angles shown in FIGS. 3A to 3B is held at a position further advanced through the lens barrel together with an image of the opening of the PBDA 212. The beam 222 is shown off-center (in the + X direction) within the focus 2 electrode assembly 217 for combined beam deflection at the upper and lower main field deflectors 213 and 214. By appropriately setting the dipole excitation voltage of the electrodes 241 to 248 in the focus 2 electrode assembly 217, the XY beam position leaving the focus 2 electrode assembly 217 is the same as the XY beam position entering the focus 1 electrode assembly 216. become. The focus 2 electrode assembly 217 includes (1) a support electrode 240 having a large cylindrical ID (see FIG. 2C), and (2) a very short electrode length parallel to the optical (Z) axis and an ID smaller than the support electrode 240. And electrodes 241 to 248 forming conventional electrostatic octupoles. The dipole electrostatic excitation of the octupoles 241-248 varies in proportion to the excitation of the upper and lower main field deflectors 213 and 214 to maintain the effective axis of the focal two-electrode assembly 217 concentric with the beam 222. Let

  The common mode voltage of electrodes 240-248 is determined by the on-axis focusing requirements for the specific square beam size desired. The general focusing voltage is as follows.

  The typical electrode excitation voltage for +25 μm deflection in the X direction is as follows (describing a common mode voltage of 5007.4 V for a 40 nm beam):

  The individual values shown in the above table were theoretically determined by the procedure described later in FIG. Each of the electron beam segments 334 is uniformly separated on a grid substantially corresponding to the X axis 319 and the Y axis 320 while maintaining substantially the same relative position as the corresponding electron beam segment 302 of FIG. 3A.

  FIG. 3M is an axial sectional view of the electron beam 335 inside the field free tube 218. The initial layered distribution of angles shown in FIGS. 3A-3B is substantially retained along with the image of the PBDA 212 opening. Beam 222 is shown off-center within field free tube 218 (in the + X direction) for combined beam deflection at upper and lower main field deflectors 213 and 214. By appropriately setting the voltages at the electrodes 230-238 of the focus 1 electrode assembly 216 and the electrodes 240-248 of the focus 2 electrode assembly 217, the XY beam position entering the field free tube 218 is changed to the focus 1 electrode assembly 216. It is almost the same as the XY beam position entering. At the entrance of the field free tube 218 is a divergent accelerating lens caused by the penetration of an electric field between the focus 2 assembly 217 and the field free tube 218, which diverts the beam deflection at the wafer into the main deflector. In general, increase by 25% compared to 213 and 214. Note that the beam 222 is significantly smaller in diameter than the entrance of the focus 1 electrode assembly 216 due to the focusing effect of the main lens.

  The use of a “moving lens” as described above has two important advantages.

  1) Since the beam is always on the axis of symmetry of the lens field, all off-axis aberrations are both geometric (coma, astigmatism, curvature of field, distortion) and chromatic aberration (magnification change). Essentially eliminated.

2) The beam impinging on the wafer surface is always scanned telecentrically, i.e. perpendicular to the wafer surface, thereby improving the depth of focus.
[Considerations for beam-limited aperture position]

The position of the beam limiting aperture 212 in the present invention is an important consideration in determining the effectiveness of the beam patterning process on the substrate surface 221. The following positions in the lens barrel can be considered.
1. Near source chip 201

FIG. 3A is a diagram of the profile of the beam 222 in the vicinity of the source chip 201. The distribution of the uniform intervals of the electron beam sections 302 in the beam 222 can be clearly confirmed. At position 301, the beam 222 is very close to the tip and has little time to deviate from the ideal evenly spaced angular distribution, so any effects due to spherical or chromatic aberration are minimal. The spherical aberration has a tendency to bend the outer electron beam back toward the optical (symmetric) axis, and appears in a form in which the distance between the outer electron beams is narrower than that of the inner electron beam. Chromatic aberration tends to separate low energy electrons radially from high energy electrons. The disadvantage of positioning the beam limited aperture 212 in the vicinity of the source chip 201 is that electrons impinging on the aperture are backscattered and collide with the source chip 201, causing heating and outgassing.
2. The top of the acceleration column 209

FIG. 3B shows the beam profile at the top of the acceleration column 209. In this position (position 303), the electron beam segments 304 in the beam 222 are still evenly separated, and spherical and chromatic aberrations due to the source lens (including electrodes 202, 203, 204, and 205) are still minimal. is there.
3. Directly above the main deflector

  FIG. 2B shows a third possible position of the beam limiting aperture 212, which is the position selected for the embodiments described herein. The main advantage of placing the beam limiting aperture 212 at position 3 relates to the optical alignment of the lens barrel. By arranging the beam trimming aperture 276 at the bottom of the acceleration column 209, the main field deflectors 213 and 214, the subfield deflector / astigmatism corrector 215, and the main lens are passed through the beam limiting aperture 212. It can be used to assist in setting the excitation of alignment deflectors 207 and 208 for the purpose of directing the beam to the bottom of the containing barrel. Proper alignment of the beam through the main lens is important in obtaining a proper patterned beam at the substrate surface 221.

Note: The electron beam energy at the point in the column where the beam limiting aperture is located is a factor in determining the amount of aperture heating that can occur. Aperture heating should be considered when determining the position of the beam limited aperture.
[Another consideration for optimizing the placement of beam-limited apertures]

  The optimization process that can be used in the process shown in FIG. 1 is repeated for various positions of the beam limited aperture in the column. This generates a beam profile and current density distribution on the substrate and a design of the beam limited aperture for each position. The best position is determined taking into account:

a) Beam profile on the substrate b) Beam current density on the substrate c) Manufacturability of beam limited aperture

  FIG. 4A shows a calculated circular beam profile on the surface of the substrate 221 centered on the optical axis, representative of that generated in block 108 of FIG. 1, prior to insertion of the beam pattern limiting aperture 212 into the column of FIG. 2A. Show. The cross section of the beam 222 is a graph with respect to two axes of X401 and Y402. Although the desired patterned beam shape 403 is illustrated here as a square, it can have any practical shape and is superimposed on the beam cross section. The electron beam 404 reaches the outside of the desired shape 403, while the electron beam 405 reaches the inside of the desired shape 403. For each of the electron beams 404 and 405, the XY coordinates in the beam limiting aperture 212 are recorded along with the corresponding XY coordinates on the wafer surface 221 shown here.

  The calculation in block 108 of FIG. 1 adjusts the diameter of the circular beam to just surround the desired beam pattern 403 at the corners of the illustrated (square) pattern 403 with only a few electron beams lost. is doing. Thereby, since the maximum efficiency is ensured in the use of the emission current from the source chip 201, the current density of the beam 222 in the substrate 221 is maximized. In the illustrated example, the desired beam pattern 403 is a 40 nm square, and the diameter of the circular beam is adjusted to be slightly larger than √2 × 40 nm≈56 nm corresponding to the distance between the diagonals of the square beam pattern 403. Has been.

  FIG. 4B shows a substrate 221 centered off +12.5 μm from the optical (Z) axis, representative of what is generated in block 108 of FIG. 1 prior to insertion of the beam pattern limiting aperture 212 into the column of FIG. Figure 3 shows a calculated circular beam profile at the surface. The cross section of the beam 222 is a graph with respect to the two axes X406 and Y402, and uses the same Y axis 402 as in FIG. Corresponding to X = + 12.5 μm of the X axis 401, different X axes 406 are defined. The desired patterned beam shape 403 is illustrated as a square and is the same as FIG. 4A. The electron beam 407 reaches the outside of the desired shape 403, while the electron beam 408 reaches the inside of the desired shape 403. For each of the electron beams 407 and 408, the XY coordinates in the beam limiting aperture 212 are recorded together with the corresponding XY coordinates in the wafer surface 221 shown here.

  The design of the optical column must ensure that the beam remains circular with approximately the same diameter, even when deflected off the +12.5 μm axis as shown. This is important to achieve the same beam size and current density when deflected off-axis, such as when the beam is near the optical axis. Furthermore, it should be noted that the set of electron beams 408 reaching the inside of the desired pattern 403 shown here does not necessarily correspond to the set of electron beams 405 reaching the inside of the desired pattern 403 in FIG. 4A. It is also important. This is due to the off-axis geometric aberration of the optical column.

  FIG. 4C shows the substrate 221 surface centered off +25 μm from the optical (Z) axis, representative of that generated in block 108 of FIG. A calculated circular beam profile is shown. The cross section of beam 222 is graphed with respect to the two axes of X409 and Y402, using the same Y axis 402 as in FIGS. 4A and 4B, but for X deflection of +25 μm, X = 0 of X axis 409 Corresponding to X = + 25 μm of the X axis 401, different X axes 406 are defined. The desired patterned beam shape 403 is illustrated as a square and is the same as in FIGS. 4A and 4B. The electron beam 410 reaches the outside of the desired shape 403, while the electron beam 411 reaches the inside of the desired shape 403. For each of the electron beams 410 and 411, the XY coordinates in the beam limiting aperture 212 are recorded together with the corresponding XY coordinates in the wafer surface 221 shown here.

  The design of the optical column must ensure that the beam remains circular with approximately the same diameter even when deflected off the +25 μm axis as shown. This is important to achieve the same beam size and current density when maximum distance deflection occurs off-axis, such as when the beam is near the optical (Z) axis. Furthermore, the set of electron beams 411 reaching the inside of the desired pattern 403 shown in FIG. 4A is changed to the set of electron beams 405 reaching the inside of the desired pattern 403 of FIG. It is also important to note that it does not necessarily correspond to a set of electron beams 408 that reach the interior of the pattern 403. This is due to the off-axis geometric aberration of the optical column.

  FIG. 5 shows a calculated square beam profile on the surface of the substrate 221 centered on the optical axis, representative of what is generated in blocks 108 to 110 after insertion of the beam pattern limiting aperture 212 into the column of FIG. 2A. Yes. The cross section of the beam 222 is a graph with respect to two axes of X401 and Y402. All electron beams 505 are inside the desired patterned beam shape 403 shown in FIG. 4A.

  Note that the calculation of block 110 determines the intersection of the next five sets of electron beams.

1) A set of electron beams 405 in FIG. 4A (corresponding to no deflection)
2) A set of electron beams 408 in FIG. 4B (corresponding to X deflection of +12.5 μm)
3) A set of electron beams 411 in FIG. 4C (corresponding to X deflection of +25 μm)
4) A set of electron beams 408 (FIG. 4B) mirrored about the Y axis on the wafer surface 221 (with the X coordinate inverted), thereby generating data corresponding to an X deflection of −12.5 μm. Note that the electron beam in the beam limiting aperture 212 is also mirrored by
5) A set of electron beams 411 in FIG. 4C mirrored about the Y axis on the wafer surface 221 (with the X coordinate inverted) (this generates data corresponding to an X deflection of −25 μm).

  These five sets of intersections are generally 5 to 10% smaller than any of the five individual sets. This is due to off-axis geometric aberrations that distort the original circular beam profile. The goal of the optical design process in block 108 of FIG. 1 is to minimize off-axis geometric aberrations, thereby maximizing the size of the intersection of the five sets of electron beams. The set of electron beams 505 shown in FIG. 5 is a subset of the set of electron beams 405 of FIG. 4A resulting from this effect. In this particular example, a set of electron beams 405 has 237 components, while a set of electron beams 505 has 223 components, a decrease of 5.9%.

  FIG. 6A is a graph of a set of ideal electron beams 605, 606, and 607 to be transmitted to the wafer at the beam limiting aperture 212, corresponding to the electron beam 505 of FIG. This is a result of the ideal aperture design created in block 112 of FIG. Since the electron beam 605 in the center set in the region having the XY coordinates satisfying −40 μm <X <+40 μm and −40 μm <Y <+40 μm corresponds to one having a small angle with the optical axis in the source chip 201, Corresponds to the electron beam with the least spherical aberration. The outer set of electron beams 606, where X <−40 μm, X> +40 μm, Y <−40 μm, and / or Y> +40 μm, may be desired even if the primary optical system deviates from the desired beam profile 403. It corresponds to an electron beam that undergoes a large spherical aberration that is “bent” over the beam profile 403. Note that there are two holes in the electron beam distribution parallel to the Y-axis 602. Two other holes 604 are present in the electron beam distribution parallel to the X-axis 601. These four holes 603 and 604 correspond to electron beams (such as 404, 407, or 410) that deviate from the desired beam profile 403. Along the diagonals of the Y-axis 602 and the X-axis 601, all of the electron beam is transmitted, which carefully carefully converts the original (circular) beam profile to the desired size beam shape 403 along the diagonal dimension. It is the result of matching. The holes 603 and 604 correspond to regions of the beam pattern limited aperture 212 that are opaque to the beam 222, ie, a solid pattern. However, as can be seen in FIG. 6A, such a solid pattern cannot be physically realized because it is separated from the outside of the beam limited aperture 212. The electron beam 607 corresponds to the electron beam that needs to be sacrificed to achieve a practical aperture design as shown in FIG. 8A.

  FIG. 6B is a graph of a set of ideal electron beams 610 and 611 that should be shielded by the beam limiting aperture 212 so as not to reach the wafer surface 221. This set of electron beams is complementary to the set shown in FIG. 6A and is the result of the ideal aperture design created in block 112 of FIG. A set of electron beams 610 corresponds to the electron beams that need to be shielded to produce the desired square beam profile 403, and these electron beams 610 correspond to the holes 603 and 604 in FIG. 6A. Along the diagonals of the X and Y axes, the electron beam is not shielded at all, which carefully matches the original (circular) beam profile to the desired size beam shape 403 along the diagonal dimension. It is a result. The surrounding electron beam 611 is caused by the fact that the initial (circular) beam profile is slightly longer than the diagonal dimension of the desired beam shape 403, otherwise it is the corner corner in the actual beam profile. This is desirable because omissions can occur.

  FIG. 7A is a graph of the actual set of electron beams that were transmitted to the wafer by the practical aperture design shown in FIG. 8A created in block 114 of FIG. The only change in the transmitted electron beam between FIG. 6A and FIG. 7A is the position 701 where no transmitted electrons are present. This corresponds to the disappearance of the six electron beams 607 in FIG. 6A.

  FIG. 7B is a graph of an actual set of electron beams shielded by the aperture design shown in FIG. 8A. This set of electron beams is complementary to the set shown in FIG. 7A. The only change in the shielded electron beam between FIG. 6B and FIG. 7B is the position 702 with the addition of six shielded electrons corresponding to the electron beam 701 lost from FIG. 7A.

  FIG. 8A shows the result of the final change to the aperture design made in block 114 of FIG. 1 to obtain a practical beam pattern limited aperture 212 design. Electron beam segments 605 and 606 correspond to FIGS. 6A and 7A, and struts 812 correspond to the additional aperture material required to support the four central structures 805 that shield the electron beam corresponding to electron beam segment 610. . Various chamfers 803 are added to the aperture design to add strength and ease of manufacture, however, the size of these chamfers is kept to a minimum and avoids shielding of excessive beam current. Should. The design is not quadrusymmetric due to the fact that the main field deflection is always parallel to the X-axis 601 and the deflection parallel to the Y-axis 602 is only a little (± 1 μm). This non-quadru symmetry is manifested by the addition of two small protrusions 804 within the central square opening. Since the beam change about both the X-axis 601 and the Y-axis 602 is symmetric, the resulting beam pattern limited aperture design is mirror symmetric about both the X-axis 601 and the Y-axis 602.

  The beam pattern limited aperture (PBDA) 212 is a conductive structure and is maintained at a fixed potential. The metal foil, the conductive film, the conductive coating film, or an equivalent thin material can be manufactured by machining, laser ablation, micromachining, or the like. There is an option to use a patterned thick film supported by a continuous film that is “electron permeable” for applications that use the PBDA 12 at a position where high energy electrons are incident in the lens barrel. With this design, PBDA can be manufactured without the need for mechanical support struts 812 and chamfers 803.

  Another important consideration is, for example, in a Schottky emitter, the energy spread of the electron beam 222 above 1.0 eV FWHM. In all optical elements above the upper barrel, ie, the beam limiting aperture 212, the effect of chromatic aberration is to blur the beam 222 at the beam limiting aperture 212. In an optical design without intermediate crossing over the PBDA 212, this blur generally appears as high energy electrons far from the optical axis and low energy electrons near the optical axis. If the color beam blur is too large, the electron beam that intersects the beam pattern limited aperture will not be properly apertured. FIG. 8A shows the intersection of the electron beam with the plane of the beam limiting aperture 212 corresponding to a nominal energy electron, in this example 5000 eV.

  FIG. 8B is a diagram showing a part of the mapping of the electron beam passing through the PBDA 212 to the electron beam intersection on the wafer surface 221. A square beam 840 on the wafer surface 221 is shown relative to an electron beam passing through the center (substantially square) opening 860 of the PBDA 212. The electron beam 841 passes through the upper right corner of the opening 860 and reaches the wafer surface 221 at the upper right corner of the square beam 840. Similarly, the electron beam 845 passes through the lower left corner of the opening 860 and reaches the wafer surface 221 at the lower left corner of the square beam 840. The mapping of the electron beam passing through the central opening of the PBDA 212 follows the same pattern for the electron beams 842-844, which is basically a prior art shaped beam that provides a relatively low current density at the wafer surface 221. In the system, it is identical to the way the patterned beam is formed.

  The mapping of the electron beam between the PBDA and the wafer shown in FIG. 8B is basically an image of the central square opening 860 with respect to the wafer surface 221. In prior art shaped beam systems, the beam shape at the wafer surface 221 is typically due to a 1: 1 mapping of the (X, Y) coordinates of the beam shaping aperture (s) to the (X, Y) coordinates of the wafer surface 221. One or more aperture images are reduced on the wafer surface 221. The 1: 1 mapping means that exactly one (X, Y) coordinate exists on the plane of the wafer 221 for all (X, Y) coordinates on the plane of the PBDA 212. In order to achieve 1: 1 mapping, it is necessary to minimize all aberrations of the optical system so that imaging by the primary optics of the system dominates. This constraint affects the total amount of current that can be focused on the beam at the wafer surface 221 because aberrations need to be minimized by limiting the range of beam angles at the source chip 201 that are transmitted to the wafer surface 221. .

  However, it is not necessary to utilize such a 1: 1 mapping to form a shaped beam at the wafer surface. N: 1 mapping from the PBDA plane to the wafer plane can be used (where N is an integer, N ≧ 2, N = 3 in embodiments of the present invention). In this case, at N = 3, there are exactly three different (X, Y) coordinates in the plane of the PBDA 212 that map essentially all (X, Y) coordinates in the plane of the wafer surface 221. The only exception to this N: 1 mapping is (0,0) = beam center, but it does not affect the beam shape because it is far from the edge of the beam. Because the irradiation to the PBDA 12 is uniform, a 3: 1 mapping can focus a larger current onto the shaped beam on the wafer surface 221 than is possible with a 1: 1 mapping. The reason why the mapping of the present invention is 3: 1 (rather than 2: 1, 4: 1 etc.) is because spherical aberration is the dominant aberration on the axis. To maintain a square beam shape off-axis, it is further necessary to minimize off-axis aberrations, which otherwise would degrade the 3: 1 mapping and cause the desired patterned beam shape to be lost. This is because In the present invention, a “moving lens” (see FIGS. 3K and 3L) is used to maintain the effective axis of the main lens that is concentric with the beam for all off-axis beam deflections (see FIG. 23). Since the beam is always on the effective axis of the main lens, all off-axis aberrations (geometric and chromatic aberration) are almost eliminated and the predominance of spherical aberration (not changing off-axis) is retained.

  N: 1 mapping is possible because individual electron beams (each electron beam represents a single electron trajectory) can overlap each other with little interaction. Any residual interaction is called “space charge beam diffusion”. For beam currents in the nA range, these effects are minimal at the beam energy (50 keV) employed in the present invention.

  In order to increase the current density of the shaped beam 840, in addition to the electron beams 841 to 845, a set of electron beams is illustrated in FIG. 8C and passes through the outer openings 861 and 862 of the PBDA 212. The electron beam 846 passes through an opening 861 just outside the central square opening 860 and reaches the wafer surface 221 at the lower left corner of the shaped beam 840. The electron beam 851 passes through the opening 861 at approximately the middle between the inner edge and the outer edge of the opening 861 and reaches the wafer at the center of the shaped beam 840. The electron beam 849 passes near the outer edge of the opening 861 and reaches the wafer at the upper right corner of the shaped beam 840. The mapping of the electron beams 847, 850, and 848 that pass through the opening 862 is similar as shown. Unlike the aperture 860, the electron beams that pass through the apertures 861 and 862 are “folded” and the electron beams that have passed through the apertures 861 and 862 farthest from the center of the PBDA 212 are opposite corners of the shaped beam 840. This reaches a state where the wafer surface 221 is reached. Since the current density irradiated on the PBDA 212 is uniform, the total beam current reaching the shaped beam 40 is proportional to the area of the openings 860 to 862. Since the focusing of the electron beams 841 to 845 is similar to that used to generate the shaped beam of the prior art, the area of the opening 860 is compared to the total area of the openings 860 to 862. Determine the increase in beam current density at the wafer surface 221 resulting from the use of the invention. In the example of FIGS. 8B and 8C, the total area of the openings 861 and 862 is six to seven times the area of the opening 860, so an increase in current density is possible using only the central opening 860. About 7 to 8 times that of things.

  FIG. 9 shows a diagram of the various beam positions A to D on the wafer surface used for the calculation of the beam profile. Using only the subfield deflector / astigmatism 215 (see FIG. 2A), the maximum deflection is ± 1 μm in the X direction 910 and ± 1 μm in the Y direction 911, and the beam 222 moves from the central position A901 to the position B902. To do. If only the main field deflectors 213 and 214 are used, the maximum deflection shown is ± 25 μm in the X direction 910 and the beam 222 moves to position C 903. Using both main field deflectors 213 and 214 and subfield deflector / astigmatism 215, the beam moves to position D904. A typical 2 μm square subfield is illustrated and defines the width of the scan stripe 2 μm 914. The main field deflectors 213 and 214 move the beam as a whole along the X axis 910, while the wafer stage moves in a direction 915 parallel to the Y axis 911 in the meander pattern shown in FIG. 27B.

FIG. 10A shows the calculated exposure dose at position A in FIG. 9 by one flash of the square electron beam. The coordinate axes on the substrate surface 221 are X1001 and Y1002. Region 1004 corresponds to a beam current density of 3000 A / cm ² or higher. In the illustrated example, when the resist sensitivity is assumed to be 5 μC / cm ² and the residence time is 1.67 ns, the following results.

Dose to substrate = (current density) (residence time) = (3000 A / cm ² ) (1.67 ns) = 5 μC / cm ² = resist sensitivity

Thus, the resist is completely exposed inside the region 1004. In region 1003, the exposure dose is less than 5 μC / cm ² and therefore the resist is not fully exposed. FIG. 10A shows that region 1004 is a square of about 40 nm. The combined effects of virtual source size, chromatic aberration (for all orders), and spherical aberration (for all orders) are all considered in this calculation. The beam profile plot of FIG. 13 corresponds to the current between the two lines 1005 and 1006, ie, the current density over one side of the square beam shape.

FIG. 10B shows the calculated exposure dose at position B in FIG. 9 by one flash of the square electron beam. The coordinate axes on the substrate surface 221 are X1011 and Y1012. Region 1014 corresponds to a beam current density of 3000 A / cm ² or higher. With a dwell time of 1.67 ns and resist sensitivity of 5 μC / cm ² , the resist is completely exposed inside the region 1014. In region 1013, the exposure dose is less than 5 μC / cm ² , so the resist is not fully exposed. FIG. 10B also shows that region 1014 is a square of about 40 nm, very similar to region 1004. The similarity between regions 1004 and 1014 indicates that the full X and Y subfield deflection of +1 μm has a minor effect on the beam shape. The combined effects of virtual source size, chromatic aberration (for all orders), spherical aberration (for all orders), and off-axis aberration (both geometric and chromatic aberration for all orders) are all taken into account in this calculation.

FIG. 10C shows the calculated exposure dose at position C in FIG. 9 by one flash of the square electron beam. The coordinate axes on the substrate surface are X1021 and Y1022. Region 1024 corresponds to a beam current density of 3000 A / cm ² or higher. With a dwell time of 1.67 ns and resist sensitivity of 5 μC / cm ² , the resist is completely exposed inside the region 1024. In region 1023, the exposure dose is less than 5 μC / cm ² , so the resist is not fully exposed. FIG. 10C also shows that region 1024 is a square of about 40 nm, very similar to regions 1004 and 1014. The similarity between regions 1004 and 1024 indicates that the full main field deflection of +25 μm has a small effect on the beam shape. The combined effects of virtual source size, chromatic aberration (for all orders), spherical aberration (for all orders), and off-axis aberration (both geometric and chromatic aberration for all orders) are all taken into account in this calculation.

FIG. 10D shows the calculated exposure dose at position D in FIG. 9 by one flash of the square electron beam. The coordinate axes on the substrate surface are X1031 and Y1032. Region 1034 corresponds to a beam current density of 3000 A / cm ² or higher. With a dwell time of 1.67 ns and resist sensitivity of 5 μC / cm ² , the resist is completely exposed inside the region 1034. In region 1033, the exposure dose is less than 5 μC / cm ² , so the resist is not fully exposed. FIG. 10D similarly shows that region 1034 is a square of about 40 nm, very similar to regions 1004, 1014, and 1024. The similarity between regions 1004 and 1034 indicates that the +25 μm full main field deflection combined with the +1 μm full X and Y subfield deflection has a minor effect on the beam shape. The combined effects of virtual source size, chromatic aberration (for all orders), spherical aberration (for all orders), and off-axis aberration (both geometric and chromatic aberration for all orders) are all taken into account in this calculation.

FIG. 11 shows the calculated exposure dose at position A in FIG. 9 by a single flash of adjacent square electron beams (same as FIG. 10A) in the “L” pattern with a 40 nm spacing between the beam centers. The coordinate axes on the substrate surface are X1101 and Y1102. Region 1104 corresponds to a beam current density of 3000 A / cm ² or higher. With a dwell time of 1.67 ns and resist sensitivity of 5 μC / cm ² , the resist is completely exposed inside the region 1104. In region 1013, the exposure dose is less than 5 μC / cm ² , so the resist is not fully exposed. The exposure region 1104 is substantially “L-shaped”, and has some chamfers 1105 at the “L-shaped” bent portion. The width of the “L-shaped” arm is about 40 nm and corresponds to the 40 nm square region 1004. FIG. 11 demonstrates that a complex pattern can be exposed by adjoining the square beam on the substrate surface 221. The combined effects of virtual source size, chromatic aberration (for all orders), and spherical aberration (for all orders) are all considered in this calculation.

FIG. 12 shows the calculated exposure dose at position A in FIG. 9 with two overlapping flashes of a square electron beam and one separate flash (all flashes are the same as in FIG. 10A). The coordinate axes on the substrate surface are X1201 and Y1202. Regions 1204 and 1205 correspond to a beam current density of 3000 A / cm ² or higher. With a dwell time of 1.67 ns and a resist sensitivity of 5 μC / cm ² , the resist is completely exposed within regions 1204 and 1205. In region 1203, the exposure dose is less than 5 μC / cm ² , so the resist is not fully exposed. The exposure area 1204 is the same as the exposure area 1004. Region 1205 corresponds to two flashes of the 40 nm square beam of FIG. 10A with a 10 nm overlap between the centers by 30 nm. This overlap causes the central area of region 1205 to be overexposed, resulting in a 2 to 3 nm bump 1206 in the pattern. FIG. 12 demonstrates that overlapping square beams can expose patterns that do not correspond to integer multiples of the square beam size (40 nm in this case). The combined effects of virtual source size, chromatic aberration (for all orders), and spherical aberration (for all orders) are all considered in this calculation.

FIG. 13 shows a graph 1302 of the calculated beam current density at position A in FIG. 9 for a single square beam (same as FIG. 10A) and a single Gaussian beam along the X-axis 1301. The height of the Gaussian beam 1304 is the beam current density as shown by the intersection point 1305 with the square beam current density curve 1303 calculated by averaging the current density along the Y axis between lines 1005 and 1006 in FIG. 10A. The FWHM was adjusted to about 40 nm at 3000 A / cm ² (dose = 5 μC / cm ² , assuming a residence time of 1.67 ns). The square beam current density drops from 3000 A / cm ² (intersection 1305, ± 20 nm from the beam center) to less than 1000 A / cm ² at only 5 nm away from the beam center (at ± 25 nm position).

FIG. 14 shows 40 nm center-to-center spacing along the X-axis 1401 compared to the current density produced by three adjacent Gaussian beams (all at 3000 A / cm ² with FWHM 40 nm and center-to-center spacing 40 nm). A graph 1402 of the calculated beam current density of the three square beams obtained at position A in FIG. 9 is shown. The Gaussian beam is adjusted in the same way as in FIG. 13 so that an intersection 1405 occurs at a beam current density of 3000 A / cm ² (assuming a dose = 5 μC / cm ² and a dwell time of 1.67 ns). The tails of the three square beam curves 1403 are compared to the tail drop of the three Gaussian beam curves 1404, −20 nm and +100 nm (= width 120 nm = 3 × 40 nm, where 40 nm = each square beam (Width) is descending very quickly from the beam edge. FIG. 14 shows that the sharp drop in current density at the edges of a square beam causes these beams to be adjacent, producing large size features with essentially the same maximum current density as seen along the edge of a single beam. It shows that it is possible. This explains why beam flashes can be combined to form large patterns (such as pattern 1104 in FIG. 11) with minimal bulges due to individual beam flash current tails.

FIG. 15 shows a graph 1502 of calculated beam current density at position A in FIG. 9 of three square beams adjacent along the X-axis 1501 with a 40 nm center-to-center spacing, with individual square beam profiles 1503, 1504 and 1505. And the three beam composite profile 1403 of FIG. The relatively flat top of curve 1403 (ranging from 6045 A / cm ² to 6667 A / cm ² ) has a net of less than ± 5% due to the sum of the steep sides of each square beam profile 1503, 1504, and 1505. It shows how the current density fluctuates.

FIG. 16 shows a state in which each separated Gaussian beam has a FWHM of 40 nm at an exposure amount of 3000 A / cm ² (residence time 1.67 ns and resist sensitivity 5 μC / cm) with the beam centers separated by 40 nm along the X-axis 1601. ² ), a graph 1602 of the calculated beam current density of three combined Gaussian beams 1404 (FIG. 14) and three separated Gaussian beams 1603, 1604 and 1605 is shown. A long tail portion can be clearly confirmed outside the edge of the desired exposure region of −20 nm to +100 nm. Such a long tail portion reduces the process flexibility necessary to maintain the desired pattern critical dimension (CD).

  Comparing curve 1403 in FIG. 15 with curve 1404 in FIG. 16, the combination of the three adjacent square beams of FIGS. It can be seen that it generally demonstrates a sharp edge (rapid drop in current density).

  FIG. 17A shows a diagram of a beam scanning method that can be used to set up an optical system that generates an optimized square beam profile. The desired square beam profile at location 1701 is located on the central surface 1723 of the special mask structure that may be included on the wafer stage of the system or on a special setting wafer installed in the system. Surface 1713 is configured to provide a larger imaging signal when irradiated with beam 222 compared to the imaging signal resulting from central surface 1723. This imaging contrast can be achieved by connecting surface 1713 to the imaging system and not connecting surface 1723 to the imaging system. While beam 222 is at location 1701, beam current flows to surface 1723 that is not connected to the imaging system. Scanning beam 222 in direction 1702 to edge 1710 increases the portion of the beam current that impinges on surface 1713 connected to the imaging system. In order to avoid charging effects, the current flowing on both surfaces 1713 and 1723 should eventually flow to ground. At position 1703, half of the beam is in the collection region 1713, so half of the beam current is collected. When scanning the beam 222 in the direction 1702, the square edge of the beam profile is parallel to the edge 1710, so the signal is from 0% over a distance equal to the beam dimension D1720 parallel to the scan direction 1702. Note that it is up to 100%.

  Scanning beam 222 in direction 1704 to edge 1711 increases the portion of the beam current that impinges on surface 1713, thus increasing the portion that is recovered and provides the imaging signal. At position 1705, half of the beam 222 is in the collection region 1713, so half of the beam current is collected. When scanning the beam 222 in the direction 1704, the square edge of the beam profile is 45 ° to the edge 1711, so the signal is at a distance equal to the diagonal dimension of the beam parallel to the scanning direction 1704 √2D1722. Note that the transition from 0% to 100% over time.

  When scanning the beam 222 in the direction 1706, assuming that the beam profile is a square having a dimension D1721 parallel to the scanning direction 1706, the result is similar to that described for the scanning direction 1702 above. At position 1707, half of the beam current is recovered because half of the beam exceeds the edge 1712 of region 1713.

  Another way to generate image contrast is to make regions 1713 and 1723 with materials having different secondary electron emission coefficients, thereby allowing the use of the detector optics of FIG. 2A. Methods of imaging within an electron beam system are well known to those skilled in the art.

  FIG. 17B is a diagram illustrating a computed line scan for scan directions 1702 and 1704 of FIG. 17A, showing a possible method for setting an optimized square beam. The illustrated axis 1740 corresponds to a position along one of the scan directions 1702, 1704, or 1706. The intensity axis 1741 is relative from 0.0 corresponding to no detected imaging signal to 1.0 corresponding to detection of the largest image signal. Curve 1742 corresponds to either scan direction 1702 or 1706. The increase in intensity from 0.0 to 1.0 on curve 1742 is from -27.5 nm to +27.5 nm. A 45 ° curve 1743 corresponds to the scan direction 1704. The increase in the intensity of the curve 1743 from 0.0 to 1.0 is −42.5 nm to +42.5 nm, which is a significantly slower increase than the curve 1742. The difference in intensity increase between the two curves 1742 and 1743 can be used to adjust the optical column to produce a beam profile that is as close to a square as possible. Center point 1745 corresponds to the beam at position 1703, 1705, or 1707.

  For comparison, the intensity curve 1744 was plotted for a single Gaussian beam with a FWHM of 40 nm. This curve overlaps curve 1743 at a position far from the center of the 0 nm beam. Near the center of the 0 nm beam, this curve rises more slowly than either curve 1742 or 1743. The important difference is that the Gaussian curve has the same shape for any of the scan directions 1702, 1704, or 1706, and the Gaussian beam distinguishes it from the square beam when imaged using the detector surface 1713. It is to be done.

FIG. 18 shows the calculated exposure dose at position A in FIG. 9 with a single flash of a 30 nm square electron beam using the beam pattern limiting aperture 212 of FIG. 8A. The coordinate axes on the substrate surface 221 are X1801 and Y1802. When patterning various shapes on a substrate by lithography, it would be beneficial if a range of shaped beam sizes could be generated without the need for mechanical replacement of the beam pattern limited aperture 212. By adjusting the electron optical system in the upper part of the lens barrel (usually by changing the common voltage between the first source lens electrode and the beam limiting aperture), the beam trimming aperture (BTA) 276 is irradiated with the circular beam 222. The diameter can be adjusted to change the shaped beam size at the wafer surface 221 (see FIG. 3D). In the example shown in FIG. 18, the optical system of the upper barrel is adjusted to produce a 30 nm square beam 1804 (instead of the 40 nm square beam shown in FIG. 10A). In this configuration, the exposure beam current density is reduced to 2000 A / cm ² because the entire optical system cannot be fully optimized to provide the highest current density beam (requires different BTA 276 and PBDA 212). A residence time of 2.50 ns is required for a resist sensitivity of 5 μC / cm ² . Region 1804 corresponds to a current density of 2000 A / cm ² or higher and is a square of about 30 nm. In area 1803, the exposure dose is less than 5 μC / cm ² , so the resist is not fully exposed. The combined effects of virtual source size, chromatic aberration (for all orders), and spherical aberration (for all orders) are all considered in this calculation. FIG. 18 demonstrates that the optical system of FIG. 2A, optimized for generating 40 nm square beams, can also be used for generating 30 nm square beams. The performance shown in FIG. 18 sufficiently enables the patterning of features with a resolution reduced to 30 nm, with a slightly longer residence time than the optimized 40 nm case shown in FIGS. 10A-15.

FIG. 19 shows the calculated exposure dose at position A in FIG. 9 with a single flash of a square electron beam of about 80 nm using the beam pattern limiting aperture 212 of FIG. 8A. The coordinate axes on the substrate surface 221 are X1901 and Y1902. In this example, the optics at the top of the barrel is adjusted in the opposite direction to that shown in FIG. 18 to provide a square beam 1904 that is about twice as large (about 80 nm) as in FIG. 10A. Yes. In this configuration, the exposure beam current density is reduced to 2500 A / cm ² because the entire optical system cannot be fully optimized to provide a beam with the highest current density (requires different BTA 276 and PBDA 212). A residence time of 2.00 ns is required for a resist sensitivity of 5 μC / cm ² . Region 1904 is a square of about 80 nm, corresponding to a current density of 2500 A / cm ² or higher. In area 1903, the exposure dose is less than 5 μC / cm ² , so the resist is not fully exposed. The combined effects of virtual source size, chromatic aberration (for all orders), and spherical aberration (for all orders) are all considered in this calculation. FIG. 19 demonstrates that the optical system of FIG. 2A, optimized for the generation of 40 nm square beams, can also be used to generate approximately 80 nm square beams. The performance shown in FIG. 19 fully enables feature patterning at a resolution of 80 nm with a slightly longer residence time than the optimized 40 nm case shown in FIGS. 10A-15.

FIG. 20 shows the calculated exposure dose at position A in FIG. 9 by a single flash of a 120 nm square electron beam using the beam pattern limiting aperture 212 of FIG. 8A. The coordinate axes on the substrate surface 221 are X2001 and Y2002. In this example, the optical system at the top of the barrel is further adjusted in the same direction as shown in FIG. 19 to provide a beam 2004 that is three times as large (120 nm) as in FIG. 10A. In this configuration, the exposure beam current density is reduced to 2000 A / cm ² because the entire optical system cannot be fully optimized to provide the highest current density beam (requires different BTA 276 and PBDA 212). A residence time of 2.50 ns is required for a resist sensitivity of 5 μC / cm ² . Region 2004 corresponds to a current density of 2000 A / cm ² or higher and is a 120 nm square. In area 2003, the exposure dose is less than 5 μC / cm ² and therefore the resist is not fully exposed. The combined effects of virtual source size, chromatic aberration (for all orders), and spherical aberration (for all orders) are all considered in this calculation. FIG. 20 demonstrates that the optical system of FIG. 2A optimized for the generation of a 40 nm square beam can also be used for the generation of a 120 nm square beam. The performance shown in FIG. 20 fully enables feature patterning at 120 nm resolution with a slightly longer dwell time than the optimized 40 nm case shown in FIGS. 10A-15. With a 120 nm square beam, a 2 μm square subfield can be completely drawn with 256 flashes (inter-center spacing of 125 nm), which is necessary to fill a wide area to be drawn, such as a bonding pad.

  FIG. 21A is a graph of source lens focusing voltage 2105 (left axis 2101) and main lens focusing voltage 2104 (right axis 2102) in the barrel of FIG. 2A for square beam size 2103. A source lens voltage 2105 is applied to both the first lens electrode 203 and the beam limiting aperture 204. The main lens voltage 2104 is applied to the focus 2 support electrode 240 and is also a common mode voltage for the octupole electrodes 241 to 248. FIG. 21A shows a number of source lens voltage 2105 and main lens voltage 2104 values for various desired square beam sizes 2103 ranging from 30 nm to 120 nm. Curve 2105 is much lower than the energy of incident electrons from source chip 201, which is typically greater than 2800 eV, indicating that the source lens is a decelerating electrostatic lens. Curve 2104 ranges from below to above 5000 eV, which is the energy of electrons entering the main lens. The main focusing effect occurs between the focus 2 assembly 217 in the range of 4906.5V to 5217.2V and the field free tube 218 at 49986V for writing on the wafer 221 at 50000 eV.

  FIG. 21B is a graph of the half-angle 2114 (left axis 2111) at the source chip 201 and the beam current 2115 (right axis 2112) at the wafer surface 221 for a square beam size 2113. The half angle 2114 in the source chip 201 has the following relationship with the beam current 2115 on the wafer surface 221.

I _S = source angular intensity at the radial solid angle used to illuminate the beam pattern limited aperture 212 (usually this angular intensity ranges from 100 μA / sr to over 500 μA / sr). In the table below, I _S = 500 μA / sr was assumed. The angular intensity is generally almost unchanged within a half angle of a few degrees of the optical axis.

α = half angle of beam 222 in source chip 201 (unit: degree)
I _beam = beam current at wafer surface 221
= I _S [π (απ / 180 °) ² ]

  The optical design in block 106 of FIG. 1 has been optimized to produce a 40 nm square beam with the highest possible current density. This means that the optics are not optimized for the other beam sizes (30 nm, 80 nm, and 120 nm) shown in the table and FIGS. 21A-21D. This is the reason for the decrease in current density identified in the table for both smaller and larger beams (see curve 2125 in FIG. 21C). The optical design in block 106 of FIG. 1 can be optimized for beams larger or smaller than 40 nm, particularly optimized for 30 nm or smaller for tool scalability for future device generations. Can be executed. In this case, the performance for large beams is likely to decline, but is not expected to decline significantly.

FIG. 21C is a graph of current density 2125 (right axis 2122) at flash time 2124 (left axis 2121) and wafer surface 221 (assuming 5 μC / cm ² resist sensitivity) versus square beam size 2123. Flash time and current density have an inverse relationship.
(Current density) = (5 μC / cm ² ) / (flash time)

  Thus, as the current density 2125 increases, the flash time 2124 decreases conversely. Again, because the optical design is optimized for a 40 nm beam, the performance at all other beam sizes is inferior in terms of flash time 2124 for both larger and smaller 40 nm (shortest) Flash time is best). Degradation from 40 nm to 30 nm is most noticeable, indicating that an optical design optimized for 30 nm can function very well above 40 nm (though not shown in FIG. 21C). As can be seen, the closer the beam size is to the optimal size, the better the performance in terms of flash time.

  FIG. 21D is a graph of virtual source magnification 2134 (left axis 2131) versus square beam size 2133. FIG. The magnification 2134 determines the amount by which the corners of the square beam are rounded by the virtual source image. The lower the magnification 2134, the sharper the corners. In the Schottky electron source, since the radius of the virtual source is 10 nm, the magnification of 0.17 corresponds to the radius (0.17) (10 nm) = 1.7 nm of the square beam corner by the virtual source. Further rounding occurs due to chromatic and geometric aberrations.

  FIG. 22 is a diagram of a beam blanking strategy that can be used to vary the exposure per subfield to achieve proximity effect correction. For simplicity, in FIG. 22, the upper blanker is shown as two flat electrodes 2202 and 2203, one on each side of the beam 222. Similarly, the lower blanker is shown as two flat electrodes 2204 and 2205, one on each side of the beam 222. The electrons emitted from the source chip 201 are converged by the source lens 2201 and become a substantially parallel beam 222 that irradiates a beam limited aperture (BDA) 212 supported by the BDA attachment unit 211. Figure (a) shows an unblanked beam 222 that passes through a beam limiting aperture 212 into the lower part of the optical column where it is focused on the wafer surface 221 by the main lens assembly. In this case, the blanker plates 2202, 2203, 2204, and 2205 are at the same voltage (usually 5000V) and therefore do not induce a transverse electric field. In the absence of a transverse electric field, no beam deflection occurs in the blanker.

  Figure (b) shows the blanked beam. The voltage of the electrode 2202 has been changed by + 1.7V, and the voltage of the electrode 2203 has been changed by -1.7V, thereby forming a lateral electric field 2240 that deflects the beam 222 upward as it passes through the upper blanker. The Similarly, the voltage on electrode 2204 has been changed by -1.54V, and the voltage on electrode 2205 has been changed by + 1.54V, which is the upper blanker that deflects the beam downwards as it passes through the lower blanker. A transverse electric field 2241 in the opposite direction is formed. The net result of the two deflections is that the beam 222 reaches the plane of the beam pattern limited aperture (PBDA) 212 off axis and does not pass through the opening. With proper adjustment of the upper and lower blanker voltages, the position of the virtual source remains on the axis, providing conjugate blanking.

  FIG. 8C is a timing diagram illustrating a possible method for controlling the exposure on the wafer surface 221 as part of the method for proximity effect correction. The center of beam 222 has three possible positions in PBDA 212: + d, 0 (non-blanking), and -d. The beam may dynamically have an intermediate position between -d and + d when sweeping the PBDA 212. Five intervals 2221 to 2225 each having a length of T are illustrated, and a total of 5T is illustrated along the time axis 2245. The displacement at PBDA 212 is plotted on axis 2210. Proximity effect correction (PEC) requirements may require changing the dose for various subfields, as shown at intervals 2222, 2224, and 2225.

  (1) At the first interval 2221, the beam is blanked by being held off-axis at a distance + d as shown in FIG.

  (2) A second interval 2222 indicates a high-dose blanking signal that causes the beam 222 to linearly tilt 2232 across the PBDA 212. Since the slope 2232 spends the entire interval period T, it represents the maximum possible exposure amount corresponding to the drawing in the sparse pattern area where the proximity effect correction is essentially unnecessary.

  (3) The third interval 2223 indicates a blanking position different from the interval 2221. At the interval 2223, the beam is held at a position of a distance + d2233 corresponding to the mirror image (centered on the optical axis) in FIG.

  (4) The fourth interval 2224 shows a very low dose blanking signal, and the beam 222 ramps quickly 2234 from -d to + d across the beam limited aperture 212 and then for the remainder of the interval period T, Held at + d2235. This corresponds to drawing in a dense region of a pattern having a large proximity effect correction.

  (5) The last interval 2225 shows an intermediate situation between intervals 2222 and 2224, where the beam slopes 2236 from + d to -d for the majority of interval 2225, and then for the remainder of interval period T, -Held at d2237. This corresponds to drawing in an area having a pattern density lower than the interval 2224 but higher than the interval 2222.

  A possible advantage of this blanking technique is the ease of electronic implementation, because it can be very difficult to generate ns blanking pulses with sub-ns accuracy in pulse length. . In this proposed scheme, only the ramp rate needs to be controlled and there is no short rise and fall requirement in the voltage applied to the blanker plates 2202, 2203, 2204, and 2205, so the required blanking bandwidth. Becomes lower.

  Another blanking method employs a conventional technique in which a beam is quickly deflected from a first blanking position (for example, + d) to the optical axis (this cancels the blanking of the beam). After the beam remains at the center of the PBDA 212 for the required exposure time, the beam is quickly deflected to a second blanking position (eg, -d). The disadvantage of this approach is that a high bandwidth blanker is required because the slew rate becomes implied by a possible pixel exposure time error. The advantage of moving from the first blanking position and ending at the second blanking position is that all points on the PBDA 212 have the same total beam dwell time, which results in a uniform dose across the shaped beam. It is to become.

  For the next pixel to be exposed, the first blanking position is -d and the second blanking position is + d. Consecutive pixels are exposed using alternating blanking positions by reciprocating the beam across the PBDA 212 as shown in FIG.

  FIG. 23 is a cross-sectional close-up side view of the main lens showing the settings calculated for the focus 1 and focus 2 octupole voltages. The beam 222 exits the subfield deflector / astigmatism 215 and then enters the main lens. The beam at this point may already be deflected up to ± 20 μm off-axis, avoiding off-axis geometric aberrations (coma, astigmatism, curvature of field, distortion) and off-axis chromatic aberration (magnification change). In order to do this, the electrostatic fields generated by the focus 1 assembly 216 and the focus 2 assembly 217 also need to be moved off-axis by ± 20 μm. The prior art uses various complex schemes to achieve a “moving objective lens” or “variable axis lens”, and uses higher order derivatives of the on-axis electrostatic and / or magnetic lens field. Then, the application of dipoles, quadrupoles, hexapoles, octupoles, and higher-order fields to the on-axis lens field is controlled, and the effective lens axis is offset to match the beam deflection. If a large off-axis deflection of the shaped beam is required, it may be necessary to incorporate a more complex moving lens scheme that utilizes some or all of these additional optical elements. The present invention proposes a much simpler approach to adding a pure dipole field to the focus 1 216 and focus 2 217 fields.

  The focus 1 assembly 216 includes a support electrode 230 and octupole electrodes 231 to 238 (only the electrodes 232 and 237 are visible in the cross-sectional view of FIG. 23). In the description of FIGS. 3K-3L, the various voltages used to offset the electrostatic field to match the beam deflection have been described. FIG. 23 shows the resulting electrostatic equipotential lines. Line 2301 bulges into the area between the subfield deflector / astigmatism 215 and the focus 1 electrode assembly 216, and line 2310 is between the focus 1 assembly 216 and the focus 2 assembly 217. Bulges into the area. The shape of the lines 2301 and 2310 is determined by the voltages at the subfield deflector / astigmatism corrector 215, the support electrode 230, and the eight focus 1 octupole electrodes 231-238. The voltage at the focal point 2 assembly 217 has a small effect on the line 2301 but has an important effect on the line 2310. The inner diameter (ID) of the eight octupole electrodes 231 to 238 is smaller than the ID of the support electrode 230, and the electrodes 231 to 238 seem to have a dominant influence on the position and shape of the equipotential lines 2301 and 2310. become. Lines 2301 and 2310 are added by adding a small (less than 3V) electrostatic dipole element over the 5000V common mode voltage (see the description table of FIG. 3K) for the octupole electrodes 231-238. Can be moved ± 20 μm off-axis to match the beam deflection caused by the main field deflectors 213 and 214. The radial and off-axis beam position is theoretically determined at position 2305, after which the voltage on electrodes 231 through 238 is adjusted to eliminate any deflection of beam 222 as it passes through focus 1 assembly 216. . The lack of beam deflection is taken to indicate that line 2301 is properly offset and coincides with the offset of beam 222.

  Beam 222 enters focus 2 assembly 217 after exiting focus 1 assembly 216. The beam 222 at this point should be deviated and removed by the focus 1 assembly 216 (if the above setting procedure is performed correctly), so that the beam 222 can be deflected off axis by ± 20 μm. The focal point 2 assembly 217 includes a support electrode 240 and octupole electrodes 241 to 248 (only the electrodes 242 and 247 are visible in the cross-sectional view of FIG. 23). In FIG. 3L, the various voltages used to offset the electrostatic field to match the beam deflection have been described. FIG. 23 shows the resulting electrostatic equipotential lines. Lines 2303 and 2310 bulge into the area between the focus 1 assembly 216 and the focus 2 assembly 217. The shapes of the lines 2303 and 2310 are determined by the voltages at the eight focal point 1 octupole electrodes 231 to 238, the support electrode 240, and the eight focal point 2 octupole electrodes 241 to 248. The voltage of the field free tube 218 has a small influence on the distance between the lines 2303 and 2302. The IDs of the eight focal two octupole electrodes 241-248 are significantly smaller than the ID of the support electrode 240 so that the electrodes 241-248 have a dominant influence on the position and shape of the lines 2303 and 2310. Become. By adding an electrostatic dipole element (approximately 100V) to the electrodes 241-248 above its 4900-5200V common mode voltage (see the description table in FIG. 3L), lines 2303 and 2310 are drawn. It can be moved off-axis to match the beam deflection caused by the main field deflectors 213 and 214. The radial and off-axis beam position is theoretically determined at position 2306, and then the voltages on electrodes 241 through 248 are adjusted to remove almost all deflection of beam 222 as it passes through focus 2 assembly 217. To do. The lack of beam deflection is taken to indicate that lines 2303 and 2301 are properly offset and coincide with the offset of beam 222.

  The above procedure for setting the voltages for the focus 1 octupoles 231 to 238 and the focus 2 octupoles 241 to 248 has been determined theoretically. In practice, it typically requires at least two iterations of setting the focus 1 216 and setting the focus 2 217 before removing both deflections at the positions 2305 and 2306. In electro-optic modeling, the dipole voltages of the octupoles 231 to 238 and 241 to 248 vary linearly with respect to the beam offset, and the final sensitivity to the beam shape at the wafer 221 is excessive. Don't be. For this reason, the voltage obtained by theoretical modeling (the tables in the description of FIGS. 3K and 3L) is sufficient to realize the proposal of this lens offset method in combination with the line scanning procedure described in FIGS. 17A and 17B. Should be.

  FIG. 24 is a schematic diagram of an embodiment of a wafer stage and a position sensor. In a lithographic system, the substrate is illustrated here as a 300 mm wafer 2401, which is typically capable of XY motion and, in some cases, yawing (rotation about the Z axis perpendicular to the wafer), Z motion, Additional motion axes such as roll and pitch (rotation about two vertical axes X and Y in the plane of the wafer) are mounted on a precision stage 2402 that is capable. Here, only the movements in the first three axes X, Y, and yawing are considered. Y interferometer # 1 2406 and Y interferometer # 2 2407 send respective laser beams 2416 and 2417 to stage mirror 2404. Any relative movement between the wafer 2401 and the mirror 2404 adversely affects the positioning accuracy of the beam 222 in the Y direction and yawing axis on the wafer surface 221, so that the wafer 2401 is securely clamped with respect to the stage 2404. , And the mirror 2404 is very flat and is firmly attached to the stage 2402. The X interferometer 2405 sends the laser beam 2415 to a stage mirror 2403 that needs to be very flat and rigidly attached to the stage 2402 to avoid beam positioning errors in the X direction. The movement of the stage 2402 in the X and Y axes is determined by the relative positioning of the mirrors 2403 and 2404, and if these mirrors are not perpendicular to each other, the X and Y axes are also not perpendicular. The X, Y, and yaw positions of the wafer 2401 relative to the center 2422 of the stage 2402 are calculated using the following equations:

Y = [(Y interferometer # 1 2406 data) + (Y interferometer # 2 2407 data)] / 2 × K ₁
X = (X interferometer 2405 data) × K ₂
Yawing = [(Y interferometer # 1 2406 data) − (Y interferometer # 2 2407 data)] / 2 × K ₃
K ₁ , K ₂ , and K ₃ are scale factors.

  Each column in the column array has its own XY displacement vector from stage center 2422 such as 2420 and 2421. Next, by combining the following data, the position of each die on the wafer relative to a specific lens barrel is calculated.

1) (X, Y, yawing) position of stage from (0, 0, 0) position 2) (X, Y) displacement vector of specific lens barrel from stage (0, 0, 0) position 3) Stage (X, Y, yawing) position of the upper wafer (measured by imaging several alignment marks on the wafer)

  This wafer position measurement scheme is well known to those skilled in the art, and only additional considerations arise from the use of multiple barrels. An example of a wafer stage suitable for use in a composite beam column assembly is described in US Pat. No. 6,355,994, which is hereby incorporated by reference. An example of a control system that incorporates a wafer position measurement scheme is described in US patent application Ser. No. 10 / 059,048, which is incorporated herein by reference.

  FIG. 25 is a schematic diagram of an embodiment of an optical column and control electronics (compare with FIG. 2A). The source and lens control unit 2510 includes an electronic source chip 201, a source heater filament (not shown), a suppressor electrode 2501, an extraction electrode 202, a first source lens electrode 203, a beam limiting aperture 204, a second A voltage is applied to the source lens electrode 205. The alignment deflector control unit 2512 applies a voltage to the eight electrodes 260 to 267 in the upper alignment deflector / astigmatism corrector 207 and the eight electrodes 268 to 275 in the lower alignment deflector 208. The acceleration column controller 2513 applies a voltage to all the electrodes in the acceleration assembly 209 and the optical system mounting plate 210. The beam blanker driver 2511 applies a voltage to the electrodes 280 to 282 in the upper blanker 277 and the electrodes 283 to 285 in the lower blanker 278. The main field deflector controller 2514 applies four voltages to the electrodes 4001, 4009, 4012, and 4020 in the upper main field deflector 213, and the electrodes 4101, 4109, 4112 in the lower main field deflector 214. , And 4120 are applied with the same four voltages (connected to opposite polarities—see FIGS. 3H and 3I). The subfield deflector / astigmatism corrector controller 2515 supplies a voltage to the eight electrodes 223 to 230 of the subfield deflector / astigmatism corrector 215. The main lens and wafer bias controller 2516 includes nine electrodes 230 to 238 of the focus 1 electrode assembly 216, nine electrodes 240 to 248 of the focus 2 electrode assembly 217, the field free tube 218, and voltage contrast. A voltage is supplied to the plate 220 and the wafer 221. The main lens and wafer bias control 2516 further supplies a common mode voltage to the detector controller 2517. The detector control unit 2517 supplies a bias voltage to the detector assembly 219.

  In the compound barrel assembly, when all of the plurality of beams pass through the same acceleration column, only one acceleration column controller 2513 may be necessary. In some embodiments, a single main field deflector controller 2514 may be available. The other lens barrel controllers 2510 to 2512 and 2515 to 2517 are generally applied to only one lens barrel.

  FIG. 26 is a schematic diagram of one embodiment of a data path and system control electronics. Data block 2601 receives X, Y1, and Y2 data (see FIG. 24) from three interferometers 2405, 2406, and 2407, respectively, and three high-speed data links X2602, (Y1 + Y2) / 2 2603, and (Y1- Y2) / 2 along 2604 to the data path and system control electronics. Data links X2602 and (Y1 + Y2) / 2 2603 are connected to block 2605, which determines the (X, Y) position of the center of wafer stage 2402 relative to the (X, Y) position of the center of the lens barrel array. To do. Data link (Y1-Y2) / 2 2604 connects to block 2606, which determines the yaw angle of wafer stage 2402 relative to column array 2640. Block 2607 includes the (X, Y) coordinates of each of the lens barrels (i, j) in the lens barrel array 2640, determined in advance empirically, and the (X, Y) coordinate data is data link 2609. To the block 2608, where the data supplied by the data links 2609, 2610, and 2641 are combined to determine the (X, Y) position of each barrel relative to the wafer 2401. Block 2614 uses the data from block 2608 to determine the (X, Y) coordinates of the subfield to be drawn by each barrel (i, j). Block 2618 receives the (X, Y) subfield coordinates generated by block 2614 via data link 2619. After being given (X, Y) subfield coordinates, block 2618 receives subfield pattern data from pattern library storage 2615 via data link 2616, where each column contains ( Stripe width 50 μm) / (subfield width 2 μm) = 25 subfield pattern data sets are required. In the illustrated embodiment, since there are 6 × 6 = 36 barrels, the total number of data sets downloaded to the block 2618 via the data link 2616 is 25 × 36 = 900 subfield data sets. . Block 2618 is connected to system control computer 2650 via data link 2617. Subfield pattern data from block 2618 is transmitted to data processor 2621 via data link 2620. The subfield pattern data is supplied from the data processor 2621 to the stripe data buffer 2623 (one for each lens barrel) in 36 parallel data links 2622, and this data is buffered in preparation for drawing. The lens barrel controller 2625 (one for each lens barrel) receives this data via 36 data links 2624. The 36 lens barrel controllers 2625 transmit the subfield data via the data link 2626 to various lens power sources shown in FIG. 26, that is, the source and lens control unit 2510, the alignment deflector control unit 2512, the beam The blanker driver 2511, the main field deflector controller 2514, the subfield deflector / astigmatism corrector controller 2515, the main lens and wafer bias controller 2516, and the detector controller 2517 are supplied.

  FIG. 27A (a) shows how a 50 mm × 50 mm barrel drawing area 2701 can be divided into 50 μm wide stripes 2702 (in the example of a 300 mm wafer and a 6 × 6 barrel arrangement). The number of drawing stripes 2702 in the area 2701 is as follows.

Number of stripes = (Distance between lens barrels) / (Stripe width)
= (50 mm) / (50 μm) = 1000 drawn stripes

  FIG. (B) shows the end of a typical drawing stripe 2702, with individual 2 μm square subfields 2703 shown. The total number of subfields 2703 for each scan is as follows.

# Subfield / scan = (stripe width) / (subfield size)
= (50 μm) / (2 μm) = 25 subfields

  A close-up of one subfield is shown in FIG. (C), with the 1 nm XY address grid enlarged in the lower right corner. Figure (d) shows 1 nm address grids 2704 and 2705. The number of steps in the address grid per subfield is as follows:

#Address step = (Subfield size) / (Address grid)
= (2 μm) / (1 nm) = 2000≡2 ¹¹

To address about 2 ¹¹ address grid steps in one axis, it is necessary to address bits of the next number.
#Address bit = log ₂ (#address step) ≡11 bits

  Since addressing is two-dimensional, a total of 22 address bits is sufficient to position the beam anywhere in any 2 μm square subfield 2703.

  FIG. 27B is an illustration of an embodiment of a method for simultaneously drawing a 50 μm wide stripe with multiple lens barrels positioned in an XY array. FIG. 5A is a perspective view of an array of beams 2706 (each equal to the beam 222 in FIG. 2A) for drawing in parallel on a 300 mm wafer 2401. FIG. Each beam 2706 draws a region 2701.

  FIG. 2B is a close-up of one area 2701 and shows a beam 2706 for drawing the area 2710. Note that scan deflection 2740 maintains beam 2706 always perpendicular to wafer surface 221, thereby providing telecentric scanning. In FIG. 27A, in each scan, 25 subfields 2703 each having a 2 μm square are drawn. The wafer stage 2402 is alternately + Y, -Y, + Y. . . The beam scans in the X direction, but proceeds in a meandering pattern 2711 in the direction. Between scans 2707, the wafer advances stepwise along the X direction to the beginning of the next stripe 2702 to be drawn. FIG. 7C is a detailed view of the scan 2712, showing 25 subfields 2703 and a scan width 2715 of 50 μm. During the drawing of scan 2715, the stage moves a distance 2714 in an "on-the-fly writing" process well known to those skilled in the art.

  FIG. 27C is a diagram showing an example of the correspondence between the arrangement of dies on a general 300 mm wafer 2401 and the XY arrangement of the lens barrel together with the X axis 2730 and the Y axis 2731. The lens barrel drawing area 2701 has XY dimensions determined by the calculation of FIG. 27A, and in this example (6 × 6 lens barrel arrangement), is 50 mm × 50 mm. As shown, the corner corner barrels [(0,0), (0,5), (5,0), (5,5), etc.] draw a very small area on the wafer. . In larger barrel arrays (7 × 7, 8 × 8,..., Etc.), one or more barrels located at each of the four corners of the barrel array can be removed. The lens barrel has an X label 2721 and a Y label 2720, and the label range is 0 to 5 with respect to the 6 × 6 lens barrel arrangement. The center of the lens barrel array is between the lens barrels in the even-numbered array (6 × 6, 8 × 8,... Become the center.

  In this example, the dimensions of the die are assumed to be X dimension = 22 mm and Y dimension = 19.5 mm. There is no edge exclusion region at the edge of the wafer 2401, which results in a total of 143 dies. There is no requirement for the XY interval of the lens barrel array to match the XY interval of the die array on the wafer 2401.

  FIG. 28A is a diagram of integrated circuit (IC) pattern data 2801 divided into subfields 2804 having an X dimension 2805 of 2 μm and a Y dimension 2806 of 2 μm, respectively. Each subfield 2804 has a 1 nm XY address grid 2807 and 2808. Exactly the same subfield and address grid values are used for the beam addressing of the lens barrel and the IC pattern data, and the requirements for data path electronics are to adjust the position of the drawing beam (maximum offset ± 1 μm in XY) The pattern data is put on the drawing grid. The IC pattern data is aligned with the X axis 2809 and the Y axis 2810 corresponding to the X axis 2730 and the Y axis 2731 for the lens barrel arrangement in FIG. 27C.

  FIG. 28B shows an example of the subfield header data format. Subfield addressing requirements for IC pattern data are as follows.

Maximum IC size 64mm x 64mm (4096mm ² )
IC subfield address grid 2μm × 2μm
16-bit X address of subfield in IC 16-bit Y address of subfield in IC

Within each 2 μm square subfield, the pattern addressing requirements are as follows:
Sub-field pattern addressing grid 1nm x 1nm
16-bit X address of subfield pattern 16 bit Y address of subfield pattern

  For each subfield 2804 in the IC pattern data 2801, a subfield header 2811 is defined which is composed of 9 bytes 2817 and includes the following data.

Bytes # 0 to 1 2812 = 2 μm The total number of patterns to be exposed in the square subfield—the maximum number is 2 ¹⁶ −1 = 65535
Bytes # 2 to 3 2813 = X address of subfield (32768 μm to +32768 μm in units of 2 μm)
Bytes # 4 to 5 2814 = Y address of subfield (32768 μm to +32768 μm in 2 μm units)
Byte # 6 2815 = PEC dose level (no correction = 255 to maximum correction = 0—see FIG. 29C)
Byte # 7 2816 = ratio of drawn subfield area (no drawing = 0 to complete drawing = 255—see FIG. 29A)
Byte # 8 2818 = square beam size for drawing this subfield (beam size = value of byte # 8 in nm: 0 nm to 255 nm)

  If it is more efficient to draw subfields with two or more beam sizes, multiple subfield data definitions (each with its own subfield header) are required.

  FIG. 28C is a schematic diagram of an example of a pattern data format for drawing a single flash and a plurality of flashes. A data format 2820 for a single flash requires 5 bytes 2817.

Byte # 0 2821 = pattern type (= 1)
Bytes # 1 to 2 2822 = Flash X address (-1000 nm to 1000 nm)
Bytes # 3 to 4 2823 = Y address of flash (-1000 nm to 1000 nm)

  The data format 2838 for multiple (number = N) flashes requires 4N + 2 bytes 2817.

Byte # 0 2821 = pattern type (= 2)
Byte # 1 2825 = number of flashes (2 to 255)
Bytes # 2 to 3 2826 = X address of flash # 1 (-1000 nm to 1000 nm)
Bytes # 4 to 5 2827 = Y address of flash # 1 (−1000 nm to 1000 nm)
Bytes # 2 to 3 2829 = X address of flash # 2 (-1000 nm to 1000 nm)
Bytes # 4 to 5 2830 = Y address of flash # 2 (−1000 nm to 1000 nm)
Bytes # 2-3 2832 = Flash # 3 X address (-1000 nm to 1000 nm)
Bytes # 4 to 5 2833 = Y address of flash # 3 (-1000 nm to 1000 nm)
. . . . . .
Bytes # 2 to 3 2835 = X address of flash #N (−1000 nm to 1000 nm)
Bytes # 4 to 5 2836 = Y address of flash #N (−1000 nm to 1000 nm)

  FIG. 28D is a schematic diagram of an example of a pattern data format for drawing a single line and a polyline. The data format 2840 for a single line requires 9 bytes 2817.

Byte # 0 2821 = pattern type (= 3)
Bytes # 1 to 2 2841 = X address of line start (-1000 nm to 1000 nm)
Bytes # 3 to 4 2842 = line start Y address (-1000 nm to 1000 nm)
Bytes # 5 to 6 2844 = X address of line end (-1000 nm to 1000 nm)
Bytes # 7 to 8 2845 = Y address of line end (-1000 nm to 1000 nm)

  The data format 2847 for polyline (number = N) requires 4N + 6 bytes 2817.

Byte # 0 2821 = pattern type (= 4)
Byte # 1 2848 = number of lines in polyline (2 to 255)
Bytes # 2 to 3 2849 = X address of line # 1 start (-1000 nm to 1000 nm)
Bytes # 4 to 5 2850 = Y address of line # 1 start (-1000 nm to 1000 nm)
Bytes # 6 to 7 2852 = X address of end of line # 1 (−1000 nm to 1000 nm)
= Line # 2 start X address (-1000nm to 1000nm)
Bytes # 8 to 9 2853 = Y address at the end of line # 1 (−1000 nm to 1000 nm)
= Y address at the start of line # 2 (-1000 nm to 1000 nm)
Bytes # 10 to 11 2855 = X address of end of line # 2 (−1000 nm to 1000 nm)
= Line # 3 start X address (-1000nm to 1000nm)
Bytes # 12 to 13 2856 = Y address at the end of line # 2 (−1000 nm to 1000 nm)
= Y address at the start of line # 3 (-1000 nm to 1000 nm)
. . . . . .
Byte # 4N + 2 to 4N + 3 2858 = X address of line #N end (-1000 nm to 1000 nm)
Byte # 4N + 4 to 4N + 5 2859 = Y address of end of line #N (−1000 nm to 1000 nm)

  FIG. 28E is a schematic diagram of an example of a pattern data format for drawing the entire subfield, drawing a rectangle, or drawing a triangle in the upper right quadrant. The data format 2861 for filling the entire subfield requires only one byte 2817.

  Byte # 0 2821 = pattern type (= 5)

  The data format 2862 for the rectangle requires 9 bytes 2817.

Byte # 0 2821 = pattern type (= 6)
Bytes # 1 to 2 2863 = X address in upper left corner (-1000 nm to 1000 nm)
Bytes # 3 to 4 2864 = Y address in upper left corner (-1000 nm to 1000 nm)
Bytes # 5 to 6 2866 = X address in lower right corner (-1000 nm to 1000 nm)
Bytes # 7 to 8 2867 = Y address in lower right corner (-1000 nm to 1000 nm)

  The data format 2869 for the triangle in the upper right quadrant requires 9 bytes 2817.

Byte # 0 2821 = pattern type (= 7)
Bytes # 1 to 2 2870 = X address of upper left corner (-1000 nm to 1000 nm)
Bytes # 3 to 42871 = Y address in upper left corner (-1000 nm to 1000 nm)
Bytes # 5 to 6 2873 = X address in lower right corner (-1000 nm to 1000 nm)
Bytes # 7 to 8 2874 = Y address in lower right corner (-1000 nm to 1000 nm)

  To draw triangles in the upper left, lower left, and lower right quadrants, the pattern types are 8, 9, and 10, respectively. The data format of the pattern types 8 to 10 is the same as that of the pattern type 7.

  FIG. 28F shows a specific example of a general subfield 2804 including a plurality of drawing pattern types.

Type # 1--a single flash at position (X _sf , Y _sf ) 2901, the subscript “sf” means a single flash.

Type # 2--multiple flashes 2902 at positions ( _Xmf1 , _Ymf1 ) 2903, ( _Xmf2 , _Ymf2 ) 2904, ( _Xmf3 , _Ymf3 ) 2905, and ( _Xmf4 , _Ymf4 ) 2906. The subscript “mfX” represents multiple flashes, and X represents a flash number (1 to 4 in this example).

Type # 3—single line 2907 starting at position (X _sl0 , Y _sl0 ) 2908 and _{ending at} position (X _sl1 , Y _sl1 ) 2909. The subscript “sl” represents a single line.

Started in the type # 4 - position _{(X pl0, Y pl0) 2911} , proceed to the point _{(X pl1, Y pl1) 2912} , then proceeds to the point _{(X pl2, Y pl2) 2913} , the point (X _pl3, Y _pl3 ) polyline 2910 ending at 2914. The subscript “pl” represents a polyline.

Type # 6—A quadrangle 2918 whose upper left is located in (X _r0 , Y _r0 ) 2919 and whose upper right is located in (X _r1 , Y _r1 ) 2920. The subscript “r” represents a rectangle.

Type # 9—Triangle 2915 in the lower left quadrant where the upper left corner is located at (X _tc0 , Y _tc0 ) 2916 and the lower right corner is located at (X _tc1 , Y _tc1 ) 2917. The subscript “tc” represents a triangle type c (lower left quadrant). Other triangle types are type a (upper right quadrant) “ta”, type b (upper left quadrant) “tb”, and type d (lower right quadrant) “td”.

FIG. 29A is a schematic diagram showing a first step in the proximity effect correction (PEC) method for calculating a section drawn in each subfield 2804. For the IC pattern data 2801 defined for the X axis 2809 and the Y axis 2810, the numbers of subfields M _X 2923 and M _Y 2922 along each axis are as follows.

M _X = (X dimension of IC in μm) / (2 μm) (rounded down)
M _Y = (Y dimension of IC in μm) / (2 μm) (rounded down)

The numbering of the subfields is 0 to M _X 2923 along the X axis 2809 and 0 to M _Y 2922 along the Y axis 2810, so the total number of subfields in the IC pattern data ≡M = (M _X + 1) ( _MY + 1). If all subfields are to be rendered with a single beam size, M _datasets = M, where M _datasets is the number of required subfield _datasets . When some subfields are to be drawn due to a plurality of beam sizes, M _datasets > M.

となる。 For example,
N _j = j is the number of subfields to be drawn with j different beam sizes, j = 1, 2,. . . If it is,

It becomes.

  Due to the overhead time required to set each beam size, it is unlikely that more than a few beam sizes will be optimal for any one subfield.

  Next, for each subfield 2804, the total ratio of subfield areas to be drawn is calculated. An example is as follows.

  Since the entire subfield (2, 1) 2925 is drawn [can be part of a bonding pad], the ratio = 1.0, and the value of byte # 7 2816 in the subfield header 2811 = 255. .

  In the subfield (5, 4) 2926, since 70% of the subfield area is densely drawn, the ratio = about 0.7, and the value of byte # 7 2816 in the subfield header 2811 = 179≈ 0.7 × 255.

  Since the subfield (11, 7) 2927 is drawn sparsely by 15%, the ratio is about 0.15, and the value of byte # 7 2816 in the subfield header 2811 = 38≈0.15 × 255.

When the total ratio to be drawn is calculated for each subfield 2804, the data is stored in byte # 7 2816 of the subfield data header. If multiple beam sizes are used for any particular subfield, the value of the total area to be drawn is stored in byte # 7 2816 of each subfield data set, where the total area to be drawn is It is the sum of the values of byte # 7 2816 in all data sets for that particular subfield 2804. Note that the calculation of the total area to be drawn for each subfield 2804 is completely independent of the area to be drawn in the other subfields 2804. FIG. 29B is a schematic diagram of a second step in the PEC scheme that calculates the total backscattered electron (BSE) dose for subfield 2937, assuming that there is no PEC correction for the primary beam dose in any subfield 2804. Indicates. IC pattern data 2801 is the same as FIG. 29A. For each subfield (i, j) [i = 0 to M _X and j = 0 to M _Y ], the weighted sum of backscattered electron doses from all adjacent subfields is calculated using the illustrated relative intensity graph. calculate. As radius 2931 increases away from subfield (i, j) 2937, the contribution decreases as shown by curve 2934 plotted against relative intensity scale 2932. The BSE distribution is isotropic and the BSE contribution to the total dose in subfield 2937 is considered to be the same around circle 2938. The maximum possible backscatter contribution is η2936, the backscattered electron coefficient, which is related to the eight subfields directly surrounding subfield (i, j) and the subfield (i, j) itself. The contribution from each subfield is proportional to the drawing ratio of that subfield calculated in step 1 (FIG. 29A). The formula for the total backscatter dose d (i, j) in subfield (i, j) is:

R (m, n; i, j) ≡√ [(m−i) ² + (n−j) ² ] (2 μm) = radius from (m, n) to (i, j) s [R (m , N; i, j)] = relative intensity in subfield (i, j) due to BSE scattering in subfield (m, n) f (m, n) = area drawn in subfield (m, n) P (m, n) = currently 1 (changed in next step)
K = scale factor

  Note that the sum of m and n includes subfield (i, j) because backscattered electrons from inside the subfield (i, j) being calculated also contribute to the BSE background dose.

  FIG. 29C shows a schematic diagram of the third step in the PEC method for calculating the total dose of each subfield by combining the primary beam dose and the BSE dose. The physical process that occurs is within each pattern being drawn in subfield (i, j), and there are three types of contribution to the resist dose.

(1) Primary electrons in the drawing beam for the pattern being drawn (2) Backscattered electrons from other patterns in subfield (i, j) (3) Backscattered electrons from adjacent subfield (m, n)

  FIG. 29C shows how the writing dose can be reduced to correct contributions (2) and (3), a process known to those skilled in the art called proximity effect correction (PEC). The dose profile 2942 on the left (plotted relative axis 2941) corresponds to the case where no PEC is required since it is the minimum BSE background dose. To maximize process freedom, it is useful to adjust the writing dose so that the resist exposure dose occurs at the point of maximum slope of the dose profile, in this case, point 2945. This minimizes the effect of changes in resist processing or writing beam current on line width changes. The drawing dose 2951 is twice the level of the required exposure 2950 in the absence of PEC.

  The middle example shows the case of a medium level BSE background 2948 which is about 30% of the exposure dose. To maintain process freedom, as shown, the drawing dose 2943 is reduced by the same percentage or twice the BSE background 2948, thereby maintaining the exposure point 2946 indicated by the dashed line. is doing. Since the drawing dose 2951 is twice the exposure amount 2950, the exposure amount is reduced by a double amount.

  The example on the right represents dense lines and spaces, with the largest BSE background. In this example, the BSE background 2949 is about 60% of the exposure dose 2950 and requires a 60% reduction in the writing dose.

All i = 0,. . . M _X and j = 0,. . . After BSE background against M _Y has been determined, you can execute dose correction at the first pass. A correction coefficient p (i, j) is calculated for all subfields (i, j).
p (i, j) = 1-2d (i, j)

  Then, using these new values for p (i, j), recalculating all values of d (i, j) results in a small value for d (i, j), which This time a large value for p (i, j) occurs. Thus, the process of finding a self-consistent solution for p (i, j) oscillates, but usually converges within a few cycles. After the value for p (i, j) no longer changes due to some pre-set limit, the process ends and the calculated value for p (i, j) is multiplied by 255 and stored in byte # 6 2815.

Typical electrode voltage for the lens barrel of FIG. 2A

  One possible drawing strategy for drawing a series of stripes on a resist-coated wafer using one or more electro-optic columns is shown in FIG. 27B. As an example, assume that a 6 × 6 array of identical barrels is positioned above a 300 mm wafer. The distance between the lens barrels can be determined as follows.

Distance between lens barrels = (wafer diameter) / √ (number of lens barrels)
= (300 mm) / √ (6 × 6) = (300 mm) / 6 = 50 mm

  Each lens barrel needs to draw only a 50 mm × 50 mm square area of the wafer. When a specific lens barrel completes drawing of its own region, each of the other 35 lens barrels simultaneously completes drawing of their own region, completing the wafer. As shown in FIG. 27A, the region to be drawn by each lens barrel is divided into a series of 50 μm-wide parallel stripes, and here, the total number of stripes is determined to be 1000.

Within each stripe, the lens barrel draws on the resist in a series of “flashes” and each flash uses a high current density beam over the resist with a given sensitivity (assuming 5 μC / cm ² ). To expose the square area. The time per flash is determined by the beam current density at the outer edge of the shaped beam. As shown in FIG. 21C, the beam current density ranges from 2000 A / cm ² (30 nm and 120 nm square beams) to 2500 A / cm ² (approximately 80 nm square beams) and up to 3000 A / cm ² (40 nm square beams). It becomes.

Time per flash = (resist sensitivity) / (beam current density)
= (5 μC / cm ² ) / (3000 A / cm ² ) = 1.67 ns
= (5 μC / cm ² ) / (2500 A / cm ² ) = 2.00 ns
= (5 μC / cm ² ) / (2000 A / cm ² ) = 2.50 ns
These calculated flash times are graphed in FIG. 21C.

  Each flash requires a setup time of 1 ns because it is necessary to deflect the beam to a new position on the wafer surface. Therefore, the total pixel drawing time is determined as follows.

Total pixel drawing time = (number of flashes / subfield) [(time / flash) + (setup time)]
Here, the subfield is assumed to be 2.0 μm × 2.0 μm.

  The overall pattern density determines the average number of flashes / subfield, which remains as a variable in FIG. As expected, high pattern density requires, on average, a high flash / subfield 3001, resulting in low pattern formation throughput 3002.

As the stage moves along the length of the stripe (or in the + Y and −Y directions 2702—see FIG. 27B), the electron beam spans the width 2715 of the stripe and is perpendicular 2740 (along the X axis). And electrostatic scanning. With the assumed 50 μm stripe, the total number of subfields / scans is determined to be 25 in FIG. 27A. Including the 10 ns scan regression time, the total time per scan is:
Time / scan = (number of subfields / scan) (time / subfield) + (scan regression time)

In this drawing strategy, scanning is accomplished using the deflection of the electrostatic beam relative to the optical axis = ± (scan width) / 2. The total number of scans required to draw the length of the entire stripe is determined as follows (where stripe length = lens barrel spacing).
Number of scans = (barrel interval) / (subfield size)
= (50 mm) / (2.0 μm) = 25000

One strategy for moving the stage during drawing is to move it continuously under the drawing beam (one per column). In this case, the stage speed is as follows.
Stage speed = (subfield size) / (time / scan)

With the number of flashes / subfield as a variable, the stage speed ranges from approximately 550 mm / s (limited by the stage acceleration of 1 g = 9800 mm / s ² and the length of the stripe 2702 of 50 mm) to less than 50 mm / s. It decreases as the number / subfield increases or the beam current density decreases.

  Finally, each wafer to allow wafer transfer, global alignment, local alignment, and all other functions required between the end of drawing one wafer and starting drawing the next. In contrast, an overhead of 45 seconds was assumed. Therefore, the total time per wafer is composed of the writing time of all stripes + the stage turnaround time for all stripes + the overhead of the wafer. Throughput is inversely proportional to the total time per wafer.

FIG. 30 shows a graph of the computational throughput 3002 versus the average number of flashes 3001 per subfield in a drawing module with various numbers of barrels from 6 × 6 to 10 × 10, and the required exposure. The current 3003 is assumed to be 3000 A / cm ² , which corresponds to an optimized 40 nm square beam. The subfield is assumed to be 2 μm square, and the average number of flash / subfields is in the range of 40 to 360. A curve 3005 corresponds to a 6 × 6 barrel arrangement, and each barrel draws an area of 50 mm × 50 mm. A curve 3006 corresponds to a 7 × 7 column arrangement, and each column draws 42.9 mm × 42.9 mm or less. A curve 3007 corresponds to an 8 × 8 barrel arrangement, and each barrel draws 37.5 mm × 37.5 mm or less. A curve 3008 corresponds to a 9 × 9 barrel arrangement, and each barrel draws 33.4 mm × 33.4 mm or less. Finally, the curve 3009 corresponds to a 10 × 10 column array, and each column draws 30 mm × 30 mm or less. For a large number of flashes (over 160 / subfield), the throughput is approximately inversely proportional to the average number of flashes, whereas for a small number of flashes in region 3004, the throughput is the maximum stage acceleration (9800 mm / s ² = 1g)). If the entire 2 μm square subfield is to be drawn with a 40 nm beam, this requires an extraordinary number of flashes, ie [(2 μm) / (40 nm)] ² = 2500 flashes, which is a variable size beam. The need for is clearly shown.

  Similar graphs can be plotted using the current density shown in FIG. 21B for 30 nm, about 80 nm, and 120 nm beams. For all beam sizes other than the (optimized) 40 nm square beam, the throughput is slightly lower than that shown in FIG.

For a 120 nm square beam with 125 nm spacing between centers, a 2 μm square subfield can be drawn with [(2 μm) / (125 nm)] ² = 256 flash, which is the required 2.50 ns dwell. Even in anticipation of time, it is an acceptable number. In many barrels that are possible using the barrel design described here, at least one barrel always draws a bonding pad while the other barrel draws an area with fine features. Therefore, it is important that the entire subfield can be completely drawn in a reasonable time. The drawing strategy requires that all barrels continue to be synchronized while drawing, so the system needs to be able to maintain the drawing speed of all the barrels in order to maintain throughput.

  The invention described above can be extended to include the case where there are two beam limiting apertures in a single electron beam column. The two apertures can be used together to limit the beam at the substrate. The two apertures are both centered on the optical axis and are separated from each other in the axial direction. As described above, additional beam pattern limited apertures can be added to provide multiple apertures in a single barrel.

  Multiple beam limiting apertures may be incorporated into the electron beam column to allow selection of various beam shapes on the substrate. The aperture can be attached to a single aperture blade and moved on axis as needed, or it can be attached to a separate square aperture blade in different parts of the barrel, and between the upper and lower deflection optics Position the beam close to the optical axis so that the desired aperture can be selected by deflecting the beam (using the upper deflection optics) and using the lower deflection optics (after passing the selected aperture) For example, re-biasing on the optical axis can be performed.

  The electron optics may be configured to allow the electron beam to expand, contract, or deform so that the beam at the substrate is an expanded, contracted, or deformed shape determined by the beam limited aperture. An example of a useful deformation of a square beam is to shrink along one axis to form a triangle. Other variations include conversion from a square beam to a beam formed as a parallelogram. Electro-optical elements that can be used to achieve these effects include quadrupole and octupole lenses.

  Although the numerical method is adopted in the design procedure of the beam pattern limited aperture 212 described in the present specification, another analysis method is also possible. Referring to FIG. 1, the function of blocks 102-106 does not change, but in block 108, instead of numerical electron beam tracing, an analytical model of the behavior of the optical system is generated, which is then used to Determine which electron beam of the circular beam passes through the desired patterned beam profile and which electron beam passes outside the desired patterned beam profile. Although the analytical model of the optical system is generated using electron beam tracing, this analysis technique has the opportunity to improve the design of the PBDA 212 by reducing the impact of numerical errors in the calculation of electron beam tracing. I will provide a. This improvement is achieved by smoothing the analytical model derived from electron beam tracing calculations to smooth out small (nm scale) variations in the value of the electron beam intercept at the wafer surface 221. After generating the PBDA design at block 112 using this analysis procedure, the remaining PBDA design procedure described in FIG. 1 is the same.

  In the design illustrated in the examples herein, the practical range of beam sizes is approximately 30 nm to 120 nm. In other electro-optic designs that implement the present invention, the beam size range can be from 15-20 nm to 400-500 nm.

6 is a flowchart of a design procedure for an electron optical system that utilizes a beam pattern limited aperture for use in generating a high intensity patterned electron beam. 3 is a cross-sectional view of an electron column that utilizes a non-circular aperture to generate a high current density patterned electron beam. FIG. It is sectional drawing of the bottom part of the electron optical lens-barrel of FIG. 2A. It is various figures which show the electron beam near the source chip 201. FIG. FIG. 6 is various views showing electron beams in the upper alignment deflector / astigmatism corrector 207. FIG. 6 is various views showing an electron beam in the lower alignment deflector 208. FIG. 6 is various views showing an electron beam immediately above the beam trimming aperture 276. It is various figures which show the electron beam in a beam blanker. It is a figure which shows the electron beam in the lower beam blanker 278. FIG. FIG. 4 is various views showing an electron beam immediately above the beam limiting aperture 212. FIG. It is a figure which shows the electron beam inside an upper main field deflector 213, and the outline of a main field deflector. It is a figure which shows the electron beam inside a lower main field deflector. It is a figure which shows the electron beam inside a subfield deflector / astigmatism corrector 215. FIG. 4 is a diagram showing an electron beam inside a focus 1 electrode assembly 216. It is a figure which shows the electron beam inside a focus 2 electrode assembly. It is a figure which shows the electron beam inside the field free pipe | tube 218. FIG. 2B is a graph of the first circular beam profile in the wafer plane centered on the optical axis (0,0) before inserting the first patterned aperture into the lens barrel of FIG. FIG. 2B is a graph of a first circular beam profile at the wafer plane, centered off +12.5 μm from the optical axis, prior to insertion of the first patterned aperture into the lens barrel of FIG. FIG. 2B is a graph of the first circular beam profile in the wafer plane, centered off +25 μm from the optical axis, prior to insertion of the first patterned aperture into the barrel of FIG. 2A, with the desired square profile superimposed on the graph. FIG. FIG. 2B is a diagram showing an ideal beam profile in the wafer plane around the optical axis (0, 0) after the patterned aperture is inserted into the lens barrel of FIG. 2A. It is a figure which shows the graph of the ideal beam which the beam pattern limited aperture 212 permeate | transmits in the lens barrel of FIG. 2A. It is a figure which shows the graph of the ideal beam which the beam pattern limited aperture 212 shields in the lens barrel of FIG. 2A. It is a figure which shows the graph of the actual beam which the beam pattern limited aperture 212 permeate | transmits in the lens barrel of FIG. 2A. It is a figure which shows the graph of the actual beam which the beam pattern limited aperture 212 shields in the lens barrel of FIG. 2A. FIG. 2 shows a beam pattern limited aperture (PBDA) designed to generate a high current density square electron beam and an electron beam with a beam energy of 5000 eV on a wafer. It is a figure which shows the mapping of the electron beam which passes the center part of PBDA with respect to the electron beam intersection of a wafer surface. It is a figure which shows the mapping of the electron beam which passes the outer part of PBDA with respect to the electron beam intersection of a wafer surface. FIG. 6 shows various beam positions A to D on the wafer surface used for calculating the beam profile. FIG. 10 is a diagram showing a calculated exposure amount by one flash of a 40 nm square electron beam when the beam is at the position A in FIG. 9. FIG. 10 is a diagram showing a calculation exposure by one flash of a 40 nm square electron beam when the beam is at position B in FIG. 9. FIG. 10 is a diagram showing a calculated exposure amount by one flash of a 40 nm square electron beam when the beam is at a position C in FIG. 9. FIG. 10 is a diagram showing a calculated exposure amount by one flash of a 40 nm square electron beam when the beam is at a position D in FIG. 9. FIG. 10 shows the calculated exposure dose by three flashes (all flashes of FIG. 10A) of adjacent 40 nm square electron beams in an “L” pattern at 40 nm intervals when the beam is at position A in FIG. 9. FIG. 10 shows the calculated exposure with two overlapping flashes of a 40 nm square electron beam and a separate single flash (all flashes in FIG. 10A) when the beam is at position A in FIG. FIG. 10 shows graphs of calculated beam current density for a single 40 nm square beam (of FIG. 10A) and a single 40 nm FWHM Gaussian beam when the beam is at position A in FIG. When the beam is at position A in FIG. 9, a graph of the calculated beam current density for three combined 40 nm square beams (all flash in FIG. 10A, 40 nm spacing) and three composite 40 nm FWHM Gaussian beams also at 40 nm spacing. FIG. When the beam is at position A in FIG. 9, the calculated beam current density graph for three combined 40 nm square beams and three separate 40 nm square beams (all flash in FIG. 10A) with a beam spacing of 40 nm. FIG. It is a figure which shows the graph of the calculation beam current density in three 40nm FWHM Gaussian beams and three separated 40nm FWHM Gaussian beams by setting all beam intervals as 40 nm. FIG. 5 illustrates a beam scanning method that can be used to set up an optical system that generates an optimized square beam profile. FIG. 17B illustrates a calculated line scan for the various scan directions of FIG. 17A, showing a possible method for setting an optimized 40 nm square beam profile. FIG. 8B is a diagram showing a calculated exposure amount by one flash of a 30 nm square electron beam when the beam is at the position A in FIG. 9 using the beam pattern limiting aperture 212 of FIG. 8A. FIG. 8B is a diagram showing a calculated exposure dose by one flash of a square electron beam of about 80 nm when the beam is at the position A of FIG. 9 using the beam pattern limiting aperture 212 of FIG. 8A. FIG. 8B is a diagram illustrating a calculated exposure dose by one flash of a 120 nm square electron beam when the beam is at position A in FIG. 9 using the beam pattern limiting aperture 212 of FIG. 8A. FIG. 6 is a graph showing source lens and main lens focusing voltage versus desired square beam size. FIG. 5 is a graph of chip half-angle and beam current at the wafer for a desired square beam size. It is a figure which shows the graph of the beam flash time with respect to a desired square beam size, and the assumption exposure current density on the wafer (resist sensitivity is assumed 5 microC / cm < ² >). FIG. 6 is a graph showing a magnification of a virtual source on a wafer surface against a desired square beam size. FIG. 6 shows a beam blanking strategy that can be used to change the exposure for each subfield to achieve proximity effect correction. FIG. 6 is a cross-sectional close-up side view of the main lens showing the settings calculated for the focus 1 and focus 2 octupole voltages. It is a schematic diagram of one Embodiment of a wafer stage and a position sensor. It is the schematic of embodiment of a lens-barrel and its control electronics. 2 is a schematic diagram of one embodiment of a data path and system control electronics. FIG. It is a figure which shows how a lens-barrel drawing area | region can be divided | segmented and can be made into a 50 micrometer width stripe each subdivided into a 2 micrometer square subfield which has XY address grid of 1 nm. FIG. 6 is a diagram of an embodiment of a method for simultaneously drawing a 50 μm wide stripe by a number of lens barrels positioned in an XY array. It is a figure which shows the example of the correspondence between arrangement | positioning of die | dye on a general 300 mm wafer, and XY arrangement | sequence of a lens-barrel. FIG. 5 is a diagram of integrated circuit (IC) pattern data divided into 2 μm square subfields with a 1 nm XY address grid. It is a schematic diagram of the example of a subfield header data format. It is a schematic diagram of the example of the pattern data format for drawing a single flash and a some flash. It is a schematic diagram of the example of the pattern data format for drawing a single line and a polyline. It is a schematic diagram of the example of the pattern data format for drawing the whole subfield, a rectangle, or a triangle of the upper right quadrant. It is a figure which shows the specific example of the general subfield containing a some drawing pattern type. It is a schematic diagram which shows the 1st step in a proximity effect correction (PEC) system which calculates the division drawn in each subfield. FIG. 5 is a schematic diagram of a second step in the PEC scheme that calculates the total backscattered electron (BSE) dose for each subfield, assuming no PEC correction for the primary beam dose. It is a schematic diagram of the 3rd step in a PEC system which calculates the total dose of each subfield by combining a primary beam dose and a BSE dose. Assuming a required exposure current of 3000 A / cm ² , a graph of the computational throughput against the average number of flashes per subfield in one drawing module with various numbers of barrels from 6 × 6 to 10 × 10 FIG.

Explanation of symbols

201 Electron beam source chip 202 Extraction electrode 203 First source lens electrode 205 First source lens electrode 207 Upper alignment deflector / astigmatism corrector 208 Lower alignment deflector 212 Beam pattern limited aperture 214 Lower main field deflector 215 Sub Field deflector / astigmatism corrector 216 Focus 1 electrode assembly 217 Focus 2 electrode assembly 276 Beam trimming aperture 2101 Source lens (V)
2103 Square beam size (nm)
2102 Main lens (V)
2510 Source and Lens Control Unit 2512 Alignment Deflector Control Unit 2513 Acceleration Column Control Unit 2511 Beam Blanker Driver 2514 Main Field Deflector Control Unit 2515 Subfield Deflector / Astigmatism Corrector Control Unit 2516 Main Lens and Wafer Bias Control Unit 2517 Detection Controller 2601 Wafer stage interferometer 2650 System control computer 2615 Pattern library storage 2618 Controller 2623 Stripe data buffer 2625 Lens barrel controller 2621 Data processor 2640 6 × 6 lens array 2510 Source lens 2512 Alignment deflector 2511 Blanker 2514 Main field deflection 2515 Subfield deflector and astigmatism corrector 2516 Main lens 2517 Detector

Claims (43)

A lithography tool for patterning a resist-coated substrate,
A charged particle source configured to generate a charged particle beam;
A first lens positioned below the charged particle source and configured to turn the charged particle beam into a substantially layered charged particle beam;
A stage positioned below the first lens and carrying the resist-coated substrate;
A second lens positioned between the first lens and the stage and configured to focus the substantially layered charged particle beam onto a surface of the resist-coated substrate;
Charge in the substantially layered charged particle beam positioned between the first lens and the second lens and unable to focus to a predetermined beam profile on the surface of the resist-coated substrate by the second lens A lithography tool comprising: a beam pattern limited aperture configured to shield a majority of the particles.   The lithography tool of claim 1, further comprising a beam blanker positioned between the first lens and the beam pattern limiting aperture and blanking the substantially layered charged particle beam.   The lithography tool of claim 2, wherein the beam blanker is a dual deflection beam blanker configured to project an effective blanking surface back to the position of the virtual source.   The lithography tool according to claim 2, further comprising a beam trimming aperture positioned between the first lens and the beam blanker.   The beam trimming aperture provides equal exposure time for all aperture areas of the beam pattern limited aperture when sweeping the substantially layered beam onto the beam pattern limited aperture during blanking. The lithography tool of claim 4, wherein the lithography tool is configured and the beam trimming aperture is configured to minimize blanking time.   The lithography tool according to claim 5, wherein the beam trimming aperture has an opening having a square shape, and the beam pattern limiting aperture has an opening having a shape derived from a square.   The lithography tool according to claim 1, further comprising a beam deflector positioned between the beam pattern limiting aperture and the second lens.   The lithography tool of claim 7, wherein the beam deflector is a double deflector configured to allow telecentric scanning of the beam.   The lithographic tool of claim 8, wherein the second lens is configured such that an effective axis of the second lens is movable paraxially with the beam being scanned.   The beam pattern limiting aperture is further adapted to transmit most of the charged particles in the substantially layered beam that can be focused to the predetermined beam profile at the surface of the resist-coated substrate by the second lens. The lithography tool according to claim 1, wherein the lithography tool is configured as follows.   The lithography tool according to claim 1, wherein the beam pattern limiting aperture comprises a patterned conductive material.   The lithography tool of claim 1, wherein the beam pattern limiting aperture comprises a patterned thick film supported by a continuous charged particle permeable film.   The lithography tool according to claim 1, wherein the charged particles are electrons.   The lithography tool of claim 1, wherein the beam pattern limiting aperture is configured to generate a non-circular beam.   The lithography tool of claim 1, wherein the beam pattern limiting aperture is configured to generate a square beam. A lithography tool for patterning a resist-coated substrate,
A charged particle source configured to generate a charged particle beam;
A first lens positioned below the particle source and configured to make the charged particle beam a substantially layered charged particle beam;
A stage positioned below the first lens and carrying the resist-coated substrate;
A second lens positioned between the first lens and the stage and configured to focus the substantially layered charged particle beam onto a surface of the resist-coated substrate;
Of charged particles in the substantially layered beam that are positioned between the first lens and the second lens and cannot be focused to a predetermined beam profile on the surface of the resist-coated substrate by the second lens. A lithography tool comprising a plurality of beam pattern limiting apertures configured to shield a majority.   The plurality of beam pattern limiting apertures further transmit most of the charged particles in the substantially layered beam that can be focused to the predetermined beam profile on the surface of the resist-coated substrate by the second lens. The lithography tool of claim 16, wherein the lithography tool is configured to:   All of the plurality of beam pattern limiting apertures are axially separated from each other along the optical axis of the tool, and all of the apertures act on the beam to form the predetermined surface on the surface of the resist-coated substrate. The lithography tool of claim 16, wherein the lithography tool generates a beam profile of:   The lithography tool of claim 18, wherein the plurality of beam pattern limiting apertures are two beam pattern limiting apertures.   The plurality of beam pattern limited apertures are positioned on a single aperture blade, and the aperture blade is configured to allow insertion of any of the plurality of beam pattern limited apertures into the beam. A lithography tool according to claim 16.   A beam pattern limited aperture in a charged particle column, wherein the aperture and the column are configured to provide an N: 1 mapping of points in the plane of the aperture to the object plane of the column , N is an integer greater than 1, and the aperture is configured to generate a non-circular beam.   The beam pattern limited aperture according to claim 21, wherein the aperture is configured to exclude a majority of charged particles that do not contribute to a desired beam pattern at the object plane of the barrel.   The aperture limited to a beam pattern according to claim 21, wherein the aperture is positioned below a beam blanker in the lens barrel.   The beam pattern limited aperture according to claim 23, wherein the beam pattern limited aperture also functions as a blanking aperture.   The beam pattern limited aperture of claim 21, wherein the beam pattern limited aperture is configured to generate a square beam at the object plane.   The beam pattern limited aperture according to claim 21, wherein N is equal to three. A method for designing a beam pattern limited aperture in a charged particle column for generating a shaped charged particle beam,
Calculating a charged particle trajectory of the charged particle column;
Determining whether the trajectory falls within a desired beam profile at the object plane of the charged particle column;
Defining an ideal beam pattern limited aperture that shields all of the trajectories that do not contribute to the desired beam profile at the object plane.   28. The method of claim 27, wherein the calculating step includes calculating a plurality of sets of trajectories, each set of trajectories corresponding to a different beam position within a scan field of the object plane.   The method further comprises mapping intersections of the plurality of sets of trajectories and the plane of the beam pattern limiting aperture to define a charged particle transmission region, the defining step further comprising: 30. The method of claim 28, including the step of incorporating into an ideal beam pattern limiting aperture.   The method further comprises the step of fabricating a feasible beam pattern limited aperture, wherein the feasible aperture is very close to the ideal beam pattern limited aperture and does not contribute to the desired beam profile. 28. The method of claim 27, wherein a majority of is shielded by the realizable aperture.   The method further comprises the step of fabricating a feasible beam pattern limited aperture, wherein the feasible aperture is very close to the ideal beam pattern limited aperture and contributes to the desired beam profile. 28. The method of claim 27, wherein a majority of is made transparent through the feasible aperture.   32. The method of claim 31, wherein a majority of the charged particle trajectories that do not contribute to the desired beam profile are shielded by the feasible aperture.   32. The method of claim 31, wherein the realizable aperture has additional structures that are not present in the ideal aperture, and the structures provide mechanical integrity.   32. The method of claim 31, wherein the realizable aperture comprises a patterned conductive material.   32. The method of claim 31, wherein the feasible aperture comprises a patterned thick film supported on a continuous charged particle permeable membrane.   28. The method of claim 27, wherein the charged particle is an electron.   28. The method of claim 27, wherein the calculating step comprises using a ray tracing numerical method to generate the charged particle trajectory.   28. The method of claim 27, wherein the calculating includes using an analytical model of the barrel to generate the charged particle trajectory. A method for optimizing the position of a beam pattern limited aperture in a charged particle column,
(A) calculating a charged particle trajectory of the charged particle column; (b) determining whether the trajectory falls within a desired beam profile at the objective surface of the charged particle column;
(C) defining an ideal beam pattern limited aperture that blocks all of the trajectories that do not contribute to the desired beam profile at the object plane;
(D) producing a feasible beam pattern limited aperture that is very close to the ideal beam pattern limited aperture and shields most of the charged particle trajectories that do not contribute to the desired beam profile; ,
(E) performing steps (a) to (d) for various positions of the beam pattern limited aperture in the barrel;
(F) selecting an optimal position of the beam pattern limited aperture such that a beam profile generated by the feasible beam pattern limited aperture at the object plane is closest to the desired beam profile at the object plane; A method comprising:   40. The method of claim 39, wherein the step of selecting includes consideration of manufacturability of the feasible aperture.   40. The method of claim 39, wherein the step of selecting includes a consideration of a current density of the charged particle beam at the object plane.   40. The method of claim 39, wherein the step of selecting includes consideration of a current density profile of the charged particle beam at the object plane.   40. The method of claim 39, wherein the step of selecting includes consideration of heating of the feasible aperture by charged particle collisions.

Images