Pattern data processing for programmable direct-write apparatus

Abstract

To provide an architecture of a pattern data processing which can take advantage of high redundancy and maintains pattern fidelity.SOLUTION: In a charged-particle multi-beam apparatus (1A, 1B), a desired pattern is provided as input pattern data (INPDAT) in a vector format and processed through a pattern data processing flow. A data preprocessing system (101) receives the input pattern data (INPDAT) and preprocesses the input pattern data independently of a writing process using writing parameter data provided to the data preprocessing system, and writes intermediate pattern data (IMDAT) thus obtained to a data storage (11M). In the writing process, its writing control system (14A, 14B) reads the intermediate pattern data from the data storage (11M), converts the data into pattern streaming data (SBUF), and performs real-time transmission of the pattern streaming data to a writing apparatus.SELECTED DRAWING: Figure 7

Description

[Technical Field]

[0001] This application claims the benefit of priority under the Paris Convention for European Patent Application No. 21175588.9, filed on 25 May 2021, the entire contents of said application incorporated herein by reference.

[0002] The present invention relates to a drawing process in a programmable charged particle multibeam apparatus for processing (particularly nanopattern formation or semiconductor lithography) or inspecting a target. More specifically, the present invention relates to a method for performing a drawing process of a desired pattern on a target using at least one charged particle multibeam apparatus, wherein the desired pattern is provided as input pattern data encoded in vector form, the apparatus includes an illumination system, a pattern definition system and a projection optical system, the illumination system is configured to generate a charged particle beam and direct the charged particle beam to a pattern definition system, the pattern definition system is configured to form the charged particle beam into a plurality of subbeams representing the pattern (a plurality of subbeams representing the pattern from the charged particle beam), and the projection optical system is configured to project the plurality of subbeams onto the target, the method and the multibeam apparatus of this type. [Background technology]

[0003] This type of lithography process and multibeam configuration is described in the applicant's US7,214,951, US8,183,543 and US9,443,699B2.

[0004] The applicant has developed several suitable solutions and technologies in the field of charged particle multibeam lithography and nanopatterning and related technologies, including: CHARPAN (charged particle nanopatterning) when using ion multibeams; eMET (electron mask exposure tool) or MBMW (multi-beam mask writer) for mask writing when using electron multibeams; and PML2 (Projection Mask-Less Lithography) for direct writing lithography on substrates, particularly silicon substrates. In this regard, relevant patent documents under the applicant's name include US7,199,373, US7,214,951, US8,304,749, US8,183,543, and US8,222,621. The disclosures of these documents are incorporated herein by reference. [Prior art documents] [Patent Documents]

[0005] [Patent Document 1] US7,214,951 [Patent Document 2] US8,183,543 [Patent Document 3] US9,443,699B2 [Patent Document 4] US7,199,373 [Patent Document 5] US8,304,749 [Patent Document 6] US8,222,621 [Overview of the Initiative] [Problems that the invention aims to solve]

[0006] During pattern drawing on targets such as 6" (inch) mask blanks or silicon wafer substrates, non-ideal situations must be considered during the multibeam lithography process to achieve a mask or wafer with the desired feature and pattern fidelity. Possible issues to consider include, for example, substrate distortion due to processing, beam array field distortion that cannot be corrected by optical deflection means, "loading effects" in the subsequent etching process, or charge-induced influences. These effects are simulateable and / or measurable. In particular, the corresponding data can be provided in advance when creating data files for patterns to be drawn by the multibeam lithography machine ("data type correction"), or some or all of these effects can be calculated based on a model of the effect ("model type correction"). In most cases, it will be found that corrections need to be performed on portions of substrate area ("substrate segments") smaller than the size (area) of the beam field array in order to meet the high demands of current and future nanolithography technology.

[0007] For industrial applications, such requirements are expressed as performance requirements relating to achieving small critical dimensions (CD) and small registration errors (REG), and, in particular, achieving nanometer-level 3σ or 6σ variation of local CD uniformity (LCDU) and local registration error (LREG) within small fields (e.g., the area of ​​an MBW beam array field), and achieving nanometer-level 3σ or 6σ variation of global CD uniformity (GCDU) and global registration error (GREG) across the entire MBW lithography field on a substrate (e.g., a 6" mask blank or a 300nm silicon wafer).

[0008] A further important aspect of industrial applications is throughput in mass production, particularly in applications of maskless lithography or direct writing to silicon wafers. A simple approach of scaling up the supply of the required number of charged particle multibeam lithography machines to increase throughput by several orders of magnitude would be problematic, as it ignores the high redundancy (duplication) of the repetition of processing steps, especially with respect to the pattern pre-processing steps.

[0009] From the above perspective, the object of the present invention is to overcome the above-mentioned problems, and in particular to provide a pattern data processing architecture that can utilize high redundancy and address the above-mentioned adverse effects that may affect pattern reliability, while at the same time achieving efficient handling of computer computation and time requirements. This also includes appropriate handling of real-time effects. [Means for solving the problem]

[0010] From the first perspective of the present invention, A method is provided for performing a process of drawing a desired pattern on a target using at least one charged particle multibeam apparatus. In the method, the desired pattern is provided as input pattern data encoded in vector form, the apparatus includes an illumination system, a pattern delimiting system and a projection optical system, the illumination system is configured to generate a charged particle beam and direct the charged particle beam to the pattern delimiting system, the pattern delimiting system is configured to form the charged particle beam into a plurality of subbeams representing the pattern (a plurality of subbeams representing the pattern from the charged particle beam) and the projection optical system is configured to project the plurality of subbeams onto the target. The above method consists of at least the following steps, which are performed independently of the drawing process in the data preprocessing system: Receiving the aforementioned input pattern data, In order to obtain intermediate pattern data, the input pattern data is preprocessed using the drawing parameter data provided to the data preprocessing system. The intermediate pattern data is written to a data storage device. Includes, The method further includes at least the following steps, performed by at least one drawing control system associated with the at least one (charged particle multibeam) device during the initialization and / or execution of the drawing process using the at least one (charged particle multibeam) device: Reading the aforementioned intermediate pattern data from the data storage device, To obtain pattern streaming (real-time transmission) data, the intermediate pattern data is converted. The pattern streaming data is transmitted (in real time) to at least one pattern definition system of the at least one device, wherein each pattern definition system forms the charged particle beam of each of the at least one device into the plurality of subbeams (from the charged particle beam of each of the at least one device to the plurality of subbeams) during the drawing process in accordance with the pattern streaming data. It is characterized by including (Form 1). A second aspect of the present invention provides a charged particle multibeam apparatus configured to perform a drawing process for drawing a desired pattern onto a target. The apparatus includes an illumination system, a pattern definition system, and a projection optical system, wherein the illumination system is configured to generate a charged particle beam and direct the charged particle beam towards the pattern definition system, the pattern definition system is configured to form the charged particle beam into a plurality of subbeams representing the pattern, and the projection optical system is configured to project the plurality of subbeams onto the target. The apparatus further includes a drawing control system, The aforementioned drawing control system is Access the data storage device to read intermediate pattern data including the preprocessed pattern data of the desired pattern. Convert the intermediate pattern data to obtain pattern streaming (real-time transmission) data. Send (stream) the pattern streaming data to the pattern definition system in real time, and instruct the pattern definition system to form the charged particle beam into the plurality of sub-beams during the drawing process according to the pattern streaming data. Characterized in being configured (Form 10).

Embodiments for Carrying out the Invention

[0011] (Form 1) Refer to the first perspective of the present invention above. (Form 2) In the method of Form 1, it is preferable that the preprocessing of the input data in the data preprocessing system includes at least one of the following: Dividing the input pattern data encoded in vector format into a plurality of geometric primitive elements; Applying geometry and / or dose correction to the input pattern data or geometric primitive elements to compensate for the deviation from the ideal drawing state determined prior to the drawing process, provided that the geometry and / or dose correction is performed using correction data supplied as external data based on previous drawing processes and / or inspection processes and / or model-based simulations. (Form 3) In the method of Form 2, the preprocessing of the input data in the data preprocessing system generates preprocessed intermediate pattern data in vector format, and the preprocessed intermediate pattern data is written to the data storage device as intermediate pattern data, and The initialization and / or execution of the drawing process by the at least one drawing control system preferably includes reading the intermediate pattern data from the data storage device and then converting the input pattern data to obtain raster graphics data that represents at least a portion of the desired pattern in raster graphics format. (Form 4) In the method of Form 1 or 2, the intermediate pattern data is raster graphics data for a raster to be used during the drawing process. Preferably, the preprocessing of the input data in the data preprocessing system includes converting the input pattern data in order to obtain raster graphics data that represents at least a portion of the desired pattern in raster graphics format. (Embodiment 5) In the method of embodiment 3 or 4, the steps(s) performed by the at least one drawing control system during the drawing process preferably further include at least one of the following: To compensate for deviations from the ideal rendering state determined during the rendering process, apply geometric and / or dose corrections to the raster graphics data; Dithering pixel data, including transforming the data to match a predetermined grayscale; The pixel data is packaged into a data format that enables it to be transmitted (in real time) at the transmission speed required by the drawing process. (Embodiment 6) In the method of embodiment 4, the preprocessing of the input data in the data preprocessing system preferably includes at least one of the following after conversion to raster graphics data: To compensate for deviations from the ideal rendering state determined during the rendering process, apply geometric and / or dose corrections to the raster graphics data; Dithering pixel data, including transforming the data to match a predetermined grayscale; The pixel data is packaged into a data format that enables the pixel data to be transmitted (in real time) at the transmission speed required by the drawing process. Subsequently, the processed data is written to the data storage device as intermediate pattern data. (Form 7) In any of Forms 1 to 6, the charged particle multibeam apparatus includes two or more multibeam columns arranged in parallel, and The step of reading the intermediate pattern data from the data storage device and the subsequent steps are preferably performed at least partially in parallel with respect to one or more targets in the multi-beam column. (Embodiment 8) In the method of Embodiment 7, the multi-beam columns (or more) arranged in parallel include at least two different illumination systems, and preferably each of the multi-beam columns (or more) includes its own illumination system. (Form 9) In any of the methods of Forms 1 to 8, the step of reading the intermediate pattern data from the data storage device and the subsequent steps are preferably performed by two or more separate drawing processes that are executed simultaneously in multiple different charged particle multibeam devices. (Form 10) Refer to the second perspective of the present invention described above. (Embodiment 11) In the apparatus of embodiment 10, it is preferable that the drawing control system is configured to read and process the intermediate pattern data during the initialization and / or execution of the drawing process performed in the (charged particle multibeam) apparatus. (Embodiment 12) In the apparatus of embodiment 10 or 11, the drawing control system is preferably further configured to perform at least one of the following steps after reading the intermediate pattern data: Converting the pattern data in order to obtain raster graphics data that represents at least a portion of the desired pattern in raster graphics format; To compensate for deviations from the ideal rendering state determined during the rendering process, apply geometric and / or dose corrections to the raster graphics data; Dithering pixel data, including transforming the data to match a predetermined grayscale; The pixel data is packaged into a data format that enables it to be transmitted (in real time) at the transmission speed required by the drawing process. (Embodiment 13) In any of the apparatuses of Embodiments 10 to 12, the charged particle multibeam apparatus preferably includes two or more multibeam columns arranged in parallel, configured to perform at least partially in parallel with respect to one or more targets the step of accessing the data storage device to read intermediate pattern data and the subsequent steps(s). (Embodiment 14) In the apparatus of Embodiment 13, the plurality of multi-beam columns arranged in parallel include at least two different illumination systems, and preferably each of the plurality of multi-beam columns includes its own illumination system.

[0012] A method for performing the above-described type of drawing process in accordance with the present invention includes at least the following steps, preferably performed independently of the drawing process—for example, well prior to the drawing process—in a data preprocessing system: - Receiving input pattern data, - Preprocessing the input pattern data using the drawing parameter data provided to the data preprocessing system in order to obtain intermediate pattern data, and, - Writing intermediate pattern data to a data storage device. Includes; The method further includes at least the following steps, performed during the initialization and / or execution of a drawing process using at least one (charged particle multibeam) device by at least one drawing control system associated with at least one charged particle multibeam device: - Reading intermediate pattern data from data storage device, - To obtain pattern streaming data, intermediate pattern data is transformed, and - Transmitting pattern streaming data in real time to at least one pattern definition system of at least one (charged particle multibeam) device, wherein each pattern definition system forms multiple subbeams from each charged particle beam of at least one (charged particle multibeam) device during the drawing process according to the pattern streaming data. Includes.

[0013] This solution divides the processing of pattern data into two main sections: a first section typically performed in a data preprocessing system before the actual drawing process, and a second section typically performed when the drawing process is running. This enables time-efficient handling of pattern data and keeps the computational complexity during the drawing process low.

[0014] In many embodiments, it may be advantageous for the data preprocessing system to include: dividing input pattern data encoded in vector form into multiple geometric base elements; and / or applying geometric and / or dose corrections to the input pattern data or geometric base elements to compensate for deviations from an ideal drawing state determined prior to the drawing process, wherein the geometric and / or dose corrections are performed using correction data supplied as external data. The external data may be derived from the preceding drawing process and / or inspection process and / or model-based simulations of such processes.

[0015] In many embodiments of the present invention, the intermediate pattern data is in vector format. In such cases, preprocessing of input data in a data processing system generates preprocessed intermediate pattern data in vector format, and the preprocessed intermediate pattern data is written to a data storage device as intermediate pattern data. Furthermore, initialization of a drawing process, or the drawing process itself, performed by one or more drawing control systems, may include reading the intermediate pattern data from the data storage device and then transforming the input pattern data to obtain raster graphics data that represents at least a portion of the desired pattern in raster graphics format.

[0016] In other embodiments, the intermediate pattern data is raster graphics data for a raster to be used during the drawing process. In such cases, preprocessing of the input data in the data preprocessing system typically includes transforming the input pattern data to obtain raster graphics data that represents at least a portion of the desired pattern in raster graphics format. Furthermore, steps performed during the drawing process further include at least one of the following: applying geometric and / or dose corrections to the raster graphics data to compensate for deviations from an ideal drawing state determined during the drawing process; and / or dithering the pixel data so that it is transformed to conform to a predetermined grayscale scale; and / or packaging the pixel data into a data format that allows the pixel data to be transmitted (in real time) at the transmission rate required by the drawing process. The above dose corrections to the raster graphics data may include, for example, global dose correction for areas covering a large number of pixels, and / or local dose correction, such as aperture-specific dose correction, or in other words, pixel graylevel correction. The step of packaging the pixel data may include reordering the data to match the drawing sequence of pixels (in other words, the positioning grid sequence) on the target. This provides a sequence of pixel data corresponding to the sequence of aperture images generated on the target and sorted according to the positioning grid, at the transmission rate required by the drawing process.

[0017] In further embodiments, the preprocessing of input data in a data preprocessing system may include, as further processing steps, after conversion to raster graphics data but before writing the thus processed data to a data storage device as intermediate pattern data: applying geometric and / or dose corrections to the raster graphics data to compensate for deviations from the ideal drawing state determined during the drawing process; and / or dithering the pixel data, which includes converting the data to conform to a predetermined gray value scale; and / or packaging the pixel data into a data format that enables the pixel data to be transmitted (in real time) at the transmission rate required by the drawing process. The step of packaging the pixel data may include rearranging the data as described above.

[0018] The drawing process according to the present invention can be performed simultaneously in multiple drawing devices. Therefore, throughput can be greatly increased by having the step of reading intermediate pattern data from a data storage device and subsequent steps(s) performed by two or more separate drawing processes, preferably simultaneously, in different charged particle multibeam devices(s). Furthermore, the charged particle multibeam device may include two or more multibeam columns arranged in parallel (as shown in Figure 1; see also Figure 7), and the step of reading intermediate pattern data from a data storage device and subsequent steps are performed at least partially in parallel for one or more targets in the multibeam columns(s). Advantageously, the parallel-arranged multibeam columns(s) may include at least two different illumination systems, each illumination system including its own particle source; therefore, each illumination system will supply beam particles to one or more columns different from the one or more columns associated with the other illumination systems(s). Preferably, each of the multibeam columns(s) includes its own illumination system and, in particular, its own particle source.

[0019] The method according to the present invention is preferably performed in a charged particle multibeam apparatus configured to perform such a drawing process and equipped with a drawing control system. The drawing control system is configured to access a data storage device to read intermediate pattern data containing preprocessed pattern data of a desired pattern, to convert the intermediate pattern data to obtain pattern streaming data, to transmit pattern streaming data to a pattern definition system (in real time), and to instruct the pattern definition system to form the charged particle beam into the plurality of subbeams (from the charged particle beam) during the drawing process in accordance with the pattern streaming data. The drawing control system may further be configured to read and process the intermediate pattern data during the initialization and / or execution of the drawing process performed in the charged particle multibeam apparatus(s). Additionally, the drawing control system may be configured to perform one, some, or all of the following steps after reading the intermediate pattern data: converting the pattern data to obtain raster graphics data representing at least a portion of the desired pattern in raster graphics format; and / or applying geometric and / or dose corrections to the raster graphics data to compensate for deviations from the ideal drawing state determined during the drawing process; and / or dithering the pixel data to conform to a predetermined grayscale; and / or packaging the pixel data into a data format that enables the pixel data to be transmitted (in real time) at the transmission rate required by the drawing process (where packaging the pixel data may include the data reordering as described above).

[0020] To further illustrate the present invention, exemplary and non-limiting embodiments, such as those schematically shown in the drawings, are discussed below. [Brief explanation of the drawing]

[0021] [Figure 1] A schematic cross-sectional view of an example of a multi-column plotting tool according to the present invention. [Figure 2] A vertical cross-sectional view of an example of a pattern definition system based on conventional technology. [Figure 3A] A partial plan view (rectangular arrangement) of the first arrangement of columns (multiple) relative to the target. [Figure 3B] Second arrangement of columns (diamond arrangement). [Figure 4A] Basic drawing strategies for targets using multiple stripes. [Figure 4B] An illustrative arrangement of multiple apertures imaged onto a target. [Figure 5A] An example of a pixel map of an exemplary pattern to be exposed. [Figure 5B] An example of a pixel map of an exemplary pattern to be exposed. [Figure 6A] A configuration of multiple apertures with M=2 and N=2. [Figure 6B] An example of oversampling of pixels (multiple) in a "double grid" configuration. [Figure 7] An overview of an example of equipment related to a pattern data processing flow that is performed partly upstream (as a pre-process) and partly during the drawing process in one or more drawing tools. [Figure 8] An example flowchart of a pattern data processing pipeline according to the first embodiment of the pattern data processing flow. [Figure 9] An example flowchart of a pattern data processing pipeline according to another embodiment of the pattern data processing flow. [Figure 10] An example flowchart of a pattern data processing pipeline according to yet another embodiment of the pattern data processing flow. [Figure 11] An example of a two-tier streaming architecture for data streaming processes to drawing tools. [Examples]

[0022] The detailed discussion of the exemplary embodiments of the Invention given below discloses the basic idea and further advantageous developments of the Invention. It will be apparent to those skilled in the art that some or all of the embodiments discussed herein may be arbitrarily combined in any way that is deemed suitable for a particular application of the Invention. Throughout this disclosure, terms such as “advantageous,” “exemplary,” or “preferred” describe elements or dimensions that are particularly suitable (but not essential) for the Invention or one embodiment thereof, and can be modified unless expressly required, if deemed suitable by those skilled in the art. It goes without saying that the Invention is not limited to the exemplary embodiments discussed below, which are given for illustrative purposes and merely present preferred examples of the Invention. Within the scope of this disclosure, terms relating to the vertical direction, such as “up” or “down,” should be understood as referring to the direction of a beam that is considered to propagate downward along the vertical axis. This vertical axis is further considered to be identical to the Z direction, which is intersected by the X and Y directions.

[0023] The embodiments discussed below disclose approaches for efficient handling (processing) of pattern image data to be drawn on a target in a drawing tool.

[0024] Lithography device

[0025] Figure 1 shows a schematic cross-sectional view of an example of a multi-column lithography tool 1 equipped with a vacuum housing 10 and a target chamber 3 for a multi-column charged particle optical system 2. The multi-column charged particle optical system is mounted in the target chamber 3 by a column base plate 4. Inside the target chamber 3 is an XY stage 5, for example, a laser interferometer-controlled air-bearing vacuum stage, on which a substrate chuck 6 is loaded using a suitable operating system. The chuck 6, which is preferably an electrostatic chuck, holds a substrate 7, such as a silicon wafer, which serves as the target. For charged particle multi-beam lithography, the substrate is coated with, for example, an electron or ion beam-sensitive resist layer [7a].

[0026] The multi-column optical system 2 includes multiple sub-columns 9 (the number of columns shown in the illustration is reduced for clarity, but represents a much larger number of columns present in an actual multi-column device). Preferably, the sub-columns 9 have the same configuration or setup and are arranged side by side so that their axes are parallel to each other. Each subcolumn has an illumination system 11 including an electron or ion source 11a, an extraction system 11b, and an electrostatic multi-electrode condenser optical system 11c, and a charged particle reduction projection optical system 16, wherein the illumination system 11 (or condenser optical system 11c) supplies a broad telecentric charged particle beam to a pattern definition (PD) system 12 adapted to allow only a plurality of apertures that define the shape of the subbeam ("beamlet") passing through the aperture (beam shaping device) to pass into the beam, and the charged particle reduction projection optical system 16 consists of a plurality of electro-magneto-optical continuous projection stages, preferably including electrostatic and / or magnetic lenses and optionally other particle optics devices. In the embodiment shown in Figure 1, the projection optical system 16 includes, for example, three lenses: the first lens is an accelerating electrostatic multi-electrode lens 16a, while the second and third lenses 16b and 16c are magnetic lenses, particularly when electrons are used, or electrostatic lenses, for example, when the particles are ions.

[0027] The accelerating first lens of the projected charged particle optical system 16 operates the PD system 12 with low particle kinetic energy (e.g., 5 keV) while providing high beam energy (e.g., 50 keV) to the crossover(s) of the reduced projection optical system, thus offering the important advantage of minimizing stochastic Coulomb interaction. Furthermore, the high beam energy in the substrate is beneficial for pre-reducing the scattering of charged particles when exposing the target, particularly the charged particle-sensitive layer 8[7a].

[0028] The first lens of the projection optical system forms a first crossover, while the second lens forms a second crossover. At this position in each subcolumn is a shielding plate 15 configured to remove the beam deflected in the PD system. The third lens 16c and shielding plate 15 of the subcolumn are mounted to a reference plate 17, which is attached to the column base plate 4 by appropriate fastening means 18. The reference plate 17 is provided with a portion 19 of the off-axis optical alignment system.

[0029] The reference plate is manufactured from a suitable substrate having low thermal expansion, such as a silicon oxide or aluminum oxide-based ceramic material, which is lightweight, has a high modulus of elasticity and high thermal conductivity, and can be coated with a conductive film (coating) in at least the relevant (appropriate) portions to avoid static charge buildup.

[0030] As can be seen in the longitudinal section detail in Figure 2, the prior art PD system 12 includes three plates arranged vertically in a continuous configuration: an aperture array plate 20 (AAP), a deflection array plate 30 (DAP, also called a blanking array plate), and a field-boundary array plate 40 (FAP). Here, the term “plate” relates to the overall shape of the associated device, but does not necessarily indicate that the plate is realized as a single plate component, which is usually a preferred way of embodiment; rather, it is important to note that in certain embodiments, a “plate,” such as an AAP, may consist of multiple subplates. These plates are preferably arranged parallel to each other with common (equal) spacing along the Z direction.

[0031] The flat top surface of the AAP20 forms a defined potential interface with respect to the capacitor optical system 11c. The AAP may consist of, for example, a rectangular or square piece of silicon wafer 21 (approximately 1 mm thick) having a thinned central portion 22. This plate may be covered with a conductive protective layer 23, which would be particularly advantageous when using hydrogen or helium ions (according to US6,858,118). When using electrons or heavy ions (e.g., argon or xenon), layer 23 may also be silicon provided by the surface portions 21 and 22, respectively, in which case there is no interface between layer 23 and the bulk portions 21 / 22, respectively.

[0032] The AAP20 has multiple apertures 24 formed by holes that traverse (penetrate) the thinned portion 22 (in the vertical direction). In the illustrated embodiment, the apertures 24 are embodied to have a linear (vertical) profile (contour) portion formed in the layer 23 and a "retrograde" profile (contour) portion in the bulk layer of the AAP20, so that the lower exit (outlet) 25 of the hole is wider than the main part of the aperture 24. Both the linear profile portion and the retrograde profile portion can be formed by conventional structuring techniques such as reactive ion etching. The retrograde profile portion strongly reduces the mirror-charging effect of the beam passing through the hole.

[0033] DAP30 is a plate having a plurality of holes 33, each corresponding to the positions of apertures 24 of AAP20. These holes 33 are equipped with electrodes 35, 38 configured to selectively divert individual subbeams passing through the holes 33 from their respective paths. DAP30 can be manufactured, for example, by post-processing of a CMOS wafer equipped with an ASIC circuit. DAP30 is made, for example, from a single CMOS wafer having a square or rectangular shape and includes a thickened portion 31 that forms a frame holding a thinned (but preferably thicker than the thickness of 22) central portion 32. The aperture holes 33 in the central portion 32 are wider than 24 (for example, by about 2 μm on each side in Figure 2). CMOS electronic circuits 34 are used to control the electrodes 35, 38 provided by MEMS technology. A "ground" electrode 35 and a deflection electrode 38 are provided near each hole 33. The earth electrodes (multiple) 35 are electrically interconnected, connected to a common earth potential, and include a downwardly widened (larger diameter) portion 36 to prevent charging and an insulating portion 37 to prevent unwanted shortcuts to the CMOS circuit. The earth electrodes 35 may also be connected to portions of the CMOS circuit 34 that are at the same potential as the silicon bulk portions 31 and 32.

[0034] The deflection electrode 38 is configured to have an electrostatic potential selectively applied to it; when such an electrostatic potential is applied to the electrode 38, it causes deflection with respect to the corresponding subbeam, generating an electric field that diverts the subbeam from its nominal path. The electrode 38 may also have a downwardly widening (increased diameter) portion 39 to avoid charging. Each of the electrodes 38 is connected at its base to the respective contact portion inside the CMOS circuit 34.

[0035] The height (length) of the ground electrode 35 is greater (longer) than the height (length) of the deflection electrode 38 in order to suppress crosstalk effects between beams.

[0036] The deflection electrode 38 is selectively activated (powered) and controlled by data and control signals supplied to the DAP 30 in real time during the drawing process. These data and control signals are collectively referred to as streaming pattern data. Referring to Figure 1, the streaming pattern data is supplied to the PD system 12 by the control system 14 of the drawing machine tool 1 via a bundle of data lines (which are collectively represented as data path lines 13 in Figure 1).

[0037] The third plate 40, which functions as a FAP, is oriented downward toward the first lens portion of the charged particle reduction projection optical system 16 and has a flat surface that provides a defined potential interface with respect to the first lens 16a of the projection optical system. The thick portion 41 of the FAP 40 is made of a part of the silicon wafer and is a square or rectangular frame having a thinned central portion 42. The FAP 40 has multiple holes 43 that correspond to the holes 24 and 33 of the AAP 20 and DAP 30, but are wider than those of the AAP 20 and DAP 30.

[0038] The PD system 12, in particular its first part, the AAP 20, is irradiated with a broad charged particle beam 50 (where "broad" means the beam is wide enough to cover the entire area of ​​the aperture array formed in the AAP 20), and thus, as the beam 50 is sent out through the apertures 24, it is split into thousands of micrometer-sized beams 51. The beams 51 will traverse (pass through) the DAP and FAP without obstruction.

[0039] As previously described, when the deflection electrode 38 is powered by a CMOS electronic circuit, an electric field is generated between the deflection electrode and the corresponding ground electrode, which causes a small but sufficient deflection of the beam 52 as it passes through (Figure 2). The deflected beam can pass through the DAP and FAP without obstruction because the respective holes 33 and 34 are formed with a sufficiently wide width. However, the deflected beam 52 is removed by the sub-column shielding plate 15 (Figure 1). Therefore, only the beam unaffected by the DAP reaches the substrate.

[0040] The reduction ratio of the charged particle reduction optical system 16 is appropriately selected in terms of the beam dimensions and their inter-beam distances in the PD system 12 and the desired dimensions of the substrate in the target. This allows for micrometer-sized beams in the PD system while projecting nanometer-sized beams onto the substrate.

[0041] The beam 51 formed by the AAP (unaffected) is projected onto the substrate at a preset reduction ratio R of the charged particle projection optics. Thus, a "beam array field (BAF)" with widths BX = AX / R and BY = AY / R, where AX and AY represent the sizes of the aperture array field along the X and Y directions, respectively, is projected onto the substrate. The beam sizes of individual subbeams on the substrate are given by bX = aX / R and bY = aY / R, where aX and aY represent the sizes of the beam 51 measured along the X and Y directions at the level (height) of the DAP 30, respectively.

[0042] It is important to note that the individual beams 51, 52 shown in Figure 2 (and similar subsequent drawings) represent a much larger number of sub-beams, typically thousands, that are arranged in a two-dimensional XY array. The applicant has already realized multi-beam charged particle optics with a reduction ratio of R=200 for ion and electron multi-beam columns with thousands (e.g., 262,144) programmable beams. The applicant has also already realized such columns with a beam array field of approximately 82 μm × 82 μm on the substrate. These examples are mentioned for illustrative purposes only and should not be understood as limiting examples.

[0043] The schematic arrangement shown in Figure 3 is used to realize subcolumns with diameters such that multiple subcolumns of the above types fit within the area of ​​a substrate, such as a 300 mm silicon wafer, which is used as a substrate for the manufacture of state-of-the-art integrated circuit devices. Development of 193 nm immersion photolithography, EUV, and nanoimprint lithography tools for 450 mm silicon wafer sizes is currently underway. The multi-column configurations presented in this document can be easily adapted to any other wafer size, such as a 450 mm diameter silicon wafer, without loss of throughput by correspondingly providing a larger number of subcolumns.

[0044] The various embodiments described below relate to a multibeam tool in which a substrate (e.g., a silicon wafer) is exposed by a 193 nm immersion scanner tool having die fields, typically having DX = 33 mm and DY = 26 mm, with length DX and width DY. This type of multibeam tool is described in the applicant's US 2015 / 0311031 A1 and US 9,443,699, the teachings of which are incorporated herein by reference. A single die field may contain several chips, but usually contains several chips. Many embodiments are not limiting, and therefore the present invention may also relate to other layouts and applications.

[0045] Exemplary embodiments of compact sub-column arrangements 60 and 70 suitable for the present invention are shown in Figures 3A and 3B, respectively, which provide plan view details of the arrangement(s) with respect to the target plane. Figure 3A shows an example of a “rectangular” layout, where one sub-column 61 (symbolically represented by a circle) having an aperture array field 62 is used to expose a region 63 of one die field (each indicated by different types of hatching); thus the relative arrangement of the columns(s) reflects the relative arrangement of the die fields(s). Figure 3B shows an example of a “rhomboid” arrangement of columns, where one sub-column 71 having an aperture array field 72 is used to expose regions 73 of two die fields, but the distance between two adjacent columns corresponds to only the diagonal length of one die field.

[0046] If DX = 33 mm and DY = 26 mm, then in Figure 3A, the diameter of sub-column 61 is approximately 24 mm and the size of aperture array field 62 is approximately 8.2 mm × 8.2 mm, whereas in Figure 3B, the diameter of sub-column 71 is approximately 40 mm and the size of aperture array field 72 is approximately 16.4 mm × 16.4 mm. If the aperture pitch is 16 μm, aperture array field 62 can provide 512 × 512 = 262,144 beams, while aperture array field 72 can provide 1024 × 1024 = 1,048,576 beams.

[0047] It is clear that the multi-column layout described above requires numerous data and control signals to be supplied as input signals to the DAP(s) in order to properly control the numerous PD systems of the multi-column apparatus. Furthermore, additional control lines may exist to read from the deflection devices(s) to provide output signals. These input and output lines of the DAP(s) are collectively referred to as the data path.

[0048] In particular, the data path includes, for example, optical fibers and / or electrical line components (e.g., flatband cables) as described in the applicant's US2015 / 0311031 A1 and US 9,443,699.

[0049] Drawing patterns

[0050] Referring to Figure 4A, a pattern image pm defined by the PD system 4 is generated on the target 7. The target surface, covered with a charged particle-sensitive resist layer 8[7a], includes one or more areas r1 to be exposed. Generally, the pattern image pm exposed on the target typically has a finite size y0 that is sufficiently smaller than the width of the area r1 on which the pattern is to be formed. Thus, a scanning stripe exposure strategy is used, in which case the target is moved under the incident beam so that the beam's position on the target is constantly changing: i.e., the beam is efficiently scanned over the entire target surface. It should be emphasized here that, for the purposes of the present invention, only the relative motion of the pattern image pm on the target is important. This relative motion moves the pattern image pm over the entire area r1, forming a series of stripes s1, s2, s3, ..., sn (exposure stripes) of width y0. A complete set of stripes covers the entire area of ​​the substrate surface. The scanning direction sd may be constant (the direction of motion of the pattern image is the same between adjacent stripes), or the direction of motion from one stripe to the next may be reversed (alternating) (the direction of motion of the pattern image may be opposite between adjacent stripes).

[0051] Figure 5A shows a simple example of an imaging pattern ps with a size of 10 × 18 = 180 pixels. Here, some pixels p100 in the exposure area are exposed to a 100% gray level 401, while some other pixels p50 are exposed to only 50% of the full gray level 402. The remaining pixels are exposed to a 0% dose 403 (i.e., not exposed at all). Figure 5B shows how the 50% level is achieved: each pixel is exposed multiple times at a gray level of 0-100% per pixel, but the gray level is achieved by selecting the corresponding number of exposures for the pixels to be activated; the gray level is the ratio of activated exposures to the total number of exposures. In this example, the 50% level is achieved by selecting 2 out of 4 exposures. Of course, in the practical application of the present invention, the number of pixels in a standard image would be much larger. However, in Figures 5A and 5B, the number of pixels is limited to 180 for readability. Furthermore, generally, more gray levels are used between the 0% and 100% scale.

[0052] Thus, the pattern image pm (Figure 4A) consists of multiple pattern pixels px that are exposed with dose values ​​corresponding to the desired pattern to be exposed. However, it should be noted that since there are only a finite number of apertures in the aperture field of the PD system, only one subset of pixels px can be exposed at a time. The patterns of the apertures that are switched on are selected according to the pattern to be exposed on the substrate. Therefore, in an actual pattern, not all pixels are exposed with the full dose; some pixels are "switched off" according to the actual pattern; for each pixel (in other words, for each beamlet covering a pixel), the exposure dose may differ from pixel to pixel exposure cycle, depending on whether the pixel is "switched on" or "switched off" according to the pattern to be exposed or structured on the target.

[0053] While the substrate 16 is moving continuously, the same image element corresponding to a pattern pixel px on the target can be covered multiple times by a series of aperture images. Simultaneously, the pattern in the PD system is shifted stepwise through the apertures of the PD system. Therefore, considering a single pixel at a certain position on the target, if these apertures cover that pixel, and all apertures are switched on, the maximum exposure dose level, i.e., a "white" shade equivalent to 100%, can be obtained. In addition to the "white" shade, it is also possible to expose a single pixel on the target to a lower dose level (also called a "gray shade") that interpolates between the minimum ("black") and maximum ("white") exposure dose levels. A gray shade can be achieved, for example, by switching on only a subset of the apertures involved in drawing a single pixel; for example, four of the 16 apertures would give a 25% gray level. Another approach is to reduce the exposure time (duration) that is not blanked for the aperture(s) involved. Thus, the exposure time for a single aperture image is controlled by a grayscale code, e.g., an integer value. The exposed aperture image is one manifestation of a given number of gray shades, corresponding to zero and the maximum exposure time and (maximum) dose level. The grayscale is typically a set of gray values, e.g., 0, 1 / (n y ―1)...,i / (n y ―1), ..., 1, however n y is the number of gray values, i is an integer ("gray index", 0 ≤ i ≤ n) y ) is determined. However, generally speaking, the gray values ​​(multiple) do not need to be equally spaced, and form a non-decreasing sequence between 0 and 1.

[0054] Figure 4B shows a preferred arrangement of apertures in the aperture field of a PD system according to a basic layout, and further illustrates some quantities and abbreviations used below. The diagram shows the arrangement of aperture images b1 projected onto a target, which are shown in black shades. The principal axes X and Y correspond to the direction of target motion (scan direction sd) and its perpendicular (orthogonal) direction, respectively. Each aperture image has widths bX and bY along directions X and Y, respectively. The apertures are arranged along rows and columns, each having MX and MY apertures, respectively, with offsets between adjacent apertures in each row and column being NX·bX and NY·bY, respectively. As a result, each aperture image has a conceptual cell C1 with an area of ​​NX·bX·NY·bY, and the aperture arrangement includes MX·MY cells arranged in a rectangular pattern. Hereafter, these cells C1 will be referred to as “exposure cells”. The complete aperture arrangement, when projected onto the target, has dimensions BX (=MX·NX·bX) × BY (=MY·NY·bY). In the following discussion, we assume a square grid as a special case of a rectangular grid, where b=bX=bY, M=MX=MY, and N=NX=NY, and M is an integer. These are for the sake of explanation below and do not impose any restrictions on generalization. Thus, (one) “exposure cell” has dimensions N·b × N·b on the target substrate.

[0055] The pitch between two adjacent exposure positions is denoted by e below. Generally, the distance e can differ from the nominal width b of the aperture image. In the simplest case, b=e, as shown in Figure 6A for an example of a 2x2 exposure cell arrangement C3, where one aperture image bi0 covers one pixel (or its nominal position). In one other important case, shown in Figure 6B (and also following the teachings of US 8,222,621 and US 7,276,714), e can be a fraction of the aperture image width b, b / o, where o>1 and preferably an integer (but not necessarily an integer), and o is also called the oversampling factor. In this case, the aperture images, each on a different exposure path, spatially overlap, allowing for higher-resolution placement of the pattern to be generated. As a result, each image of an aperture can, at once, cover a large number of pixels, specifically o 2 This will cover a number of pixels. The entire area of ​​the aperture field that is imaged onto the target is (NMo) 2 This will result in the inclusion of 100 pixels. From the perspective of aperture image positioning, this oversampling corresponds to a so-called positioning grid that is different (because it is finer than the spacing) than what is needed to simply cover the target area.

[0056] Figure 6B shows an example of oversampling with o = 2 combined with a positioning grid called a "double grid". Specifically, it shows an example of an image of an aperture array by (one) exposure cell C4 having parameters o = 2 and N = 2. Therefore, four aperture images bi1 (dashed lines) are printed at each nominal position (the small square fields in Figure 6B), but these are offset (shifted) by a pitch e in both the X and Y directions on a regular grid. The size of the aperture images is the same value b, while the pitch e of the positioning grid is now b / o = b / 2. The offset with respect to the previous nominal position (the offset of the positioning grid) is also of size b / 2. At the same time, the dose and / or grayscale of each pixel can be adapted (reduced) by selecting an appropriate gray value for the aperture image covering each pixel. As a result, an area of size a is printed, but the positioning grid is finer, so the positioning accuracy is increased. By directly comparing Figure 6A and Figure 6B, the position(s) of the aperture image(s) is / are arranged on the positioning grid twice as fine (generally o times finer) than before, while the aperture image(s) themselves overlap. The exposure cell C4 now includes the position (i.e., "pixel") to be processed during the drawing process, and thus contains o times more pixels than before. Accordingly, an area bi1 having an aperture image of size b×b is associated with o = 4 pixels in the case of oversampling with o = 2 in Figure 6B (also called a "double grid"). Of course, o can take any other integer value, especially 4 ("quad grid", not shown) or 8. For the parameter o, non-integer values greater than 1, such as √2 = 1.414 or 2√2 = 2.828, corresponding to the case of the "Double-Centered Grid" shown in US 9,653,263, may be assigned. 2 of the position (i.e., "pixel") and thus o 2 times more pixels than before. 2 = 4 pixels in the case of oversampling with o = 2 in Figure 6B (also called a "double grid"). 1 / 2 = 1.414 or 2 3 / 2 = 2.828, may be assigned.

[0057] It is important to note that when using an interlocking grid (o>1), the number of gray shades can be increased through dithering, while the dose distribution remains uniform. This is because the number of gray shades for any nominal grid is equal. This means that for a double-interlocking grid, the number of achievable effective dose levels is more than four times greater than for an uninterlocking grid. Generally speaking, all oversampling exposure grids (i.e., o>1) are shifted by a distance b / o in the X and Y directions. 2 It consists of up to o nominal grids. Therefore, if the dose level of only one of these o grids is increased, the step from one dose level to the next can be divided into o substeps; this can be repeated for the other grids until all subgrids are exposed to the nominal level. As those skilled in the art will know, the beam shape on the substrate is a convolution of the mechanical blur (of the aperture plate) and the reduced aperture shape of the aperture plate. By setting the width b to a natural number multiple of the exposure grid constant e, in other words, by making o = b / e an integer, a uniform dose distribution can be obtained on the substrate. Otherwise, the dose distribution may have minimum and maximum values ​​in the period of the exposure grid due to aliasing effects. A large number of gray shades allows for better feature positioning. Therefore, increasing the gray level is important when the number of gray shades per pixel position is limited to a certain number.

[0058] Pattern data processing

[0059] In accordance with the present invention, the lithography tool includes a data processing system that converts the pattern to be drawn into a representation of dose assignments defined on a raster, such as gray level assignments or beamlet dose assignments to a pixel array on a target (each for a beam array field of individual columns formed by the respective PD system). Typically, the pattern is repeatedly drawn on the same substrate, for example, on multiple dies (chips) on a (single) silicon wafer in the case of direct lithography, and / or on a series of mask blanks in the case of mask lithography. The lithography tool processing system consists of the following two main parts (1) and (2): (1) The first part is configured to generate intermediate data by preprocessing pattern data independently of, and preferably prior to, the drawing process, and to store the intermediate data in a memory unit capable of maintaining the intermediate data for a longer period of time. This first part can be implemented as a dedicated, independent system, also referred to as a "data preprocessing unit" or "data center," shown as unit 101 in Figure 7. Those skilled in the art will see that this first part can also be implemented in other ways, such as in the form of a distributed computer network; And, (2) The second part is configured to continue the data processing flow during the drawing process, reading intermediate data from the storage unit, applying several further processing steps to the intermediate data as needed, converting the intermediate data into streaming data, and streaming the streaming data to the individual column PD system in real time during the drawing process for each substrate exposure. This second part 102 is also referred to as the "data streaming unit" or "data path". This is typically embodied by multiple individual instances in each of the control systems of the drawing tool, which are shown as multiple units associated with the corresponding drawing system or columns of the multi-column drawing tool, for example, in the example of Figure 7. This second part 102 can also be implemented using other configurations, for example, a single server computer or a computer array (each computer in the computer array supporting the corresponding group of columns (or just one single column) of the multi-column drawing tool).

[0060] Figure 7 is a simplified schematic conceptual diagram of an example of a pattern data processing flow in an example of a drawing system comprising multiple drawing tools 1A, 1B and a preprocessing system 101 for generating pattern data to be drawn on each of the targets 7, 7' processed by the drawing tools. The patterns to be drawn, input pattern data INPDAT, are supplied to the data preprocessing station 111 of the preprocessing system 101. The data preprocessing station 111 processes the input pattern data to generate intermediate data IMDAT and writes them to one or more storage devices 11M, such as a hard drive type mass storage device accessible via LAN or a dedicated storage system specifically provided for the drawing system; in the (one or more) storage devices 11M, the intermediate data is stored in an appropriate manner so that it can be accessed by the data streaming unit 162. In the example shown in Figure 7, multiple data streaming units 162 access the (one or more) storage devices 11M to read the intermediate data, which can preferably be performed simultaneously, using, depending on the case, local intermediate storage units LMDAT (e.g., SSD memory) and / or special additional data processing. Each data streaming unit 162 then streams the data (in real time) to the drawing tool columns that use the data during the drawing process, via a buffer storage unit SBUF (e.g., the control system's RAM memory). In the schematic diagram of Figure 7, two drawing tools 1A and 1B, each containing multiple columns (for clarity, only three columns c1, c2, c3 and c4, c5, c6 for each drawing tool are shown), are supplied with pattern data by a corresponding number of data streaming units (six in Figure 7), all of which ultimately access the intermediate data IMDAT from the data preprocessing unit 101. Generally, the number of drawing tools 1A and 1B can be one, two, or any other number. For each drawing tool 1A and 1B, the data streaming unit 162 belonging to each constitutes a part of the respective controller 14A and 14B (which correspond to the control system 14 in Figure 1).

[0061] Pattern data

[0062] In many embodiments, the pattern to be drawn is described in vector form as a collection of geometric elements, such as rectangles, trapezoids, or general polygons; this typically provides better data compression and thus reduces the data storage requirement. Polygon vertex density can serve as a measure of the complexity of the pattern, with a typical example of a polygon vertex value being 2000 vertices / μm 2 Furthermore, typically, patterns are repeatedly drawn on multiple substrates over a long period of time, but the initial pattern data is first pre-processed into intermediate data and then stored for streaming over that long period. In a typical example of direct drawing, the pattern covers an area of ​​a die with side lengths of 31 mm × 26.8 mm, and considering the vertex density of 2 × 4 bytes / vertex, it would require approximately 13.3 TB of memory storage space.

[0063] Figures 8 to 11 schematically illustrate an example of a pattern data processing flow. The rectangular shapes represent nodes corresponding to specific data structure or data format stages, and the edges (or arrows) represent processing and / or streaming operations. In particular, the first node represents pattern VECPAT, which is supplied to the data center 101 in vector form as input pattern data INPDAT (see Figure 7). Furthermore, in Figures 8 to 10, the left side of each data stage shows a symbolic image (illustration) of each data stage, illustrating an example of a simple geometric structure. The dashed boxes in Figures 8 to 11 symbolically represent sections / units in which each part of the processing flow is executed. Other than that, the processor and control system executing the pattern data processing flow are not shown in these figures for clarity, but they are understandable and easily embodied by those skilled in the art.

[0064] Data processing

[0065] In one group of embodiments of the present invention, the data center 101 preprocesses the pattern to generate (lower) beamlet dose assignments for individual columns, which are used as intermediate data. During the drawing process, the (drawing) tool scans the substrate, and each data path (for each single column and the PD system to which it belongs) transmits the intermediate data from the data center to its PD system (in real time).

[0066] Figure 8 shows an example flowchart of the pattern data processing pipeline in the data center 181 and data path 182 of the first embodiment of the pattern data processing flow.

[0067] In the first step 91, the pattern VECPAT is divided into multiple smaller data chunks SPLITGEOM, which may geometrically overlap. Each chunk contains, for example, one or more geometric elements. In typical embodiments, each chunk may consist of multiple polygons, each polygon having fewer vertices than a predetermined maximum number and fewer edges than a predetermined maximum edge length. For example, the maximum number of vertices may be on the order of 10,000, such as 7,560 or 8,192, and the maximum edge length may be on the order of 1 μm. This division process in step 91 does not take into account any properties of individual columns, and therefore the data does not need to be copied for individual processing in this stage (step).

[0068] The data preprocessing unit 181 may include a compensation subunit configured to compensate for non-ideal drawing conditions such as substrate distortion due to processing, beam array field distortion that cannot be corrected by optical deflection means, "loading effects" resulting from preceding, parallel, or subsequent etching processes, charge induction effects, and / or other adverse effects. For example, as step 92, it may be convenient to apply compensation in the vector domain that affects all columns in the same way in this stage (step). The corresponding processing may be performed independently, and possibly in parallel, for all chunk SPLITGEOMs, but the resulting data is classified and coded in a manner that improves the computer calculation speed of subsequent steps. Most commonly, this would involve a change in nominal dose assigned to these geometries so that effects such as fogging due to backscattered electrons, masking effects (such as etch loading effects), and proximity effects due to forward-scattered electrons can be compensated for. This is performed based on appropriate dose correction data DOSEMOD, which can be supplied as an external input, obtained, for example, from model-based and / or measurement-based analysis of the relationship between dose and the limit dimensions of the feature(s) drawn on the substrate. Appropriate methods for dose correction are well known from the prior art. Furthermore, it is possible to include correction of lateral spatial position changes with respect to the particle beam by applying transformations to the geometry, such as translation, scaling, or rotation, as described in US9,568,907B2. This can be performed based on position change correction data DISLOCMOD, which can be supplied as an external input, determined, for example, in a similar manner from analysis of the relationship between dose and feature position changes (e.g., due to charging or heating of the substrate or part of the exposure apparatus).The results of such analysis are supplied to the data center's correction subunit in an appropriate format, for example, as model parameters or processed measurement results, and used in the correction process of step 92. Step 92 provides an output, referred to as CORRGEOM in Figure 8, which is a collection of chunks, each chunk containing a collection of geometric elements.

[0069] In the next rasterization step 93, the chunks, or more precisely, the geometric elements of all chunks, are converted into a raster graphics array PIX. Here, the pixel gray level represents the physical dose of the corresponding aperture image. All pixels entirely within a single geometry are assigned the polygon's color, while the colors of pixels crossing one side of a single geometry are weighted by a fraction of the area of ​​pixels covered by that geometry. This method suggests a linear relationship between the area of ​​the geometry and the total dose after rasterization. The doses are first calculated as floating-point numbers; only then are they converted into a discrete set of dose values, as supported by the PD system. The raster graphics array PIX may also include multiple subgrid arrays (which are superimposed by being drawn in subsequent drawing scans); the subgrid arrays may correspond, for example, to positioning grids. As a result of rasterization, the pixel data of a raster graphics array PIX (or each subgrid array) represents the nominal dose value for each pixel, for example, in the form of a floating-point number. Typically, representing input pattern data in the pixel domain increases the memory requirements by an order of magnitude. Graphics processing units (GPUs) are particularly well-suited for processing this data, while data communication overhead is practically negligible, and computational operations on the data can be performed in parallel.

[0070] In one variant or complementary form of this embodiment, correction of adverse effects on drawing conditions that affect all columns in the same way can be applied to the array PIX data already in the pixel domain, rather than (or supplementing) the operation on the elements of SPLITGEOM. The same applies to the correction processes DOSEMOD and DISLOCMOD, as already considered for the input data; however, the output PIX is a collection of raster graphics arrays or subgrid arrays.

[0071] In the next step 94, the pixel data of the raster graphics array PIX may be corrected for column-specific adverse effects on the drawing state (condition). This correction step generates corrected pixel data CORRPIX as individually corrected copies of the input pixel data. For example, this correction step may include distortion of the beam array field that cannot be corrected by optical beam deflection means, spatial changes in beam blur within the beam array field, and / or spatial changes in the current density of the charged particle beam, as described in the applicant's US9,568,907B2, US9,520,268B2, and US10,325,757B2, respectively. Note that these documents describe multiple correction techniques, and the teachings from these documents are incorporated herein by reference.

[0072] An example of correction in the pixel domain is described in US9,520,268B2: a predetermined set of column-specific blur homogenization kernels, COLMOD, can be applied to the pixel data of a beam array field. This type of correction is performed by convolution of the pixel data by kernels. The convolution kernels typically act on the blur range, which is tens of nanometers, so that the pixel data of the beam array can be parallelized in an effective way, distributed across different CPUs and / or GPUs.

[0073] In one variant or complementary form of this embodiment, correction of column-specific adverse effects on drawing conditions can already be applied in the vector domain. In particular, spatial position changes are suitable for such correction, as described in the applicant's US9,568,907B2.

[0074] Next, the dithering process in step 95 converts the dose value data of the pixel data CORRPIX to gray value data DITHPIX based on a predetermined gray value scale. This is a position-dependent rounding process that ensures that rounding errors are averaged across nearby pixels, and, combined with oversampling, allows for much finer dose variation than would be possible with a discrete set of dose values ​​available for just one aperture; this can be achieved by known algorithms for converting visual image data to pixel graphics. Additional corrections can also be applied in this stage (step) immediately before or after dithering, depending on the actual corrections (e.g., bad aperture correction as described in US9,269,543B2), if they are applicable in the pixel domain. The resulting data DITHPIX is, for example, integer data represented on a raster array.

[0075] In step 96, pixel packaging may be performed. The data DITHPIX obtained from stage (step) 95 is distributed into a sequence of pixel data corresponding to a sequence of aperture images (multiple) generated on the target and sorted according to the positioning grid sequence. This helps to reorder the pixel data into a suitable form acceptable to the PD system and is compatible with the drawing method, and the resulting pixel package data PACKPIX is stored in a dedicated storage device (storage device 11M in Figure 7) capable of supporting the access speed and transmission rate required by the drawing process. For example, by the drawing method already described with reference to Figures 4 to 6, two pixels adjacent to each other within the same exposure cell are exposed to a large number of other pixels between them, for example, No 2 Pixels can be exposed by exposure up to -2. Thus, in this embodiment, the pixel package data forms intermediate data held in memory that can be accessed multiple times for repeated exposure of the same pattern.

[0076] Given the computational load of the data preprocessing unit for each column, it would be advantageous to store intermediate data for as long as possible, but at most until the required number of wafers or masks have been exposed. However, adverse effects that result in non-ideal drawing conditions change over time, and therefore the corrections applied throughout data preprocessing need to be readjusted. Given this possibility, it may be advantageous to monitor the characteristics (values) of the (drawing) tool and each column by in-situ and ex-situ processes that can provide feedback when the drawing conditions have changed sufficiently to necessitate readjusting the corrections. The in-situ process may involve, for example, the measurement of charged particle currents using a Faraday cup. On the other hand, the ex-situ process typically involves the inspection (observation) of the reference pattern drawn on the target substrate using an independent measuring device, such as a scanning electron microscope. The tool monitor data obtained from the process is then used in model-type and / or measurement-type analysis to determine the degree of correction, which is then applied as described in steps 92, 93, 94 and 95. If corrections need to be readjusted, the entire preprocessing must be started from scratch using vector-formatted input pattern data. For example, this data processing pipeline may be advantageous for multi-column tools where each sub-column requires frequent readjustment of the sub-column's illumination system correction due to variations in, for example, one or more individual sources. Such readjustments can be applied in data preprocessing units that practically require starting preprocessing from scratch. Ideally, this preprocessing can be run in parallel with the currently running streaming process so that intermediate data can only be replaced for the next streaming process after the currently running streaming process is complete. This ensures that the lithography equipment can operate without interruption while maintaining high quality.

[0077] Finally, the data streaming unit 182 reads the pixel package data PACKPIX from memory and sends the data (in real time) to the buffer memory SBUF associated with the corresponding drawing tool's PD system, symbolically shown as step 97 in Figure 8.

[0078] Vector data type intermediate data

[0079] Figure 9 shows a flowchart of the pattern data processing pipeline in data center 191 and data path 192 in another embodiment of the pattern data processing flow. In this example, the data preprocessing unit 191 converts the pattern input data in vector form, as in steps 91, 92, and 93 in the process of Figure 8, into a general raster graphics representation PIX that does not include column-specific corrections. As a result, i.e., the pixel data PIX forms intermediate data that is stored in a storage medium to ensure high transmission rate access by the data streaming unit. In this case, the data streaming unit 192 may include additional subunits for applying column-specific corrections to the pixel data during the streaming process, such as kernel COLMOD in step 94. Furthermore, in this variant, the workload of such corrections can be performed in parallel, i.e., distributed across multiple CPUs and / or multiple GPUs and / or multiple FPGAs, so that the processing can be performed in real time, i.e., without significantly reducing the data streaming rate. The following steps 95 and 96 may be performed similarly in the data streaming unit, and steps 97 onward are performed similarly to those in the process shown in Figure 8. This alternative separation of the data center and data path allows for frequent adjustment of the correction subunit(s) if the drawing conditions require it. For example, in this case of a data processing pipeline, it is particularly advantageous for multi-column tools where each column requires frequent readjustment of the correction for the sub-column's illumination system due to variations in, for example, one or more individual sources. Such readjustments are determined by in-situ measurements and can be applied immediately in the correction unit of step 94 without interrupting the operation of the lithography apparatus or incurring additional burden.

[0080] Figure 10 shows an example flowchart of a pattern data processing pipeline in a data center 131 and data path 132 in yet another embodiment of the pattern data processing flow. In this example, the data processing unit 131 transforms the pattern input data in only the same two steps 91 and 92 described above to generate a data CORRGEOM of multiple corrected geometry elements. In this example, the data CORRGEOM forms intermediate data that is stored in persistent memory and accessed by the data streaming unit 132 on demand during the drawing process. This typically reduces the amount of intermediate data required to be stored by a multiple of 10 compared to the rasterized pixel data PIX. On the other hand, in this setup, the data streaming unit will perform the rasterization process 93 and the subsequent steps described above.

[0081] Data Streaming

[0082] As described above, data streaming units 182, 192, and 132 transmit pixel package data to the buffer memory SBUF associated with the PD system (in real time). For details of the drawing process, it is advisable to simultaneously adhere to two requirements: (1) Scanning stripe exposure involves (including) the movement of the target relative to a column containing a PD system that forms a beamlet array field. (2) The pixel package data transmitted to the PD system (in real time) must correspond to the pattern to be exposed on the target, if the beamlet array field covers each area of ​​the pattern to be exposed.

[0083] These two requirements, though stringent, can be met, for example, by using a two-stage streaming architecture for the streaming process shown in Figure 11. This two-stage streaming process is executed in a data streaming unit 112 of the associated drawing tool 1A / 1B, corresponding to one of units 102, 182, 192, or 132 in Figures 7 to 10.

[0084] In the first stage, the data streaming unit 112 acquires intermediate data from the storage device 11M, which may be stored in the local intermediate storage device LMDAT. From there, the data streaming unit 112 transmits the data to the buffer subunit SBUF after applying processing 113 to the intermediate data, if applicable. The intermediate data may correspond to data PACKPIX or its precursor data, such as PIX (Figure 9) or CORRGEOM (Figure 10), with respect to Figures 8 to 10. In the case of precursor data, such data would require additional processing 113, which can be executed as part of the streaming process within the streaming unit 112, to acquire the data in PACKPIX format (or an equivalent format); for example, in the case of PIX or CORRGEOM, processing 113 would include processing steps 94 to 96 in Figure 9 or processing steps 93 to 96 in Figure 10. If the intermediate data is PACKPIX data (Figure 8), these are already suitable for the PD system, and processing 113 is skipped. Data transmission between the storage device 11M and the data streaming unit 112 can be achieved using conventional optical communication links. The resulting data (in a data format such as PACKPIX) corresponds to a collection of raster graphics arrays or subgrid arrays, which are then written to a buffer subunit SBUF.

[0085] The data for a stripe scan typically corresponds to one or more subgrid arrays. As soon as all the data for the next stripe scan becomes available in the buffer subunit SBUF, a second stage of streaming, including the drawing process for that stripe, i.e., the pixel subgrid array, can be started or continued: each buffered data 114 is read from the buffer subunit SBUF and supplied to the PD system of the drawing tool at a constant rate, during which time the target can be moved relative to the PD device at a corresponding constant speed.

[0086] It is important to note that the streaming data 114 is supplied not only to the PD device 12 but also to the beam deflection device of the projection optical system 16, as this plays a crucial role during the drawing process: the beam deflection device helps to keep the beam(s) fixed at their positioning grid positions(s) during the exposure cycle and to shift the beam(s) to their next positions on the positioning grid. During the scanning stripe process, columns 1A / 1B are moved relative to the exposure target 7 by the positioning unit (XY stage) 5, while the target positioning is monitored by the laser measurement unit 8. The latter (laser measurement unit 8) feeds back position information 115 to the buffer subunit. The position information 115 can also be transmitted using a conventional optical communication link.

[0087] This feedback cycle, based on position information 115 combined with the buffer subunit SBUF, forms a highly advantageous feature of the second stage of the streaming architecture. This ensures that the pattern data transmitted (in real time) to the PD device 12 is synchronized with one of the targets 7, and that the pattern thus exposed is faithfully positioned at its design location.

[0088] All or part of the above embodiments may be described as follows, but are not limited to them. [Note 1] A method for performing a process of drawing a desired pattern on a target using at least one charged particle multibeam apparatus. The aforementioned desired pattern is provided as input pattern data encoded in vector format. The apparatus includes an illumination system, a pattern definition system, and a projection optical system. The illumination system is configured to generate a charged particle beam and direct the charged particle beam towards a pattern definition system. The pattern definition system is configured to form the charged particle beam into a plurality of subbeams representing the pattern (from the charged particle beam to a plurality of subbeams representing the pattern). The projection optical system helps to project the plurality of subbeams onto the target. The method described above includes at least the following steps, which are performed independently of the drawing process in the data preprocessing system: To receive the aforementioned input pattern data; To obtain intermediate pattern data, preprocess the input pattern data using the drawing parameter data provided to the data preprocessing system; and, Write the aforementioned intermediate pattern data to a data storage device. The method further includes at least the following steps, performed by at least one drawing control system associated with the at least one device during the initialization and / or execution of a drawing process using the at least one device: Read the aforementioned intermediate pattern data from the data storage device; To obtain pattern streaming (real-time transmission) data, the intermediate pattern data is converted; and, Transmit the pattern streaming data to at least one pattern definition system of the at least one device (in real time), wherein each pattern definition system forms the charged particle beam of each of the at least one device into the plurality of subbeams (from the charged particle beam of each of the at least one device to the plurality of subbeams) during the drawing process in accordance with the pattern streaming data. [Note 2] In the above method, the preprocessing of the input data in the data preprocessing system includes at least one of the following: The input pattern data, encoded in vector format, is divided into multiple basic geometric elements; To compensate for deviations from the ideal drawing state determined prior to the drawing process, geometric and / or dose corrections are applied to the input pattern data or geometric basic elements, wherein the geometric and / or dose corrections are performed using correction data supplied as external data based on the preceding drawing process and / or inspection process and / or model-type simulation. [Note 3] In the above method, the preprocessing of the input data in the data preprocessing system generates preprocessed intermediate pattern data in vector format, and the preprocessed intermediate pattern data is written to the data storage device as intermediate pattern data; Initialization and / or execution of a drawing process by the at least one drawing control system includes reading the intermediate pattern data from the data storage device and then converting the input pattern data to obtain raster graphics data that represents at least a portion of the desired pattern in raster graphics format. [Note 4] In the above method, the intermediate pattern data is raster graphics data for the raster to be used during the drawing process; The preprocessing of the input data in the data preprocessing system includes converting the input pattern data in order to obtain raster graphics data that represents at least a portion of the desired pattern in raster graphics format. [Note 5] In the above method, the step(s) performed by the at least one drawing control system during the drawing process further include at least one of the following: To compensate for deviations from the ideal rendering state determined during the rendering process, apply geometric and / or dose corrections to the raster graphics data; Dithering pixel data, including converting the data to match a predetermined grayscale; and, The pixel data is packaged into a data format that enables it to be transmitted (in real time) at the transmission speed required by the drawing process. [Note 6] In the above method, the preprocessing of the input data in the data preprocessing system includes at least one of the following after conversion to raster graphics data: To compensate for deviations from the ideal rendering state determined during the rendering process, apply geometric and / or dose corrections to the raster graphics data; Dithering pixel data, including converting the data to match a predetermined grayscale; and, The pixel data is packaged into a data format that enables the pixel data to be transmitted (in real time) at the transmission speed required by the drawing process. Subsequently, the processed data is written to the data storage device as intermediate pattern data. [Note 7] In the above method, the charged particle multibeam apparatus includes two or more multibeam columns arranged in parallel; The step of reading the intermediate pattern data from the data storage device and the subsequent steps are performed at least partially in parallel with respect to one or more targets in the multi-beam column. [Note 8] In the above method, the multi-beam columns arranged in parallel include at least two different illumination systems, preferably each of the multi-beam columns includes its own illumination system. [Note 9] In the above method, the step of reading the intermediate pattern data from the data storage device and the subsequent steps are performed by two or more separate drawing processes that are preferably performed simultaneously in a plurality of different charged particle multibeam devices. [Note 10] A charged particle multibeam apparatus configured to perform a drawing process for drawing a desired pattern onto a target. The apparatus includes an illumination system, a pattern definition system, and a projection optical system; The illumination system is configured to generate a charged particle beam and direct the charged particle beam toward a pattern definition system, the pattern definition system is configured to form the charged particle beam into a plurality of sub-beams representing the pattern, and the projection optical system helps to project the plurality of sub-beams toward the target. The apparatus further comprises a drawing control system; The aforementioned drawing control system is To access a data storage device in order to read intermediate pattern data containing preprocessed pattern data of the desired pattern, To obtain pattern streaming (real-time transmission) data, the intermediate pattern data is converted as follows: The pattern streaming data is transmitted (in real time) to the pattern definition system, and the pattern definition system is instructed to form the charged particle beam into the plurality of sub-beams during the drawing process according to the pattern streaming data. It is composed of. [Note 11] In the above-described apparatus, the drawing control system is configured to read and process the intermediate pattern data during the initialization and / or execution of the drawing process performed in the apparatus. [Note 12] In the above apparatus, the drawing control system is further configured to perform at least one of the following steps after reading the intermediate pattern data: Converting the pattern data in order to obtain raster graphics data that represents at least a portion of the desired pattern in raster graphics format; To compensate for deviations from the ideal rendering state determined during the rendering process, apply geometric and / or dose corrections to the raster graphics data; Dithering pixel data, including transforming the data to match a predetermined grayscale; The pixel data is packaged into a data format that enables it to be transmitted (in real time) at the transmission speed required by the drawing process. [Note 13] In the above apparatus, the charged particle multibeam apparatus includes two or more multibeam columns arranged in parallel, configured to perform at least partially in parallel with respect to one or more targets the step of accessing the data storage device to read intermediate pattern data and the subsequent steps(s). [Note 14] In the above apparatus, the plurality of multi-beam columns arranged in parallel include at least two different illumination systems, preferably each of the plurality of multi-beam columns includes its own illumination system.

[0089] Within the framework of the full disclosure of the present invention (including the claims and drawings), further modifications and adjustments to the embodiments are possible based on the fundamental technical concept. Furthermore, within the framework of the full disclosure of the present invention, various combinations or selections (including "non-selection") of various disclosed elements (including each element of each claim, each element of each embodiment, each element of each drawing, etc.) are possible. In other words, the present invention naturally includes the full disclosure, including the claims and drawings, and various modifications and alterations that a person skilled in the art could make in accordance with the technical concept of the present invention. In particular, with respect to the numerical ranges described herein, any numerical value or sub-range included within that range should be interpreted as being specifically described, even if not otherwise stated.

[0090] Furthermore, the reference numerals in the drawings included in the claims are solely for the purpose of aiding the understanding of the invention, and are not intended to limit the present invention to the embodiments and illustrated examples.

[0091] Furthermore, the full contents of each of the above-mentioned documents are incorporated into this book through reference and are included herein.

Claims

1. A method for performing a writing process of a desired pattern on a target using at least one charged particle multi-beam device (1A, 1B), comprising: The desired pattern is provided as input pattern data (INPDAT) encoded in vector format; the apparatus includes an illumination system, a pattern definition system, and a projection optical system; the illumination system is configured to generate a charged particle beam and direct the charged particle beam towards a pattern definition system, the pattern definition system being configured to form the charged particle beam into a plurality of sub-beams representative of the pattern, and the projection optics serves to project the plurality of sub-beams onto the target (7, 7'); The method comprises at least the following steps, which are executed in a data pre-processing system (101) independently of the rendering process: receiving said input pattern data (INPDAT); pre-processing the input pattern data using writing parameter data provided to the data pre-processing system to obtain intermediate pattern data (IMDAT); writing said intermediate pattern data into a data storage device (11M); Including, The method further comprises at least the following steps, executed by at least one drawing control system (14A, 14B) associated to said at least one device (1A, 1B) during initialization and / or execution of a drawing process using said at least one device (1A, 1B): reading the intermediate pattern data from the data storage device; Transforming the intermediate pattern data (LMDAT) to obtain pattern streaming data. sending said pattern streaming data to at least one pattern definition system of said at least one apparatus, each pattern definition system forming a respective charged particle beam of said at least one apparatus into said plurality of sub-beams during said writing process in response to said pattern streaming data; Contains The method comprising:

2. 10. The method of claim 1 , The pre-processing of the input data in the data pre-processing system (101) includes at least one of the following: Splitting (91) the input pattern data (VECPAT) encoded in vector format into a plurality of geometry primitives (SPLITGEOM); applying (92) geometry and / or dose corrections to said input pattern data (VECPAT) or geometry primitives (SPLITGEOM) in order to compensate for deviations from ideal writing conditions determined prior to said writing process, said geometry and / or dose corrections being performed using correction data (DOSEMOD, DISLOCMOD) provided as external data based on a previous writing process and / or an inspection process and / or a model-based simulation; The method comprising:

3. 3. The method of claim 2, pre-processing of the input data in the data pre-processing system (101) produces pre-processed intermediate pattern data in vector form (SPLITGEOM, CORRGEOM), the pre-processed intermediate pattern data being written to a data storage device as intermediate pattern data; and said initiating and / or executing a drawing process by said at least one drawing control system (14A, 14B) includes, after reading said intermediate pattern data from said data storage device, converting (93) said input pattern data to obtain raster graphics data (PIX, PACKPIX) representing at least a portion of said desired pattern in raster graphics format. The method comprising:

4. 10. The method of claim 1 , the intermediate pattern data being raster graphics data for a raster to be used during the rendering process; Pre-processing the input data in the data pre-processing system (101) includes converting (93) the input pattern data to obtain raster graphics data (PIX, PACKPIX) representing at least a portion of the desired pattern in a raster graphics format. The method comprising:

5. The method according to claim 3 or 4, The steps performed during the drawing process by the at least one drawing control system (14A, 14B) further include at least one of the following: applying (94) geometry and / or dose corrections to the raster graphics data (PIX) to compensate for deviations from ideal rendering conditions determined during the rendering process; dithering (95) pixel data, including converting said data to conform to a predetermined grey value scale; packaging (96) the pixel data into a data format that allows said pixel data to be transmitted at a transmission rate required by said drawing process; The method comprising:

6. 5. The method of claim 4, The pre-processing of the input data in the data pre-processing system (101), after conversion to raster graphics data, includes at least one of the following: applying (94) geometry and / or dose corrections to the raster graphics data (PIX) to compensate for deviations from ideal rendering conditions determined during the rendering process; dithering (95) pixel data, including converting said data to conform to a predetermined grey value scale; packaging (96) the pixel data into a data format that allows said pixel data to be transmitted at a transmission rate required by said drawing process; Thereafter, the thus processed data is written as intermediate pattern data to the data storage device. The method comprising:

7. 10. The method of claim 1 , The charged particle multi-beam device includes two or more multi-beam columns (c1, c2, c3; c4, c5, c6) arranged in parallel; and The step of reading out the intermediate pattern data from the data storage device and subsequent steps are performed at least partially in parallel for one or more targets (7, 7') in the multi-beam column. The method comprising:

8. 8. The method of claim 7, The multi-beam columns arranged in parallel include at least two different illumination systems, each of the multi-beam columns including its own illumination system. The method comprising:

9. 10. The method of claim 1 , The step of reading out the intermediate pattern data from the data storage device and subsequent steps are performed by two or more separate writing processes performed in a plurality of different charged particle multi-beam devices (1A, 1B). The method comprising:

10. A charged particle multi-beam device (1A, 1B) configured to perform a writing process for writing a desired pattern on a target, comprising: The apparatus includes an illumination system, a pattern definition system, and a projection optical system; the illumination system is configured to generate a charged particle beam and direct the charged particle beam towards a pattern definition system, the pattern definition system being configured to form the charged particle beam into a plurality of sub-beams representative of the pattern, and the projection optics serves to project the plurality of sub-beams onto the target (7, 7'); The apparatus further comprises a drawing control system (14A, 14B); The drawing control system (14A, 14B) accessing a data storage device to read intermediate pattern data (LMDAT) including pre-processed pattern data of the desired pattern; converting said intermediate pattern data (LMDAT) to obtain pattern streaming data (SBUF); sending the pattern streaming data to the pattern definition system and instructing the pattern definition system to form the charged particle beam into the plurality of sub-beams during the writing process in response to the pattern streaming data (SBUF); What is configured An apparatus comprising:

11. 11. The apparatus of claim 10, said drawing control system being adapted to read and process said intermediate pattern data during initialization and / or execution of a drawing process executed in said device (1A, 1B); An apparatus comprising:

12. 12. The device according to claim 10 or 11, The writing control system is further configured to perform at least one of the following steps after retrieving the intermediate pattern data: Transforming (93) the pattern data to obtain raster graphics data (PIX) representing at least a portion of the desired pattern in a raster graphics format; applying (94) geometry and / or dose corrections to the raster graphics data (PIX) to compensate for deviations from ideal rendering conditions determined during the rendering process; dithering (95) pixel data, including converting said data to conform to a predetermined grey value scale; packaging (96) the pixel data into a data format that allows said pixel data to be transmitted at a transmission rate required by said drawing process; An apparatus comprising:

13. 11. The apparatus of claim 10, the charged particle multi-beam device comprises two or more parallel arranged multi-beam columns (c1, c2, c3; c4, c5, c6) configured to perform the step of accessing the data storage device to read out intermediate pattern data and subsequent steps at least partially in parallel on one or more targets (7, 7'); An apparatus comprising:

14. 14. The apparatus of claim 13, The plurality of multi-beam columns arranged in parallel include at least two different illumination systems, and each of the plurality of multi-beam columns includes its own illumination system. An apparatus comprising:

15. The method of claim 1, The step of reading out the intermediate pattern data from the data storage device and subsequent steps are performed by two or more separate writing processes performed simultaneously in a plurality of different charged particle multi-beam devices (1A, 1B). The method comprising:

Images