Semiconductor metrology with information from multiple process steps

Abstract

Metrology system (400) comprising: a lighting subsystem (402) that provides a quantity of illumination light (406) for one or more metrology targets arranged on a wafer (401) that has previously been processed through a plurality of process steps of a manufacturing process sequence; a detector subsystem (404) that detects a quantity of light (408) from the one or more metrology targets in response to the quantity of illumination light (406) and generates a quantity of measurement signals (411) in response to the quantity of detected light (408); and a computer system (430), trained to Receiving an initial amount of process information (421) from a first process machine (420) which is used to perform a first process step of the plurality of process steps on the wafer (401); Receiving a second set of process information (421) from the first process machine (420) or another process machine that is used to perform a second process step of the plurality of process steps on the wafer (401); and Estimating the value of a structural parameter of interest (440) of one or more metrology targets based on the set of measurement signals (411) and the first and second sets of process information (421).

Description

CROSS-REFERENCE TO RELATED REGISTRATION

[0001] The present patent application claims priority pursuant to 35 USC §119 of provisional US patent application 62 / 318,166 entitled “Process Information Assisted Metrology”, filed on April 4, 2016. TECHNICAL AREA

[0002] The described embodiments relate to metrology systems and methods and special methods and systems for improved measurement of parameters that characterize the dimensions of structures that have been produced by multiple structuring processes. BACKGROUND INFORMATION

[0003] Semiconductor devices, such as logic and memory devices, are typically manufactured through a series of processing steps applied to a single device. These processing steps create the various structures and multiple layers of structure found in semiconductor devices. Lithography, for example, is a semiconductor manufacturing process that involves creating a structure on a semiconductor wafer. Additional examples of semiconductor manufacturing processes include, but are not limited to, chemical-mechanical polishing, etching, deposition, and ion implantation. Multiple semiconductor devices can be fabricated on a single semiconductor wafer and then separated into individual semiconductor devices.

[0004] Multiple structuring techniques are now commonly used to increase the resolution of structures formed on the semiconductor wafer for a given lithographic system. Fig. Figures 1A-1D show a double patterning lithography technique (DPL), commonly referred to as a litho-etch-litho-etch process (LELE). Fig. Figure 1A shows a silicon substrate layer 10, an interface layer 11 (e.g., silicon dioxide), a component layer 12, a hard mask layer 13, a sacrificial layer 14, and a structured photoresist layer 15 resulting from a lithographic structuring step. The in Fig. The structure shown in Figure 1A is then subjected to exposure and etching steps, resulting in the structure shown in Figure 1A. Fig. The structure shown in Figure 1B is obtained. In this structure, the structure of the photoresist layer 15 was practically transferred to the hard mask layer 13. Both the sacrificial layer 14 and the structured photoresist layer 15 were removed. A number of deposition and lithography steps are applied to achieve the structure shown in Figure 1B. Fig. to arrive at the structure shown in 1C. Fig. Figure 1C shows a further sacrificial layer 16 and a structured photoresist layer 17, which are built up on the hard mask layer 13. The structured photoresist layer 17 contains a structure with the same pitch as the first structured photoresist layer 15 and also the same pitch as the structure etched into the hard mask layer 13. However, the structured photoresist layer 17 is offset from the structure of the hard mask layer 13 by half the pitch of the structured photoresist layer 17. The in Fig. The structure shown in Figure 1C is then subjected to exposure and etching steps, resulting in the structure shown in Figure 1C. Fig. The structure shown in 1D is shown. In this structure, the structure of the photoresist layer 17 has been practically transferred to the hard mask layer 13. Both the sacrificial layer 16 and the structured photoresist layer 17 have been removed. Fig. Figure 1D shows a structure etched into the hard mask 13 with double the pitch of the structured photoresist layers 15 and 17, which were created by the mask of the lithography system.

[0005] Fig. Figure 1D also shows the effects of a non-optimized DPL process. Ideally, the nominal pitch of the doubly structured structure should be a constant value P. However, due to deficiencies in the DPL process, the pitch of the resulting structure can vary depending on its position, due to lattice irregularities. This is commonly referred to as "pitch walk." A deviation from the nominal pitch, P, is shown in Fig. 1D is shown as ΔP. ​​In another example, a critical dimension of each resulting structure should have the same nominal value, CD. However, a critical dimension (e.g., middle critical dimension, lower critical dimension, etc.) of the resulting structure may vary depending on its position due to deficiencies in the DPL process. A deviation from the desired critical dimension, CD, is in Fig. 1D shown as ΔCD.

[0006] Pitch drift and ΔCD are examples of geometric errors caused by flaws in the DPL process, such as misalignment between the two lithographic layers, inconsistencies in focus and exposure during the lithographic process, errors in the mask structure, etc. Both pitch drift and ΔCD result in a unit cell that is larger than expected. Although pitch drift and ΔCD are specifically described, other multi-structuring errors can also be considered.

[0007] Although the LELE process refers to the Fig. As described in 1A-1D, many other multiple structuring processes that lead to similar errors can be considered (e.g., litho-litho-etch, multiple litho-etch structuring, multiple structuring defined by a spacer element, etc.). Similarly, although a double structuring process with reference to the Fig. Similar errors described in 1A-1D occur in higher-order structuring processes, such as quadruple structuring. Errors like pitch gait and ΔCD are usually more pronounced in structures resulting from higher-order structuring processes.

[0008] Metrology processes are used at various stages during semiconductor manufacturing to detect defects on wafers, thereby increasing throughput. Optical metrology techniques offer the potential for high-throughput measurements without the risk of sample destruction. A number of optical metrology-based techniques, including scatterometry and reflectometry applications, along with associated analysis algorithms, are commonly used to characterize critical dimensions, layer thicknesses, composition, and other parameters of nanoscale structures.

[0009] A common approach to controlling a semiconductor manufacturing process is to use a metrology machine after each critical process step. Generally, process control improves when critical steps are monitored more closely.

[0010] However, implementing a metrology step after each process step is expensive, both in terms of manufacturing time and cost. As the number of critical process steps has increased for advanced technology stages, inserting a metrology step after each critical process step has become prohibitively expensive. Furthermore, with each metrology stage, the number of critical process steps that can be measured cost-effectively decreases compared to the total number of critical process steps.

[0011] If metrology steps are eliminated from the process flow for advanced technology stages, such as a LELE multi-structuring process, metrology results become ineffective for process control purposes for all intermediate process steps. For example, in a LELE multi-structuring process, a metrology step can only be performed after the last process step. The results of this measurement can be effectively used to correct the last process step, but not the earlier ones. This limitation persists even when multiple metrology steps are used. The last process step before metrology can be effectively corrected, but the current metrology does not provide sufficient measurement information to correct all other process steps.

[0012] Metrology applications involving the measurement of structures generated by multiple process steps, and especially multi-structuring processes, present challenges due to the practical limitations of the number of insertion points for metrology. Increasing demands for resolution, correlation between multiple parameters, increasingly complex geometric structures, and the growing use of opaque materials exacerbate this problem and leave undesirable gaps in the control of advanced manufacturing processes. Therefore, methods and systems for improved process control of advanced manufacturing processes are highly desirable.

[0013] US 2016 / 0003609A1 describes metrology procedures that utilize a trained metrology model. Reference measurements are performed using a reference metrology system to determine the actual parameter values ​​of component structures of interest. Optical metrology measurements of neighboring metrology targets are also performed. The trained metrology model links the results of the optical metrology measurements with the reference measurements. The trained metrology model allows for the determination of the actual parameter values ​​of component structures of interest based on optical metrology measurements of neighboring metrology targets.

[0014] US 2015 / 0176985A1 concerns the use of a measurement model to determine a geometric error in a multi-structuring process on a wafer. Measurement data is collected from a large number of measurement points, each adjacent to a single-structured metrology target and a multi-structured metrology target. OVERVIEW

[0015] This paper presents methods and systems for measuring process-induced defects in a multi-structured semiconductor fabrication process. These methods are based on measurements of a sample and process information from one or more preceding process steps used to fabricate the sample. Based on the measured defects, corrected process parameter values ​​are transmitted to the appropriate process machine to improve process performance. In this way, a metrology machine provides corrections for one or more process machines used to perform any preceding process step.

[0016] The metrology system receives process information from any process machine used to perform any preceding process step. Process machines include lithography machines, etching machines, deposition machines, chemical-mechanical planarization (CMP) machines, etc. Process information received from the metrology machine includes, but is not limited to, process control parameters, process machine setting parameters, process environment parameters, process data acquired by sensors in a process machine, metrology data acquired by sensors in a process machine, etc.

[0017] In one aspect, a metrology machine is used in a metrology step after a number of process steps have been executed. The metrology machine measures structural parameters of interest of metrology targets on the wafer in its physical state and communicates correctable process parameter values ​​to one or more process machines involved in one or more of the preceding process steps. When executed by the appropriate process machine, the correctable process parameter values ​​reduce process-induced errors in the geometry of the structures produced by the process.

[0018] In another aspect, multiple metrology machines are used to control a manufacturing process in combination with process information from one or more process steps. In addition to the process information, metrology information from an additional metrology step inserted into the process is also used to improve the metrology of the structure and to enhance process control.

[0019] In some embodiments, a metrology machine uses a physically based measurement model to estimate the values ​​of structural parameters of interest from the measurement data (e.g., measured spectra) and the process information.

[0020] In some other embodiments, a metrology machine uses an input-output measurement model to estimate the values ​​of structural parameters of interest from the measurement data (e.g., measured spectra) and process information. These models include signal-response metrology models, neural network models, support vector machine models, etc.

[0021] In another aspect, a signal response metrology (SRM) measurement model is trained using measurement signals and associated process information from multiple targets integrated into a set of multiple targets, and operates on measurement signals from these same multiple targets. This approach decorrelates critical parameters to each other and to other process variations.

[0022] In some embodiments, auxiliary targets are arranged alongside the primary measurement target and are subject to the same process variations. In these embodiments, the training set of metrology targets includes a primary, nominally dimensioned target and one or more auxiliary targets, which have different nominal values ​​of the parameters of interest.

[0023] The above is an overview and therefore necessarily contains simplifications, generalizations, and omissions of details; thus, it is clear to those skilled in the art that the overview serves only for illustrative purposes and is in no way limiting. Other aspects, inventive features, and advantages of the devices and / or processes described herein will become apparent from the non-limiting detailed description set forth herein. BRIEF DESCRIPTION OF THE DRAWINGS Fig. Figures 1A-1D show selected steps of a double structuring lithography technique (DPL), which is commonly referred to as a litho-etch-litho-etch process (LELE). Fig. Figure 2 shows a manufacturing process sequence 100, which includes a sequence of manufacturing process steps and a single metrology step. Fig. Figure 3 shows an example of trench formation, which is characterized by a specific example of the in Fig. 2 shows a process flow of 100. Fig. Figure 4 shows a manufacturing process flow 200, which includes a sequence of manufacturing process steps and two metrology steps. Fig. Figure 5 shows a process flow 300, which is an example of a self-aligned octuplet structuring process. Fig. Figure 6 shows an example of the formation of rib spacing elements, illustrated by a specific example of the one in Fig. 300 can be generated using process 5 shown. Fig. Figure 7 shows a semiconductor wafer 130 with a number of metrology targets located at various measurement points on the surface of the wafer. Fig. Figure 8 shows a system 500 for measuring properties of a specimen according to the exemplary methods shown herein. Fig. Figure 9 is a flowchart showing a procedure 500 for determining one or more parameter values ​​that characterize geometric errors induced by a multiple structuring process based on measurements and process information. DETAILED DESCRIPTION

[0024] Reference will now be made in detail to background examples and some embodiments of the invention, examples of which are shown in the accompanying drawings.

[0025] This document describes methods and systems for measuring process-induced errors after several steps in a multi-structured semiconductor fabrication process, based in part on process information from one or more of the steps. Based on the measured errors, corrected process parameter values ​​are communicated to the appropriate process machine to improve process performance. In this way, the metrology machine provides corrections not only for the process machine used to perform the last process step before the metrology machine measurement, but also for one or more process machines used to perform any of the preceding process steps.

[0026] The metrology system receives process information from any process machine used to perform any of the preceding process steps. Process machines include lithography machines, etching machines, deposition machines, chemical-mechanical planarization (CMP) machines, etc. Process information received by the metrology machine includes, but is not limited to, process control parameters, process machine setting parameters, process environmental parameters, process data acquired by sensors in the process machine, metrology data acquired by sensors in the process machine, etc. In some examples, a process machine includes integrated metrology sensors to measure process information that is communicated to the metrology machine. For example, a lithography machine may include an optical reflectometer to measure wafer geometry.In another example, an etching machine can incorporate an optical emission spectroscopy sensor to monitor and control the machine's plasma source. These sensors alone are insufficient to provide the information necessary for full metrology of a device structure. However, the inventors have discovered that, in combination with metrology signals generated by a metrology machine, these signals enable the metrology and process control of multi-step manufacturing processes that would otherwise be uncontrolled.

[0027] Fig. Figure 2 shows a manufacturing process sequence 100, which includes a series of manufacturing process steps, including a lithography step 101, an etching step 102, another lithography step 103, another etching step 104, and finally a metrology step 105. An incoming wafer in a specific physical state 114 within the manufacturing process is transformed by the lithography step 101 into a different physical state 115 as a result of the lithography process. Similarly, the etching step 102 transforms the wafer from state 115 to state 116, the lithography step 103 transforms the wafer from state 116 to state 117, and the etching step 104 transforms the wafer from state 117 to state 118.

[0028] In one aspect, a metrology machine is used in metrology step 105 to measure structural parameters of interest of metrology targets on the wafer in physical state 118, and to communicate correctable process parameter values ​​to one or more process machines involved in one or more of process steps 101-104. When executed by the appropriate process machine, the correctable process parameter values ​​reduce process-induced errors in the geometry of the structures produced by process step 100.

[0029] As in Fig. As shown in Figure 2, process information 106 is communicated from a lithography machine, used to perform lithography step 101, to the metrology machine, which is used to perform metrology step 105. Similarly, process information 107 is communicated from an etching machine, used to perform etching step 102, to the metrology machine; process information 108 is communicated from a lithography machine, used to perform lithography step 103, to the metrology machine; and process information 109 is communicated from an etching machine, used to perform etching step 104, to the metrology machine.

[0030] Although, as in Fig. As shown in Figure 2, process information from each process step of process flow 100 is communicated to the metrology machine; in general, process information from any one or more of the process steps can be communicated to the metrology machine.

[0031] As in Fig. As shown in 2, the metrology machine generates correctable process parameter values ​​based on the values ​​of one or more structural parameters of the wafer in state 118, measured by the metrology machine in metrology step 105, and the received process information (e.g., any of the process information 106-109). For example, as shown in Fig. Figure 2 shows that correctable process parameter values ​​110 are communicated to the lithography machine used to perform lithography step 101. Similarly, correctable process parameter values ​​111 are communicated to an etching machine used to perform etching step 102, correctable process parameter values ​​112 are communicated to the lithography machine used to perform lithography step 103, and correctable process parameter values ​​113 are communicated to an etching machine used to perform etching step 104.

[0032] Although, as in Fig. As shown in Figure 2, correctable process parameter values ​​can be communicated to a machine that executes each process step of process sequence 100. In general, correctable process parameter values ​​can be communicated to a machine that executes one or more of the process steps.

[0033] Fig. Figure 3 shows an example of trench formation, which is caused by a specific example of the in Fig. 2 shows a process flow of 100. Fig. Figure 3 shows a metrology target at each stage of trench formation within the lithography-etch-lithography-etch (LELE) process sequence 100. In physical state 115 (after lithography step 101), the structure of interest comprises a base layer 121, a component layer 122, and a structured photoresist layer 123. In physical state 116 (after etching step 102), the structured photoresist layer 123 has been completely removed, along with a portion of the component layer 122 that was exposed to light during lithography step 101. At this point, the first trench structure of the component layer has been formed. In physical state 117, a sacrificial layer 124 and another structured photoresist layer 125 are added during lithography step 103. In physical state 118, the structured photoresist layer 125, the sacrificial layer 124 and a further part of the component layer 122 are removed by etching step 104.At this point, the second trench structure of the building element layer is formed, and the resulting structure is measured with a metrology machine in metrology step 105.

[0034] In this example, the metrology machine is able to measure a critical dimension associated with each trench feature, CD1 and CD2. However, without additional process information, the metrology system cannot determine which trench feature was produced by which lithography step. In this example, dose information 106 is communicated from the lithography machine performing lithography step 101 to the metrology machine performing metrology step 105. Additionally, dose information 108 is communicated from the lithography machine performing lithography step 103 to the metrology machine. Based on the received dose information, the metrology machine assigns each trench to the corresponding lithography step that produced that trench. In this example, a larger dose in lithography step 101 (i.e.,Dose1 > Dose2) a larger critical dimension (i.e., CD1 > CD2). In this way, the first trench, with dimension CD1, is assigned to lithography step 101, and the second trench, with dimension CD2, to lithography step 103.

[0035] In another aspect, the metrology machine generates a correctable process parameter value 110 and communicates this to the lithography machine, which executes lithography step 101 to correct the dimension of trench 1. Similarly, the metrology machine generates a correctable process parameter value 112 and communicates this to the lithography machine, which executes lithography step 103 to correct the dimension of trench 2.

[0036] In some examples, the metrology machine is capable of performing measurements of structural parameters of interest. For instance, in some embodiments, the metrology machine is capable of measuring CD1 and CD2 independently, as referenced in Fig. 2 described. However, in some other examples, the metrology machine is not able to measure all parameters of interest. Quite often, a metrology machine is able to measure the mean trench size (i.e., (CD1+CD2) / 2) because the measured signals are primarily sensitive to changes in volume rather than position. In these examples, process information received by the metrology machine allows for the measurement of both CD1 and CD2 independently. In this way, process information, combined with metrology signal information, enables improved metrology capability in addition to improved process control.

[0037] Fig. 2 shows a LELE or (LE) 2 -Manufacturing process flow. However, the methods and systems described herein can generally be applied to any multi-structuring process flow, such as a (LE) N -Manufacturing process flow involving N litho-etching steps, where N is any positive integer, any self-aligned multiple structuring technique, etc., is applied.

[0038] In another aspect, multiple metrology machines are used to control a manufacturing process, in combination with process information from one or more process steps in the process flow. Additionally, with reference to Fig. In addition to the process information described in section 2, metrology information from an additional metrology step inserted into the process flow is also used to improve the metrology of the structure and to improve process control.

[0039] Fig. Figure 4 shows a manufacturing process flow 200, which includes a sequence of manufacturing process steps and two metrology steps. Elements with the same number are analogous to those referenced in... Fig. 2 described. As in Fig. As shown in Figure 4, the process includes lithography steps 101 and 103, and etching steps 102 and 104, as described in reference to Fig. 2 described. However, process flow 200 additionally includes two metrology steps. Metrology step 201 is inserted in the middle of process flow 200, and metrology step 202 is performed at the end of process flow 200.

[0040] In one aspect, the wafer in physical state 117 is measured by a metrology machine performing metrology step 201, and a statement of these measurement results is communicated to a metrology machine performing metrology step 202 at the end of process sequence 200. In metrology step 202, a metrology machine is used to measure structural parameters of interest of the metrology targets on the wafer in physical state 118, based on process information obtained from any of the preceding process steps (i.e., steps 101-104) and the measurement results 203 obtained from the intervening metrology step 201.

[0041] As with reference to Fig. As described in section 2, the metrology machine executing metrology step 202 generates correctable process parameter values ​​for one or more of the process steps of process sequence 200 based on the measurement results obtained in metrology step 202. These correctable process parameter values ​​are communicated to one or more process machines involved in one or more of process steps 101-104. When executed by the appropriate process machine, the correctable process parameter values ​​reduce process-induced errors in the geometry of the structures produced by process sequence 200.

[0042] The Fig. Figures 2 to 4 show examples of multiple structuring processes from (LE) N-Type. However, the metrology and process control techniques described herein are also applicable to self-aligned multi-structuring processes. Advanced process stages (e.g., 5-nanometer and 3.5-nanometer process stages) require complex structuring schemes to achieve desired rib pitches. For example, to achieve a rib pitch below 20 nanometers, a self-aligned octuplet structuring process (SAOP) may be necessary.

[0043] Fig. Figure 5 shows a process flow 300, which is an example of a SAOP process. The process flow 300 includes a lithography step followed by a repeating sequence of etching and deposition steps. As shown in Fig. As shown in Figure 5, an incoming wafer in a specific physical state 310 is transformed within the manufacturing process by lithography step 301 into a different physical state 311 as a result of the lithography process. Similarly, etching step 302 transforms the wafer from state 311 to state 312, deposition step 303 transforms the wafer from state 312 to state 313, etching step 304 transforms the wafer from state 313 to state 314, deposition step 305 transforms the wafer from state 314 to state 315, etching step 306 transforms the wafer from state 315 to state 316, deposition step 307 transforms the wafer from state 316 to state 317, and etching step 308 transforms the wafer from state 317 to state 318.

[0044] In one aspect, a metrology machine is used in metrology step 309 to measure structural parameters of interest of metrology targets on the wafer in physical state 318 and to communicate correctable process parameter values ​​to one or more process machines involved in one or more of process steps 301-308. When executed by the appropriate process machine, the correctable process parameter values ​​reduce process-induced errors in the geometry of the structures produced by process 300.

[0045] As in Fig. As shown in Figure 5, process information 319 is communicated from a lithography machine used to perform lithography step 301 to the metrology machine used to perform metrology step 309. Similarly, process information 320 is communicated from an etching machine used to perform etching step 302 to the metrology machine; process information 321 is communicated from a deposition machine used to perform deposition step 303 to the metrology machine; process information 322 is communicated from an etching machine used to perform etching step 304 to the metrology machine; process information 323 is communicated from a deposition machine used to perform deposition step 305 to the metrology machine; and process information 324 is communicated from an etching machine used to perform etching step 306 to the metrology machine.Process information 325 is communicated from a deposition machine used to perform deposition step 307 to the metrology machine, and process information 326 is communicated from an etching machine used to perform etching step 308 to the metrology machine.

[0046] Although as in Fig. As shown in Figure 5, process information from each process step of process flow 300 is communicated to the metrology machine; in general, process information from any one or more of the process steps can be communicated to the metrology machine.

[0047] As in Fig. As shown in section 5, the metrology machine generates correctable process parameter values ​​based on the values ​​of one or more structural parameters of the wafer in state 318, measured by the metrology machine in metrology step 309, and the received process information (e.g., any of the process information 319-326). For example, as shown in Fig. Figure 5 shows that correctable process parameter values ​​327 are communicated to the lithography machine used to perform lithography step 301. Similarly, correctable process parameter values ​​328 are communicated to an etching machine used to perform etching step 302, correctable process parameter values ​​329 are communicated to the deposition machine used to perform deposition step 303, correctable process parameter values ​​330 are communicated to an etching machine used to perform etching step 304, correctable process parameter values ​​331 are communicated to the deposition machine used to perform deposition step 305, correctable process parameter values ​​332 are communicated to an etching machine used to perform etching step 306, and correctable process parameter values ​​333 are communicated to the deposition machine.which is used to carry out the deposition step 307, and correctable process parameter values ​​334 are communicated to an etching machine which is used to carry out the etching step 308.

[0048] Although, as in Fig. As shown in Figure 5, correctable process parameter values ​​can be communicated to a machine that executes each process step of process sequence 300. In general, correctable process parameter values ​​can be communicated to a machine that executes one or more of the process steps.

[0049] Fig. Figure 6 shows an example of the formation of rib spacing elements, which are based on a specific example of the in Fig. 300 will be generated as shown in the process flow shown in section 5. Fig. Figure 6 shows a metrology target at each stage of rib spacer formation within the SAOP process 300. In physical state 311 (after lithography step 301), the structure of interest comprises a substrate layer 330, and a repeating sequence of nitride layers 331, 333, 335 and oxide layers 332, 334 and 336, a layer 337 of an antireflective underside coating (BARC), and a structured photoresist layer 338. In physical state 313 (after deposition step 303), the structured photoresist layer 123 and the BARC layer 337 are completely removed, along with part of the nitride layer 335, leaving two spacer structures 335A-B. At this point, the first set of rib spacer structures has been formed. In physical state 315, layers 334 and 335 are removed, along with part of the nitride layer 333, leaving four spacer element structures 333A-D.At this point, the second set of rib spacer structures is formed. In physical state 317, layers 332 and 333 are removed, along with part of the nitride layer 331, leaving eight spacer structures 331A-H. At this point, the third set of rib spacer structures is formed, and the resulting structure is measured by a metrology machine in metrology step 309.

[0050] In this example, the metrology machine is not able to directly measure a critical dimension associated with each rib spacer structure 331A-H. However, with additional process information, the metrology system is able to determine the dimensions of each rib spacer structure and to determine which rib spacer structures are sensitive to each process step. In this example, dose information 319 is communicated to the metrology machine by the lithography machine, which is used to perform lithography step 301, along with a specification 321 of the critical dimension CD1 of the spacers, a specification 323 of the critical dimension CD2 of the spacers, and a specification 325 of the critical dimension CD3 of the spacers.Based on the obtained dose and dimensional information, the metrology machine associates each rib spacer element with the corresponding process steps that produced the respective rib spacer element structure. In this way, the addition of process information to the metrology data enables the control of the SAOP process with a single metrology machine.

[0051] In some embodiments, a metrology machine uses a physically based measurement model to estimate the values ​​of structural parameters of interest from measurement data (e.g., measured spectra). Metrology techniques employing physical, model-based measurements typically require a parameterized geometric model of the structure under investigation. Examples of parameters include the critical dimension, pitch amplitude, or other parameters of interest. Additionally, an accurate electromagnetic model of the interaction between the optical system and the structure under investigation is required to simulate signals generated during the measurement. Nonlinear regression of the simulated signals against measured signals is used to determine parameters of the modeled structure. This approach requires accurate modeling of the structure and its material properties.

[0052] In these embodiments, process information received from machines used to perform preceding process steps is directly fed into the measurement model. In some examples, process parameter values ​​of the measurement model are set to values ​​obtained from the process machines. In other examples, the received process information is further processed to arrive at specific values ​​of model parameters or mathematical relationships between model parameters. In this way, the received process information is used to constrain the measurement model and reduce parameter correlations.

[0053] In some other embodiments, a metrology machine uses an input-output measurement model to estimate the values ​​of structural parameters of interest from measurement data (e.g., measured spectra). These models include signal response metrology models, neural network models, support vector machine models, etc.

[0054] In another aspect, a trained input-output measurement model is used to estimate the values ​​of structural parameters of interest based on measured signals and process information obtained from machines used in previous process steps to produce the sample under investigation. The combination of measurement signals and process information provides more information needed to separate and measure critical structures than would otherwise be obtainable based on either measurement signals or process information alone.

[0055] In some examples, an SRM measurement model is generated based on process information from preceding process steps (e.g., simulated process data or actual process data associated with the fabrication of a design-of-experiment (DOE) wafer) and raw measurement data (e.g., simulated spectra or spectra acquired from the DOE wafer) acquired from measurement points, including multi-structure metrology targets (simulated or actual). Machine learning, feature extraction, and other techniques are used to generate a direct input-output model (i.e., a transfer function) that relates DOE process information and spectra of one or more multi-structure targets to corresponding reference measurements of the parameters of interest. In some embodiments, the training set of multi-structure metrology targets includes targets that are nominally identical, i.e.,The objectives differ from one another due to process variations. In some embodiments, the process variations that influence the parameters of interest are intentionally amplified for the purpose of training the model.

[0056] In one example, the transfer function combines process information and scatterometry signals with corresponding CD-SEM measurements of a [system / device]. Fig. The SAQP target shown in Figure 7 is related to this. An SRM model is generated for each parameter of interest, and the same model is used to perform subsequent measurements at other measuring points.

[0057] To train the SRM model, a computer system (e.g., computer system 330) receives a large amount of raw measurement data associated with measurements from numerous measuring points and corresponding process information from previous process steps. Each of the numerous measuring points contains a multi-structured metrology target characterized by at least one parameter of interest and generated by at least two structuring steps of a multi-structuring process. A value of the parameter(s) of interest is known at each of the numerous measuring points.

[0058] For model training purposes, measurement data can be acquired from any location with known perturbations of the design parameters, i.e., the structural or process parameters. These locations can be, for example, in the scribe line, on the device, or at other locations on the wafer where, for instance, lithographic exposure conditions or reticulation design properties vary over a range of values. Alternatively, measurement data can be acquired from different device locations (e.g., a location with dense structures and a location with isolated structures, or locations with two different CDs on the mask). Generally, the measurement data is acquired from different locations that are perturbed in a known manner. The perturbation can be known from mask data, equipment data acquisition (EDA) data, process data, etc.

[0059] In one example, arbitrary focus, exposure, and overlay are systematically varied across the device or wafer. In another example, a random focus and exposure matrix (FEM) is used to reduce the correlation with sublayer parameters, as described in US patent US 8,142,966 B2 to Izikson et al.

[0060] In a preferred embodiment, the set of systematic variations during the fabrication of an actual DOE wafer is implemented. The DOE wafer is then measured to generate the raw measurement data. A fabricated wafer contains systematic errors that cannot be easily modeled by simulation. For example, the effect of sublayers is more accurately captured by measurements on a real wafer. The contribution of the sublayers can be decorrelated to the measurement responses by modifying the process parameters during fabrication, e.g., by varying the focus and exposure for a fixed sublayer condition. In another example, the contribution of the sublayer can be mitigated by acquiring multiple datasets of structures with changing top-layer topography and a constant sublayer condition.In one example, the top layer can contain periodic structures and the bottom layer can be non-periodic.

[0061] Measurement points can be selected to increase measurement sensitivity. For example, measurements taken at line ends are most sensitive to focus changes. In general, measurements should be taken on structures that are most sensitive to changes in the parameter being measured.

[0062] Although performing actual measurements on DOE wafers is preferred, in some other examples the process information and measurement response of a DOE wafer can be simulated for other known structural parameter values. In these examples, the process information and raw measurement data are generated synthetically. For example, a process simulator such as the Positive Resist Optical Lithography (PROLITH) simulation software, available from KLA-Tencor Corporation, Milpitas, California (USA), can be used. In general, any process modeling technique or machine can be considered within the scope of this patent document (e.g., Coventor simulation software, available from Coventor, Inc., Cary, North Carolina, USA).

[0063] In some examples, the raw measurement data includes two ellipsometric parameters (ψ, Δ) over a spectral range obtained at different measurement points. However, in general, the measurement data can be any data that specifies structural or geometric properties of the structures formed on the surface of a semiconductor wafer.

[0064] In some examples, the measurement data are associated with simulations of process parameters and corresponding measurements of the measurement sites on the surface of a DOE wafer (e.g., Wafer 130). For example, the measurement data may include simulated spectral measurements associated with the multi-structure metrology target associated with each measurement site.

[0065] In some other examples, the measurement data are associated with actual measurements of the sampling points on the surface of a DOE wafer (e.g., Wafer 130). The measurement data include actual spectral measurements associated with the multi-structure metrology target associated with each sampling point.

[0066] In some examples, the measurement data are associated with measurements of the multiple measurement sites on a design-of-experiments (DOE) wafer, and the parameter(s) of interest characterizing the multi-structured metrology target is / are measured at each of the sites by a reference metrology system. The reference metrology system is a trusted metrology system, such as a scanning electron microscope (SEM), tunneling electron microscope (TEM), atomic force microscope (AFM), or an X-ray measurement system, such as a small-angle X-ray scatterometer (SAXS) or an X-ray fluorescence (XRF) system, capable of accurately measuring the parameter value. However, the reference metrology system generally lacks the capability to operate as an inline metrology system (e.g., due to low throughput, high measurement uncertainty for individual sites, etc.).

[0067] In some embodiments, process variations and associated parameter variations are organized in a design-of-experiments (DOE) structure on the surface of a semiconductor wafer (e.g., DOE wafers), for example as described herein with reference to Fig. 7 described. In this way, the measuring system queries different locations on the wafer surface, which correspond to different process and corresponding structural parameter values. In the section with reference to Fig. In the example described in point 7, the measurement data are associated with a DOE wafer, which has known fluctuations in the Fig. The 6 CD1 and CD2 are processed. For example, CD1 is varied by changing the dose during lithography, and CD2 is varied by changing the thickness of the spacer elements 335A-B. CD3 is the distance between spacer element 333D and spacer element 333E, associated with the adjacent unit cell (shown with dashed lines instead of shading). CD3 is related to CD1 and CD2 by equation (2). CD3=LithoPitch−4Tspacer−CD1−2CD2 where LithoPitch is a predefined pitch of the photoresist lattice structure, and T AbstandselementThe thickness of the spacer elements is 333A-D. Although in this example the dose and the thickness of the spacer elements are varied to generate the desired parameter fluctuations, in general measurement data associated with any known fluctuations of process parameters (e.g., lithography focus, exposure, and other local and global parameters), structural parameters, or both, can be considered.

[0068] Fig. Figure 7 shows a semiconductor wafer 130 with a number of dies (e.g., die 133) located at various measuring points on the surface of the wafer. In the Fig. In the embodiment shown in Figure 7, the dies are located at measuring points arranged in a rectangular grid pattern aligned with the x- and y-coordinate system 132 shown. Each die includes an SAQP metrology target 131. In the embodiment shown in Figure 7, the dies are located at measuring points arranged in a rectangular grid pattern aligned with the x- and y-coordinate system 132 shown. Each die includes an SAQP metrology target 131. Fig. In the embodiment shown in Figure 7, each multi-structured metrology target 131 comprises a set of lines resulting from a first structuring step, together with at least one further set of intermediate lines resulting from a subsequent step in the multi-structuring process. As a result, each multi-structured metrology target comprises a lattice structure with a repeating pattern of spacer element structures, similar to the multi-structured unit cell 131. The geometry of the multi-structured unit cell 131 is characterized by CD1, CD2, CD3, and the pitch gang, as described with reference to Figure 7. Fig. 6 described.

[0069] Wafer 130 contains an array of dies with different known structural parameter values. Thus, CD1 has different known values ​​depending on its position on wafer 130. In this way, wafer 130 can be considered a design-of-experiments (DOE) wafer. It is desirable that the DOE wafer contains a matrix of multi-structured metrology targets spanning the full range of structural parameter values ​​(e.g., CD1) expected to result from the underlying process window. As in Fig. As shown in Figure 7, the CD1 values ​​differ for different columns of dies (column index in the x-direction). Thus, the wafer contains 130 columns of dies, each with a different CD1 value depending on its position on the wafer. Furthermore, the CD1 values ​​extend beyond those expected to result from the process window.

[0070] In some embodiments, a batch of DOE wafers similar to DOE wafer 130 is produced, wherein each DOE wafer in the batch has a different known nominal CD2 value. CD2 is varied by changing the spacer thickness or the etching conditions, which affect the entire wafer. Thus, each DOE wafer is produced under slightly different process conditions to achieve different nominal CD2 values. Each DOE wafer in the batch contains a range of different known CD1 values, as previously described herein with reference to wafer 130.

[0071] In some examples, one or more features are extracted from the raw measurement data by reducing one dimension of the measurement data. Although this reduction is optional, if it is used, the SRM measurement model is at least partially determined based on the one or more extracted features. Similarly, one or more features are extracted from the process information received from previous process steps by reducing one dimension of the process information.

[0072] In general, the dimension of the measurement data, the process information, or both, can be reduced by a number of known methods, including principal component analysis, nonlinear principal component analysis, selection of individual signals from the second set of measurement data, and filtering of the second set of measurement data.

[0073] In some examples, measurement data, process information, or both are analyzed using principal component analysis (PCA), nonlinear PCA, core PCA, independent component analysis (ICA), fast fourier transform (FFT) analysis, discrete cosine transform (DCT) analysis, or a combination of these techniques to extract features that most strongly reflect the fluctuations in process parameters, structural parameters, or both occurring at the different measurement points. In other examples, a signal filtering technique may be applied to extract signal data that most strongly reflects the parameter fluctuations occurring at the different measurement points.In some other examples, individual signals that most strongly reflect the parameter fluctuations occurring at the different measurement points can be selected from several data points present in the measurement data. Although it is preferable to extract features from the measurement data and process information to reduce the size of the data subjected to subsequent analysis, this is not strictly necessary.

[0074] An SRM measurement model is determined based on process information and associated raw measurement signals, or based on reduced versions of one or both. A trained SRM measurement model is structured to receive measurement data generated by a metrology system at one or more measurement points and associated process information from preceding process steps, and to directly determine structural parameter values ​​associated with each measurement target. In a preferred embodiment, the SRM measurement model is implemented as a neural network model. In one example, the number of nodes in the neural network is selected based on the features extracted from the measurement data, the process information, or both.In other examples, the SRM measurement model can be implemented as a linear model, a polynomial model, a response surface model, a decision tree model, a random forest model, a support vector machine model, or other types of models.

[0075] The SRM measurement model is trained based on the known values ​​of the parameter of interest. In some examples, the trained SRM measurement model is generated using DOE process information, raw measurement data, and known parameter values. The model is trained so that its output matches the defined expected response for all spectra within the process variation space defined by the DOE spectra.

[0076] In some examples, the trained SRM model is used to calculate structural parameter values ​​directly from measurement data (e.g., spectra) acquired from actual device structures on other wafers (e.g., product wafers) and associated process information from preceding processes used to manufacture the measured structures. The SRM measurement model receives measurement data (e.g., measured spectra) and associated process information directly as input and provides parameter values ​​as output, thus being a trained input-output model.

[0077] Additional details regarding model creation, training, and use as part of the measurement process are described in US Patent US 8,843,875 B2 to Pandev, US Patent Publication US 2014 / 0297211 A1 by Pandev et al., US Patent Publication US 2014 / 0316730 A1 by Shchegrov et al., US Patent Publication US 2014 / 0172394 A1, US Patent Publication US 2015 / 0042984 A1 by Pandev et al., US Patent Publication US 2015 / 0046118 A1 by Pandev et al., and US Patent Publication US 2015 / 0235108 A1 by Pandev et al. US patent publication US 2016 / 0109230A1 by Pandev et al., and US patent publication US 2015 / 0323471A1 by Sapien et al.

[0078] In some examples, a computer system (e.g., computer system 330) receives a set of optical measurement data associated with measurements of a metrology target on the surface of a semiconductor wafer, and process information associated with previous process steps for fabricating the metrology target. The parameter(s) of interest indicate a geometric error induced by the multi-structuring process.

[0079] In one example, the structural parameters CD1, CD2, CD3 and pitch response of the Fig. The parameters of interest for the target structure shown are listed below. These parameters are given as a non-restrictive example. In general, many other structural parameters (e.g., slope angle, lower critical dimension, etc.) can be used to indicate geometric errors induced by a multi-structuring process.

[0080] In some embodiments, a measured product wafer contains an array of structures with nominal values. Thus, CD1, CD2, CD3, and the pitch gear have the same nominal values, regardless of their position on the wafer.

[0081] In some examples, the measurement data include two ellipsometric parameters (ψ, Δ) over a spectral range obtained at different measurement sites. The measurement data include spectral measurements associated with the multi-structured metrology target associated with each measurement site. While the measurement data in some examples are spectral measurements, in general, the measurement data can be any measurements that describe the structural or geometric properties of the structures formed on the surface of a semiconductor wafer.

[0082] The value of at least one parameter of interest associated with the metrology target is determined based on the measurement data, associated process information, and a trained SRM measurement model. The value of the parameter of interest indicates a geometric error induced by the multi-structuring process. The value of the parameter of interest is calculated directly from the trained SRM measurement model.

[0083] The value of the parameter of interest is stored in a memory (e.g. memory 332).

[0084] Due to the structural symmetry present in some targets, critical dimensions often cannot be directly measured solely from scatterometry signals derived from a single target and associated process information. For example, the scatterometry signals from a lattice structure exhibiting a positive perturbation in the critical dimension, CD (e.g., CD+x), are identical to the scatterometry signals from a lattice structure exhibiting a negative perturbation in CD (e.g., CD-x).

[0085] In another aspect, an SRM measurement model is trained based on measurement signals and associated process information from multiple targets integrated into a multi-target set, and operates with measurement signals from the same multiple targets. This approach decorrelates critical parameters to each other and to other process variations.

[0086] In some embodiments, auxiliary targets are arranged alongside the primary measurement target and are subject to the same process variations (e.g., SAQP process variations). In these embodiments, the training set of metrology targets includes a primary target with nominal dimensions and one or more auxiliary targets that have different nominal values ​​of the parameters of interest.

[0087] Auxiliary targets are formed during the lithography process steps. In some examples, a mask with a different line-to-space ratio and / or pitch can be used to generate auxiliary targets. It is preferred to position the primary target and auxiliary targets as close together as possible to increase the accuracy of the SRM measurement model. In some embodiments, both the primary target and the auxiliary targets are positioned adjacent to each other at each measurement location. By positioning the metrology targets close together, it is less likely that simplifying assumptions for linking parameters of both metrology targets will induce significant errors. For example, it is very likely that the thickness of an underlying layer will have the same value for both metrology targets as long as the targets are positioned close together.Thus, for adjacent metrology targets, the thickness of the underlying layer can be treated as the same constant value without leading to a significant error.

[0088] The use of auxiliary targets for training and for using an SRM measurement model is analogous to the single-target approach described earlier. However, training the multi-target SRM measurement model additionally requires training data from the auxiliary targets and the primary measurement target. Similarly, using the multi-target SRM measurement model requires measurement data from the auxiliary targets and the primary measurement target. However, reference measurement data for training only needs to be collected from the primary target, as the specific parameter values ​​associated with the auxiliary targets are not relevant.

[0089] In some embodiments, an SRM measurement model is trained and used based on measurement signals from a metrology target measured at multiple steps of the multi-structuring process. Measured spectra or measured parameters of interest from one or more preceding process steps are passed on for training and use of the SRM measurement model associated with the primary target. This approach also decorrelates critical parameters to each other and to other process variations.

[0090] This approach does not require additional wafer space needed for setting up extra auxiliary targets. However, this approach requires wafer measurements to be performed at multiple process steps.

[0091] Using measurement data acquired during multiple process steps to train and utilize an SRM measurement model is analogous to the approach previously described herein for a single target. However, training the SRM measurement model additionally requires measuring the primary target during at least two different process steps.

[0092] Similarly, using the SRM measurement model requires measurement data from the primary target at the various process steps. However, reference measurement data for training only needs to be acquired from the primary target at the most recent process step, since only the specific parameter values ​​of the target at this step are of interest.

[0093] As described herein, reference measurements obtained using other technologies are required to train the SRM model. CD-SEM is an example of a measurement technique known for its high measurement uncertainty.

[0094] Specific examples involving LELE and SAOP are described herein as non-restrictive examples. In general, the methods and systems described herein can be used to improve the measurement of parameters of interest generated by any multiple structuring technique (e.g., self-aligned double, triple, quadruple, octuple structuring, double litho-double etching (LELE) structuring, etc.).

[0095] Fig. Figure 8 shows a System 400 for measuring the properties of a specimen according to the exemplary methods presented herein. As in Fig. As shown in Figure 8, the system 400 can be used to perform spectroscopic ellipsometric measurements of one or more structures of an specimen 401. For this purpose, the system 400 can include a spectroscopic ellipsometer equipped with an illuminator 402 and a spectrometer 404. The illuminator 402 of the system 400 is configured to generate illumination of a selected wavelength range (e.g., 150–2000 nm) and direct it onto the structure located on the surface of the specimen 401. The spectrometer 404, in turn, is configured to receive illumination reflected from the surface of the specimen 401. It should also be noted that the light emitted from the illuminator 402 is polarized by the use of a polarization generator 407 to produce a polarized illumination beam 406.The radiation reflected from the structure arranged on specimen 401 is directed through a polarization state analyzer 409 and to the spectrometer 404. The radiation received by the spectrometer 404 in the detection beam 408 is analyzed with respect to its polarization state, which enables a spectral analysis of the radiation transmitted through the analyzer by the spectrometer. These spectra 411 are transmitted to the computer system 430 for analysis of the structure.

[0096] As in Fig. As shown in Figure 8, System 400 incorporates a single measurement technology (e.g., SE). However, System 400 can generally incorporate any number of different measurement technologies. As a non-limiting example, System 400 can be configured as a spectroscopic ellipsometer (including Mueller matrix ellipsometry), a spectroscopic reflectometer, a spectroscopic scatterometer, an overlay scatterometer, an angle-resolved beam profile reflectometer, a polarization-resolved beam profile reflectometer, a beam profile reflectometer, a beam profile ellipsometer, any single- or multiple-wavelength ellipsometer, or any combination thereof. Furthermore, measurement data acquired by different measurement technologies and analyzed according to the methods described herein can generally be acquired by multiple machines rather than by a single machine integrating multiple technologies.

[0097] In a further embodiment, the system 400 can include one or more computer systems 430, which are used to perform measurements based on measured spectra and process information as described herein. The one or more computer systems 430 can be in communication with the spectrometer 404. In one aspect, the one or more computer systems 430 are configured to receive measurement data 411 associated with measurements of the structure of the specimen 401. The one or more computer systems 430 can also be in communication with one or more process machines 420, which are configured to perform a preceding process step used to produce the measured specimen 401.In one aspect, the one or more computer systems 430 are trained to receive process information 421 that is associated with any of the preceding process steps, with measurements of the structure of the specimen 401 at any of the preceding process steps, or with a combination thereof.

[0098] The various steps described in the present disclosure can be carried out by a single computer system 430 or, alternatively, by a multiple computer system 430. Furthermore, different subsystems of the system 400, such as the spectroscopic ellipsometer 404, can include a computer system capable of performing at least some of the steps described herein. Therefore, the foregoing description should not be construed as limiting the present invention, but merely as providing an explanation. Furthermore, the one or more computer systems 430 can be configured to perform any other step(s) of any embodiment of the method described herein.

[0099] Additionally, the computer system 430 can communicate with the spectrometer 404 in any known manner. For example, one or more computer systems 430 can be connected to computer systems associated with the spectrometer 404. In another example, the spectrometer 404 can be controlled directly by a single computer system connected to computer system 430.

[0100] The computer system 430 of the metrology system 400 can be configured to receive and / or acquire data or information from the system's subsystems (e.g., spectrometer 404 and the like) or from one or more process machines 420 via a transmission medium, which may include wired and / or wireless components. In this way, the transmission medium can serve as a data link between the computer system 430 and other systems or subsystems of the system 400.

[0101] The computer system 430 of the metrology system 400 can be configured to receive and / or acquire data or information (e.g., measurement results, model inputs, model results, etc.) from other systems via a transmission medium, which may include wired and / or wireless components. In this way, the transmission medium can serve as a data link between the computer system 430 and other systems (e.g., memory in the metrology system 400, external memory, process machines 420, a source of reference measurements, or other external systems). For example, the computer system 430 can be configured to receive measurement data from a storage medium (i.e., memory 432 or external memory) via a data link. For example, spectral results obtained using the spectrometer 404 can be stored in a permanent or semi-permanent storage device (e.g., a microcontroller, ...The spectral results can be stored in the computer system 430 (internal memory or external memory). In this respect, the spectral results can be imported from internal or external memory. Furthermore, the computer system 430 can transmit data to other systems via a transmission medium. For example, a measurement model or a structural parameter value 440 determined by the computer system 430 can be transmitted and stored in external memory. In this respect, measurement results can be exported to another system.

[0102] The Computer System 430 may, but is not limited to, include a personal computer system, a mainframe system, a workstation, an image processor, a parallel processor, or any other known device. In general, the term "computer system" may be broadly defined as encompassing any device that has one or more processors which execute instructions from a storage medium.

[0103] Program instructions 434, which implement procedures such as those described herein, can be transmitted via a transmission medium, such as a wire, a cable, or a wireless transmission link. For example, as in Fig. Figure 8 shows program instructions 434 stored in memory 432 being transferred to processor 431 via bus 433. Program instructions 434 are stored in a computer-readable medium (e.g., memory 432). Examples of computer-readable media include read-only memory, random access memory, a magnetic or optical disk, or a magnetic tape.

[0104] In some embodiments, the illumination light and the light detected by the illuminated measuring point include several different wavelengths. In some embodiments, the light from the illuminated measuring point is detected at several different detection angles. By detecting light at multiple wavelengths and detection angles, the measurement sensitivity to pitch drift and variations in critical dimensions (e.g., CD) is improved. In some embodiments, the light from the illuminated measuring point is detected at several different azimuth angles. These out-of-plane measurements can also improve the measurement sensitivity to pitch drift and variations in critical dimensions. In some embodiments, the acquisition of optical measurement data for a specific set of system settings, e.g.,Spectroscopic or angle-resolved system, one or more azimuth angles, one or more wavelengths, and any combination thereof, optimized.

[0105] Fig. Figure 9 shows a procedure 500, which is suitable for implementation by a metrology system, such as the one in Fig. The metrology system 400 of the present invention, as shown in Figure 8, is suitable. In one aspect, data processing blocks of the method 500 can be executed by a pre-programmed algorithm that is executed by one or more processors of the computer system 430, or by any other general-purpose computer system. The particular structural aspects of the metrology system 400 do not constitute limitations and should be interpreted as merely illustrative.

[0106] In block 501, a wafer is received by a metrology system (e.g., metrology system 400). The wafer contains one or more metrology targets, which were produced through a multitude of process steps in a manufacturing process sequence executed by a multitude of process machines.

[0107] In block 502, an initial amount of process information is received in a computer system of the metrology system from a first process machine, which is used to carry out the first of the many process steps on the wafer.

[0108] In block 503, a second set of process information is received in a computer system of the metrology system from the first process machine or another process machine that is used to carry out a second process step of the multitude of process steps on the wafer.

[0109] In block 504, a quantity of illumination light is supplied to one or more metrology targets by a lighting subsystem of the metrology system.

[0110] In block 505, a quantity of light from one or more metrology targets is detected in response to illumination by a detector subsystem of the metrology system.

[0111] In block 506, the detector subsystem generates measurement signals in response to the amount of detected light.

[0112] In block 507, a value of a parameter of interest of one or more metrology targets is estimated based on the measurement signals and the first and second sets of process information.

[0113] In block 508, the value of a correctable parameter associated with at least one of the process steps is estimated based on the measured property of the metrology target and the first and second sets of process information. Additionally, the value of the correctable parameter is transmitted to a process machine used to perform the at least one process step.

[0114] In some examples, using measurement data associated with multiple targets for model building, training, and measurement eliminates or significantly reduces the impact of sublayers on the measurement result. In one example, measurement signals from two targets are subtracted to eliminate or significantly reduce the effect of sublayers in each measurement result. Using measurement data associated with multiple targets increases the sample and process information embedded in the model. In particular, using training data that includes measurements of several different targets at one or more measurement points enables more accurate measurements.

[0115] In one example, a measurement model is generated from spectral measurements of a DOE wafer for both isolated and dense targets. This model is then trained based on the spectral measurement data and known structural parameter values. The resulting trained models are then used to calculate structural parameter values ​​for both isolated and dense targets on sample wafers. In this way, there is a separate trained model for each parameter, which calculates the parameter value from the measured spectra (or extracted features) associated with both isolated and dense targets.

[0116] In another aspect, measurement data derived from measurements performed using a combination of several different measurement techniques are used for model building, training, and measurement. Utilizing measurement data associated with multiple different techniques increases the sample and process information embedded in the model and enables more accurate measurements. Measurement data can be derived from measurements performed using any combination of different techniques. In this way, different measurement points can be measured using multiple techniques, increasing the measurement information available for characterizing semiconductor structures.

[0117] Within the scope of this patent document, any measurement technique or combination of two or more measurement techniques can generally be considered. Examples of measurement techniques include, but are not limited to, spectroscopic ellipsometry, including Mueller matrix ellipsometry, spectroscopic reflectometry, spectroscopic scatterometry, scatterometry overlay, beam profile reflectometry, both angle-resolved and polarization-resolved, beam profile ellipsometry, ellipsometry with one or more discrete wavelengths, transmission small-angle X-ray scatterometer (TSAXS), small-angle X-ray scattering (SAXS), small-angle grazing-incidence X-ray scattering (GISAXS), wide-angle X-ray scattering (WAXS), X-ray reflectivity (XRR), X-ray diffraction (XRD), grazing-incidence X-ray diffraction (GIXRD), high-resolution X-ray diffraction (HRXRD), X-ray photoelectron spectroscopy (XPS), and X-ray fluorescence (XRF).Grazing-incidence X-ray fluorescence (GIXRF), low-energy electron scatterometry (LEXES), X-ray tomography, and X-ray ellipsometry. In general, any metrology technique applicable to the characterization of semiconductor structures can be considered, including image-based metrology techniques.

[0118] Additional sensor options include electrical sensors, such as non-contact capacitance / voltage or current / voltage sensors, which preload the device and detect the resulting preload with an optical sensor (or vice versa), or supported optical techniques, such as XRD, XRF, XPS, LEXES, SAXS, and pump-probe techniques. In one embodiment, a two-dimensional beam profile reflectometer (pupil imaging) can be used to acquire both angle-resolved and / or multispectral data in a small measurement spot. Alternatively, a UV Linnik interferometer can be used as a spectral Mueller matrix pupil imaging device.

[0119] In some examples, the modeling, training, and measurement procedures described herein will be implemented as an element of the critical dimension optical metrology system SpectraShape®, available from KLA-Tencor Corporation, Milpitas, California, USA. In this way, the model is generated and ready for use immediately after the DOE wafer spectra have been acquired by the system.

[0120] In some other examples, the model building and training procedures described herein are implemented offline, for example, by a computer system implementing the AcuShape® software available from KLA-Tencor Corporation, Milpitas, California, USA. The resulting trained model can be included as an element in an AcuShape® library accessible to a metrology system performing measurements.

[0121] In another example, the methods and systems described herein can be applied to overlay metrology. Grid measurements are particularly relevant for overlay measurement. The goal of overlay metrology is to determine shifts between different lithographic exposure steps. Performing overlay metrology on electronic components is difficult due to the small size of the component structures and the generally small value of the overlay.

[0122] For example, the pitch of typical write-line overlay metrology structures varies between 200 nanometers and 2000 nanometers. However, the pitch of device overlay metrology structures is usually 100 nanometers or less. Furthermore, in a nominal production environment, the device overlay is only a small fraction of the periodicity of the device structure. In contrast, surrogate metrology structures used in scatterometric overlay are often shifted by larger values, e.g., a quarter of the pitch, to increase signal sensitivity to the overlay.

[0123] Under these conditions, overlay metrology is performed using sensor architectures that have sufficient sensitivity to overlays with low displacement and low pitch. The methods and systems described herein can be used to obtain a measurement signal sensitive to overlays based on device structures, proxy structures, or both.

[0124] After acquisition, the measured signals are analyzed to determine the overlay error based on changes in the measured signals. In a further step, the spectral or angle-resolved data are analyzed using PCA, and an overlay model is trained to determine the overlay based on the principal components detected in the measurement signal. In one example, the overlay model is a neural network model. In this sense, the overlay model is not a parametric model and is therefore not susceptible to errors introduced by inaccurate model assumptions.

[0125] In some embodiments, the training of the overlay metrology model is based on measurements of custom metrology structures, which are nominally identical to the device structures but exhibit larger displacements. This can help overcome the sensitivity problem. These displacements can be introduced by fixed design displacements introduced between structures in the two layers to be measured during reticle design. The displacements can also be introduced by displacements during lithography exposure. The overlay error can be extracted more efficiently from the compressed signal (e.g., PCA signal) by using multiple shifted targets (e.g., pitch / 4 and pitch / 4), and the effect of the sublayer can also be reduced.

[0126] In general, the methods and systems shown herein for performing semiconductor metrology can be applied directly to actual device structures or to custom metrology targets (e.g., proxy structures) located in a die or within a write line.

[0127] In yet another aspect, the measurement techniques described herein can be used to provide active feedback to a process machine (e.g., lithography machine, etching machine, deposition machine, etc.). For example, values ​​of the structural parameters determined using the methods described herein can be communicated to a lithography machine to adjust the lithography system to achieve a desired output. Similarly, etching parameters (e.g., etching time, diffusivity, etc.) or deposition parameters (e.g., time, concentration, etc.) can be incorporated into a measurement model to provide active feedback to etching machines or deposition machines, respectively.

[0128] In general, the systems and methods described herein can be used as part of a dedicated metrology machine or alternatively as part of a process machine (e.g., a lithography machine, etching machine, etc.).

[0129] As described herein, the term "critical dimension" encompasses any critical dimension of a structure (e.g., lower critical dimension, middle critical dimension, upper critical dimension, slope angle, grid height, etc.), a critical dimension between any two or more structures (e.g., distance between two structures), and a displacement between two or more structures (e.g., overlay displacement between overlapping grid structures, etc.). Structures can include three-dimensional structures, patterned structures, overlay structures, etc.

[0130] As described herein, the terms “critical dimension application” or “critical dimension measurement application” include any measurement of the critical dimension.

[0131] As described herein, the term "metrology system" encompasses any system used, at least in part, to characterize a specimen in any respect, including measurement applications such as critical dimension metrology, overlay metrology, focus / dose metrology, and composition metrology. However, such technical terms do not limit the breadth of the term "metrology system" as described herein. Additionally, the Metrology System 400 can be configured to measure structured and / or unstructured wafers.The metrology system can be configured as an LED inspection machine, an edge inspection machine, a backside inspection machine, a macro inspection machine, or as a multi-mode inspection machine (which involves data from one or more platforms simultaneously), and as any other metrology or inspection machine that benefits from the calibration of system parameters based on data for the critical dimension.

[0132] This document describes various embodiments of a semiconductor processing system (e.g., an inspection system or a lithography system) that can be used to process a specimen. The term "specimen" is used herein to refer to a wafer, a reticle, or any other sample that can be processed (e.g., structured or inspected for defects) by known means.

[0133] As used herein, the term "wafer" generally refers to substrates formed from a semiconductor material or a non-semiconductor material. Examples include, but are not limited to, single-crystal silicon, gallium arsenide, and indium phosphide. Such substrates are commonly encountered and / or processed in semiconductor manufacturing facilities. In some cases, a wafer may consist solely of the substrate (i.e., a bare wafer). Alternatively, a wafer may contain one or more layers of different material formed on a substrate. One or more layers formed on a wafer may be "structured" or "unstructured." For example, a wafer may contain a variety of dies exhibiting repeatable structural features.

[0134] A "reticule" can be a reticule at any stage of a reticule fabrication process, or a complete reticule that may or may not be released for use in a semiconductor fabrication facility. A reticule, or "mask," is generally defined as an essentially transparent substrate with essentially opaque regions formed on it in a pattern. The substrate may, for example, be a glass material such as amorphous SiO2. A reticule can be positioned over a photoresist-coated wafer during an exposure step of a lithography process, allowing the pattern on the reticule to be transferred to the photoresist.

[0135] One or more of the layers formed on a wafer can be structured or unstructured. For example, a wafer can contain a multitude of dies, each exhibiting repeatable structural features. The formation and processing of such material layers can ultimately lead to complete devices. Many different types of devices can be formed on a wafer, and the term wafer, as used herein, is intended to encompass any wafer on which any known type of device is fabricated.

[0136] In one or more exemplary embodiments, the functions described herein may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functions may be stored as one or more instructions or code on or transmitted via a computer-readable medium. Computer-readable media include both computer storage media and communication media, including any medium that enables the transmission of a computer program from one location to another. A storage medium may be any available medium accessible by a general-purpose or specialized computer.Non-restrictive examples of such computer-readable media may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code means in the form of instructions or data structures, and that can be accessed by a general-purpose or specialized computer, or by a general-purpose or specialized processor. Likewise, any connection is correctly referred to as computer-readable medium.If the software is transmitted, for example, from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pairs, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pairs, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of medium. Disk and disc, as used herein, include compact disc (CD), laserdisc, optical disc, digital versatile disc (DVD), floppy disk, and Blu-ray disc, where "disks" generally reproduce data magnetically, while "discs" reproduce data optically using lasers. Combinations of the above are also to be included in the meaning of computer-readable media.

Claims

[1] Metrology system (400) comprising: a lighting subsystem (402) that provides a quantity of illumination light (406) for one or more metrology targets arranged on a wafer (401) that has previously been processed through a plurality of process steps of a manufacturing process sequence; a detector subsystem (404) that detects a quantity of light (408) from the one or more metrology targets in response to the quantity of illumination light (406) and generates a quantity of measurement signals (411) in response to the quantity of detected light (408); and a computer system (430), trained to Receiving an initial amount of process information (421) from a first process machine (420) which is used to perform a first process step of the plurality of process steps on the wafer (401); Receiving a second set of process information (421) from the first process machine (420) or another process machine that is used to perform a second process step of the plurality of process steps on the wafer (401); and Estimating the value of a structural parameter of interest (440) of one or more metrology targets based on the set of measurement signals (411) and the first and second sets of process information (421). [2] Metrology system (400) according to claim 1, wherein the computer system (430) is further configured to: Estimating the value of a correctable parameter associated with at least one of the process steps, based on the measured property of the metrology target and the first and second sets of process information (421); and Communicating the value of the correctable parameter to at least one process machine (420) which is used to perform at least one process step. [3] Metrology system (400) according to claim 1, wherein the first set of process information (421) includes one of the following: a process control parameter, a setting parameter for a process machine (420), an environmental parameter for a process, a set of process data acquired by sensors in the first process machine (420), and a set of metrology data acquired by sensors in the first process machine (420). [4] Metrology system (400) according to claim 1, wherein the first set of process information (421) includes a parameter value for the lithography focus, a parameter value for the lithography dose or a combination thereof. [5] Metrology system (400) according to claim 1, wherein the estimation of the value of the parameter of interest (440) of one or more metrology targets involves a physical model or a trained input-output measurement model. [6] Metrology system (400) according to claim 5, wherein the computer system (430) is further configured to: Training the input-output measurement model with simulated measurement data and simulated process information, actual measurement data and actual process information associated with a Design-Of-Experiments (DOE) wafer, or a combination thereof. [7] Metrology system (400) according to claim 1, wherein the computer system (430) is further configured to: Reducing one dimension of the first and second set of process information (421), the set of measurement signals (411) or a combination thereof. [8] Metrology system (400) according to claim 1, wherein the detector (404) is configured to detect light (408) from the target structure at multiple wavelengths, multiple detection angles, or a combination of multiple wavelengths and multiple detection angles. [9] Metrology system (400), comprising: a lighting subsystem (402) that provides a quantity of illumination light (406) for one or more metrology targets arranged on a wafer (401) that has previously been processed through a plurality of process steps of a manufacturing process sequence; a detector subsystem (404) that detects a quantity of light (408) from the one or more metrology targets in response to the quantity of illumination light (406) and generates a quantity of measurement signals (411) in response to the quantity of detected light (408); and a non-volatile, computer-readable medium, comprising: Code to cause a computer system (430) to receive a first set of process information (421) from a first process machine (420) which is used to execute a first process step of the plurality of process steps on the wafer (401); Code to cause the computer system (430) to receive a second set of process information (421) from the first process machine (420) or another process machine that is used to execute a second process step of the plurality of process steps on the wafer (401); and Code to cause the computer system (430) to estimate a value of a structural parameter of interest (440) of one or more metrology targets based on the set of measurement signals (411) and the first and second sets of process information (421). [10] Metrology system (400) according to claim 9, wherein the non-volatile computer-readable medium further comprises: Code to cause the computer system (430) to estimate a value of a correctable parameter associated with at least one of the process steps, based on the measured property of the metrology target and the first and second sets of process information (421); and Code to cause the computer system (430) to communicate the value of the correctable parameter to at least one process machine (420) which is used to perform the at least one process step. [11] Procedure encompassing: Receiving a wafer (401) containing one or more metrology targets produced by a multitude of process steps of a manufacturing process flow carried out by a multitude of process machines (420); Receiving an initial amount of process information (421) from a first process machine (420) which is used to perform a first process step of the plurality of process steps on the wafer (401); Receiving a second set of process information (421) from the first process machine (420) or another process machine that is used to perform a second process step of the plurality of process steps on the wafer (401); Providing a quantity of illumination light (406) for one or more metrology targets; Detecting a quantity of light (408) from the one or more metrology targets in response to illumination (406); Generating a quantity of measurement signals (411) in response to the quantity of detected light (408); Estimating the value of a parameter of interest (440) of one or more metrology targets based on the set of measurement signals (411) and the first and second sets of process information (421). [12] The method of claim 11, further comprising: Estimating the value of a correctable parameter associated with at least one of the process steps, based on the measured property of the metrology target and the first and second sets of process information (421); and Communicating the value of the correctable parameter to at least one process machine (420) which is used to perform at least one process step. [13] Method according to claim 11, wherein the plurality of process machines involved in the manufacturing process (420) includes at least one lithography machine and at least one etching machine. [14] Method according to claim 11, wherein the first and / or second set of process information (421) includes a process parameter value. [15] Method according to claim 14, wherein the process parameter value is a lithography focus parameter value, a lithography dose parameter value or a combination thereof. [16] Method according to claim 11, wherein the first and / or second set of process information (421) includes a property of the metrology target that is measured by a metrology system in the first process machine (420). [17] Method according to claim 11, wherein the estimation of the value of the parameter of interest (440) of one or more metrology targets involves a physical model or a trained input-output measurement model. [18] The method of claim 17, further comprising: Training the input-output measurement model with simulated measurement data and simulated process information, actual measurement data and actual process information associated with a Design-Of-Experiments (DOE) wafer, or a combination thereof. [19] The method of claim 11, further comprising: Reducing one dimension of the first and second set of process information (421), the set of measurement signals (411) or a combination thereof. [20] Method according to claim 11, wherein the one or more metrology targets comprise a nominal metrology target and at least one auxiliary metrology target, wherein both the nominal metrology target and the at least one auxiliary metrology target are each characterized by at least one parameter of interest which was generated by at least two structuring steps of a multiple structuring process.

Images