Progress for characterization and Advanced reticle repair
07/01/2000
John Morgan, FEI Company Inc., Peabody, Massachusetts*
Troy Morrison, Surface/Interface Inc., Sunnyvale, California
*Additional authors are listed in the Acknowledgments.
overview
The combined attributes of focused ion beam reticle repair and the quantitative application of atomic force microscopy, the latter implemented as SNP, provide a complete solution for advanced reticle repair and characterization. Ion beam repair offers superior accuracy and precision in both material deposition and removal repairs—removal without significant damage to underlying or adjacent quartz. SNP can provide quantitative measurement of 3D structures, including those associated with alternating phase shifters etched into quartz as well as embedded shifters. These capabilities are crucial to the performance of the advanced reticles required for subwavelength lithography.
 Figure 1. Clear defect in a binary chrome reticle a) before repair and b) after the carbon deposition repair. |
In an era when the cost of a set of advanced reticles for a semiconductor manufacturing process can approach $1 million, the value of reticle repair capability is obvious. In addition to smaller features, reticles now include complex three-dimensional (3D) reticle enhancement structures associated with phase shifting that must be characterized during reticle manufacture and accommodated by any repair process. The variants to 3D-reticle structures include alternating phase shifters etched into the quartz substrate and embedded shifters, the latter a layer of partially transmissive material, typically MoSiON or CrOx. It is the addition of these 3D structures that drives up the cost of advanced reticles and makes their repair both valuable and difficult.
 Figure 2. MoSiON-embedded shifter reticle with a) missing contact hole and b) contact hole replicated by FIB. |
Focused ion beam (FIB) systems offer the capability to precisely and accurately deposit or remove material on reticles to repair both clear and opaque defects. A relatively new 3D-metrology technology—Stylus NanoProfilometry (SNP)—offers the capability to characterize FIB repairs thoroughly and nondestructively. The role of 3D metrology is critical for characterizing the desired features as well as the defects encountered in the reticle manufacturing process.
3D characterization of shifter films also is important because the film thickness determines the degree of phase shift and transmission. The integration of these two technologies provides a complete reticle repair solution.
FIB reticle repair
 Figure 3. Tip shape removal a) provides true feature profiles from b) raw data. |
FIB-based reticle repair systems are favored over laser-based methods when critical edge placement is required, features are small, or geometries are too complex to copy with sufficient fidelity. FIB systems use a finely focused ion beam to both remove and deposit material on the quartz substrate. FIB systems can make repairs in reticle features as small as 0.3µm, with edge alignment accuracy better than 0.05µm.
To repair clear defects, FIB systems deposit an opaque film onto the exposed quartz substrate by introducing a small amount of precursor hydrocarbon gas in the vicinity of the repair. When the beam scans over the designated repair area, it decomposes the gas and deposits an opaque film consisting of carbon from the gas and gallium from the ion beam. The resulting film is durable and unaffected by standard reticle cleaning processes (Fig. 1). Moreover, varying the deposition time permits control of the film thickness.
 Figure 4. FIB repair of a clear defect in a chrome binary reticle with accompanying SNP-derived profile. |
To repair an opaque defect, the FIB system must remove the opaque film from the reticle without damaging the surrounding features or underlying substrate. The ion beam alone can physically sputter away the defect. To eliminate or substantially control ancillary damage, a reactive gas is used to enhance the etching effect of the opaque defect, while simultaneously reducing the etching effect on the quartz; this process is known as gas-assisted etching (GAE). As with the deposition technique, GAE occurs only in the area scanned by the beam, permitting precise control of the removal process.
FIB repairs to embedded shifters are analogous to repairs to the opaque film of binary reticles (Fig. 2).
 Figure 5. SNP measurement of 13nm "riverbed." |
Beam blanking and specially designed scan methodologies add further control to the FIB material removal process, enhancing it for reticle repair. Adaptive beam blanking automatically constructs a bit map of the defect from the FIB image and blanks the ion beam when it scans over exposed quartz. This technique minimizes damage to the quartz while freeing the system operator from having to carefully outline the defect. Careful design of the ion beam scanning parameters addresses the problem of excessive quartz damage at the edge of the defect area. This damage — known as "riverbedding" —results when the ion beam sputters material from the edge of the defect more rapidly than from the interior. The edge material disappears sooner, and if that material is not compensated for, the beam sputters the quartz underlying the edges while finishing the removal of the interior film.
Reticle characterization
Clearly, the capability to characterize phase-shift reticles in 3D is critical both in their original manufacture and in their repair. Knowledge of linewidth, although important, lacks the complete description of the process or defect. We need to know sidewall angle and shape, feature height or depth, and film thickness. Conven-tional metrology methods fall short for various reasons:
- Profilometers lack sufficient resolution and the capability to closely follow high aspect features.
- CD-SEMs, the industry standard for 2D linewidth measurements, cannot provide accurate measurements in the third dimension.
- Cross sectioning, whether mechanical or FIB, for SEM metrology is not viable because of its destructive nature. In addition, the thickness of reticles makes them difficult to cleave.
- Traditional atomic force microscopes (AFMs) are designed for imaging, but they lack the quantitative capabilities and process integration needed for reticle characterization.
Thus, an alternative approach is required. A quantitative application of AFM can provide the required information.
Quantitative application of AFM
 Figure 6. Image of a PSM and diagram of SNP measurements. |
SNP is an implementation of AFM technology designed specifically to provide quantitative metrology in a platform tailored to reticles and other semiconductor industry applications. By focusing on the issues that affect quantitative metrology, the SNP has been able to overcome the historical limitations of using a probe technology for CD and 3D metrology.
The concept behind SNP is elegantly simple [1]: a finely tipped probe is used to periodically touch the sample surface, creating a 3D-measurement array that accurately maps surface shape. The first issue addressed must be precise control of the contact between the sample surface and the probe. A special technique—Force Controlled Contact (FCC)—permits the system to apply a force sufficient to overcome extraneous forces, such as sample charging, while still maintaining tip and sample integrity (typically a few hundred nano-Newtons) to probe the sample without destroying it. A feedback circuit that monitors capacitance between the beam and its mount is used to measure the amount of force exerted by the tip on the sample surface.
 Figure 7. SNP contour map of a simulated quartz bump a) before and b) after repair. |
The second important attribute of SNP that contributes to accurate and precise metrology is its scan methodology. The probe assembly is mounted on a piezoelectric scanner, which brings the tip into controlled contact with the surface using purely vertical (z) motion. Upon sensing the predetermined set force, the scanner retracts the tip vertically, and only then translates the tip in the x or y direction to the next measurement point. This straight-down approach—Angle-Controlled Contact (ACC) — is critical, as the removal or deconvolution of the probe from the raw scan is required to achieve the true feature shape.
Another critical area of concern for any quantitative probe-based methodology is tip characterization. The line scan or trace created by an AFM or profilometer is a convolution of the surface shape and the tip shape used to collect the trace. If the tip shape is accurately known, it can be removed from the data set, yielding a true representation of the surface shape. SNP automatically measures tip shape via on-board characterizers and removes its shape from the raw profile measurement. The profiles in Fig. 3 demonstrate the importance of tip shape removal. The figure shows a set of closely spaced chrome lines on a reticle that was intentionally over-etched in a wet etch process. The profile without tip shape removal (Fig. 3b) accurately shows the reduction in line height, but only with tip shape removed (Fig. 3a) is the reduction in linewidth also apparent.
Combining FIB and SNP
The following examples provide clear evidence of the combined capabilities of FIB and SNP for advanced reticle repair:
- Characterization of clear defect repair. Figure 4 shows the FIB repair of a clear defect in a chrome binary reticle. The accompanying plot compares profiles of the repair and the original line (labeled "control"). In this example, repair thickness, edge position, sidewall angle, and edge profiles are all readily apparent.
- Characterization of opaque defect repair. This example shows the FIB repair of an opaque defect in a chrome binary reticle. The edge biasing can be clearly seen and measured in the SNP plot. The riverbed effect can be observed and measured at any point along the repair edges. The accompanying plot compares profiles of repaired and control sections of the line. Edge position and profile, and sidewall angle are readily apparent. Fig. 5 shows the ability of SNP to measure the riverbed. This repair shows a riverbed of 15nm. Also, a roughness of about a few nm can be measured in the repair site. Because of its unique set of capabilities, the SNP can act as a single platform to image, quantify the riverbed, and sample roughness as well measure the CD, z-height, and sidewall angle.
- Characterization of a phase-shift reticle. Figure 6 is an SNP image of a phase-shift reticle. Note the detailed 3D information available in the accompanying profile and the measurements needed to fully qualify the performance of a typical APSM reticle. Particularly critical and difficult to measure by conventional means such as optical- and SEM-based metrology is the depth of the quartz and sidewall angles of the structure at varying heights.
- Quartz bump repair. Quartz bumps are surface structures that occur on a quartz substrate. Although they can be detected by optical inspection systems, they are invisible to optical metrology systems and difficult or nearly impossible to image adequately by SEM. The repair of these defects has been identified as a roadblock to implementing FIB-based repairs for alternating phase-shift reticles. In addition to imaging difficulties, end-pointing quartz bumps on a quartz substrate are a serious problem. Because there is no compositional change between the bump and the substrate, it is difficult to know when to stop milling. A proposed strategy to address this problem is to characterize the shape and size of the bump with SNP and to use this information to program a scan strategy and beam dose profile to remove the bump.
Figure 7 shows an example of an SNP contour map of a simulated quartz bump before and after repair.
Conclusion
The use of FIB and SNP together provides a complete solution for advanced reticle repair and characterization. FIB offers superior accuracy and precision in both material deposition and removal repairs. Advanced FIB techniques permit the safe removal of opaque defects without significant damage to the underlying or adjacent quartz substrate. SNP provides quantitative measurement capabilities designed to integrate directly into existing manufacturing operations. Furthermore, SNP can measure 3D structures.
Acknowledgments
Additional authors are David Ferranti and Diane Stewart of FEI Company. Stylus NanoProfilometry, Force Controlled Contact, and Angle Controlled Contact are registered trademarks of Surface/Interface Inc.
Reference
- J.B. Bindell, et al., "Stylus NanoProfilometry: A new approach to CD metrology," Solid State Technology, pp. 45-53, June 1999.
John Morgan received his BS in chemical engineering from Clarkson University. He is product manager of Mask Repair Sys-tems at FEI Co., Micrion Products Div., One Corporation Way, Pea-body, MA 01960; ph 978/538-6700, e-mail jmorgan@ ma.feico.com.
Troy B. Morrison received his BS and MS in materials science and engineering from MIT. He has more than 8 years of experience in semiconductor process and inspection technology, and is an applications manager at Surface/Interface Inc., 260 Santa Ana Court, Sun-nyvale, CA; ph 408/732-7111, e-mail [email protected]
Mask validation: the task gets tougher
Direct validation of repairs prior to a reticle's release into production becomes more critical with each reduction in feature size; this is not a simple task. Validation looks at whether the repair itself is correct and whether the wafer will print correctly. Three possible techniques we have at our disposal are metrology on the repair (as discussed in this article); analysis to emulate a stepper tool with the Zeiss Aerial Image Measuring System (the AIMS lithography simulation system used to characterize the binary, PSM, OPC mask defects, mask repair, and mask CD errors, and to understand their printability); and an actual printability test on a wafer. Clearly, if we can correlate metrology of repairs with AIMS analysis and printability tests, we will be able to provide "go/no-go" results in the least possible time and at the lowest possible cost by using only metrology. The measurements made by Stylus NanoProfilometry will hopefully help get us to that point.