An in-process strategy for defect-free CPL masks
12/01/2005
The masks required for chromeless phase lithography can be successfully inspected for transparent defects and repaired by existing technology. An in-process inspection and repair strategy that promises to ensure shipment of defect-free reticles is proposed.
By Yasutaka Morikawa
Chromeless phase lithography (CPL) technology employs paired phase-edges to produce darkness along with a combination of high numerical-aperture (NA) projection and off-axis illumination, which improves printed image contrast and depth-of-focus (DOF). This scheme was introduced as a potential resolution enhancement technology (RET) for application to 90nm- and 65nm-node logic gate layers. It is theoretically the same as the 100% transmission halftone phase-shifting mask (HT-PSM), so the pattern layout can be the same as HT-PSM.
One of the technical challenges of the CPL mask is that it requires tritone pattern features, which means all large dark areas must be covered with opaque chrome shielding as in high-transmission HT-PSM. Another issue is the selection of mask structure. There are two CPL mask structure candidates: island or mesa (in which the regions surrounding the phase-shifter patterns are etched) and trench (in which the phase-shifter pattern regions themselves are etched). We chose the island type because it is relatively insensitive to alignment error in the logic gate maskmaking process.
 Figure 1. Process flow for fabricating an island-type CPL mask. |
Figure 1 shows the process flow for the island-type CPL mask. It is almost the same as the current HT-PSM process, except that the shifter is the quartz substrate rather than an additional film. We have been developing a quartz etch process to get better performance for alternating-aperture PSMs (AAPSM) and also found it useful for the CPL mask process. So the advantage of this technology from the viewpoint of mask manufacturing is that it is possible to prepare the data by an extension of the HT-PSM design technology and to apply almost the same process technology as HT-PSM and AAPSM. Of course, there are other technical challenges, for example, accurate overlay and CD control on the second exposure, which clears the chrome from the CPL mesas.
While a mask fabrication process is essential, inspection and repair technologies are also very important. The challenge most discussed in AAPSM inspection has been detecting quartz defects that show small optical contrast in the standard transmitted light inspection. In CPL masks, the critical features are also constructed with quartz steps. Therefore, it is easy to imagine that these main features won’t have enough contrast for the inspection systems.
We took review images of 90nm-node 1:2 line and space (L/S) CPL patterns with a UV inspection system in the transmission mode. These confirmed that the contrast was low, similar to a 75% transmission HT-PSM. To confirm CPL mask inspection capability and quartz defect detectability, we fabricated and inspected a programmed defect CPL test mask on several leading-edge inspection systems. An AIMS tool was employed to evaluate defect printability and repair capability. Based upon our data, we propose an inspection and repair plan to guarantee defect-free CPL masks using current leading-edge tools.
Experimental conditions
Test mask design and fabrication. We have designed and fabricated a CPL programmed defect test mask, which has L/S patterns for 90nm and 65nm technology nodes. Only island-type structures were used, without assist features. The pitch for the 90nm lines was 270nm while that for the 65nm node was 195nm, making the spaces twice the width of the lines. As shown in Fig. 2, there were nine different defect types: four types of extensions or bumps, and five types of intrusions or trenches. Both quartz and chrome-over-quartz defects appeared, with nominal dimensions varying from 10nm to 300nm in steps of 10nm at 4× mask dimension.
The test mask was fabricated on a standard 6025 quartz substrate covered by AR8 opaque chrome and negative chemically amplified resist. The first electron-beam write was done at 50kV, followed by dry etching for both chrome and quartz (which was etched to 180° depth for ArF wavelength). The second write step was also done using a 50kV e-beam pattern generator, but this time by exposing positive chemically amplified resist and removing the exposed chrome by dry etching.
Evaluation tools. Various tools and technologies were used in this experiment. Die-to-die (D/D) inspection was performed using KLA-Tencor’s SLF27 (algorithm: P150 XPAddt10) and 525 DUV systems (algorithms: P125 UCFddt35 + phase contrast), as well as Applied Materials’ Aera193 aerial image tool (NA = 0.7, 0.8/0.52 annular illumination). AIMS review was done with Zeiss’ MSM-193 (λ = 193nm, NA = 0.7, 0.8/0.52 annular illumination), and the repair system used was the Rave nm1300.
Experimental results
Imaging capability. The KT525 DUV system has a phase contrast option that enhances the visibility of phase defects and includes a defocus inspection function, which can help identify the signs of such defects. Images were acquired of 65nm- and 90nm-node patterns under four different conditions: SLF27 at UV with 150nm pixel size; KT525 at DUV with 125nm pixels (normal condition); KT525 with phase contrast and zero focus offset; and KT525 with phase contrast and 450nm focus offset.
For both 90nm and 65nm mask designs, shorter wavelength, phase contrast, and nonzero defocus helped increase the image contrast of CPL patterns and were expected to provide better detection performance for quartz defects.
Detectability and printability of quartz defects. Figure 3 shows a summary of all nine categories of quartz defect detectability and printability for the 90nm and 65nm nodes. Each defect category has three columns that correspond to each of the three types of inspection tools tested. The heights of the colored bars indicate 100% detection, and the values in boxes indicate the detection percentage when it is <100%. Inspection runs on the KLA tools were done five times, whereas the Aera193 was run 10 times. Red horizontal lines indicate the boundary for >10% CD change as confirmed by AIMS analysis, and blue dotted lines indicate where defects would cause >5% CD change.
These results imply that at the 90nm node, all quartz defects that cause 10% CD error on AIMS were detected by the KT525 with phase contrast, but the tool may be too sensitive to edge-type defects. Most of the quartz defects that cause 10% CD error were detected by the Aera193 with constant detection capability. It looks as though the Aera193 detectability and printability are well balanced. At the 65nm node, most of the quartz defects that cause 10% CD error were detected by the KT525 with phase contrast and by the Aera193. Quartz defects identified as printable by the AIMS tool at 65nm appeared more difficult to detect than defects at 90nm, but that may have been because 0.7NA was too low a setting on the AIMS tool for the 65nm experiment.
 Figure 4. SEM pictures and AIMS images for a) 90nm-node and b) 65nm-node defects before and after nanomachining repair. |
Repair. We tried to repair “large” quartz defects for 90nm and 65nm nodes using the Rave nm1300 nanomachining repair tool. Figure 4 shows before and after repair SEM and AIMS images. The SEM images show that the quartz defects repaired were very large, larger than the main CPL pattern in width and length. Nevertheless, the nanomachining repair shows good printing performance of the repaired region on the AIMS analysis: <±5% different from reference structure CDs through ±0.2µm DOF.
Inspection and repair plan
 Figure 5. In-process CPL mask structure a) at first inspection and detectable defect types: b) chrome and quartz extensions (orange) and c) missing material (blue). |
We have examined quartz defect inspection capability on CPL masks, but only for the main chromeless pattern and D/D inspection. A real CPL mask has a full tritone design with opaque chrome shielding as well as phase edges. Die-to-database (D/B) inspection may be required. Based on this experiment and our experience with similar designs, we can suggest an inspection and repair plan to guarantee defect-free CPL masks.
Two in-process inspections will be needed. The first inspection would be performed after the first write and etch process. The second inspection would be performed after the second process. All repairs would be performed after the second inspection because AIMS analysis doesn’t work properly on the structure produced by the first process.
At the first inspection, several types of defects are expected (Fig. 5). The extension- or bump-type defects with chrome present are easy to detect with normal D/D and D/B inspection systems because good image contrast is expected on this defect type. A quartz extension defect can also be detected by current inspection systems that have the capability to detect AAPSM quartz defects, such as the KLA-Tencor SLF/STARlight, even on a single-die mask layout.
If the mask has D/D-capable layout, other applications are possible. Clear or missing material defects should be easy to detect at this stage by normal D/D and D/B inspection systems because the absence of the expected chrome top layer will give good image contrast on these defects.
 Figure 6. In-process CPL mask structure a) at second inspection and detectable defect types: b) chrome residue and quartz bump (orange) and c) missing chrome (blue). |
At the second inspection, other types of defects are expected to appear (Fig. 6). For extension- or bump-type defects on a chromeless structure, the defect should already have been detected at the first inspection. If the layout allows D/D inspection, the KT525 with phase contrast or the Aera193 is effective for this defect type. Chrome residue on a nominally chromeless region or missing chrome on a nominally opaque region may be detected by transmitted light pattern inspection for D/D layout. D/B inspection of chromeless pattern and database rendering will require future work.
Extension-type or bump-type defects detected at either inspection can be repaired using the Rave nanomachining tool after the second inspection. Missing chrome defects will be repaired by focused ion-beam carbon deposition. If we detect missing quartz-type defects on a shifter region where the chrome has been etched (most likely seen on the first inspection), we have no reasonable repair method at this moment, but there is only a very small chance of creating this defect type with current blanks and mask processes.
Conclusion
We have fabricated a CPL quartz-defect test mask that has 90nm- and 65nm-node designs, examined defect detectability on several kinds of inspection systems, and examined quartz defect printability by using the AIMS tool. From these results, we have seen that the KT525 DUV with phase contrast or the Applied Materials Aera193 are effective for detecting quartz defects that cause 10% CD error as evaluated by AIMS. Large quartz extension defects on CPL patterns were successfully repaired by the Rave nanomachining tool. The AIMS tool confirmed good printability and CD performance through-focus at the repaired line. These results have enabled us to suggest an inspection and repair plan to guarantee defect-free CPL masks.
Acknowledgments
The author would like to thank Chikashi Ito and Eddie Nagaoka, KLA-Tencor Japan, and Hidemichi Imai, DNP, for their technical support and advice regarding KT525’s effective condition settings, as well as Applied Materials engineers for their technical support with Aera193 data collection and analysis.
References
- J. Fung Chen, et al., “Binary Halftone Chromeless PSM Technology for λ/4 Optical Lithography,” Proc. SPIE 4346, pp. 515-533, 2001.
- D.J. Van Den Broeke, et al., “Complex 2D Pattern Lithography at λ/4 Resolution Using Chromeless Phase Lithography (CPL),” Proc. SPIE 4691, pp. 196-214, 2002.
- J. Fung Chen, et al., “Mask Design Optimization for 70-nm Technology Node Using Chromeless Phase Lithography (CPL) Based on 100% Transmission Phase-shifting Mask,” Proc. SPIE 4754, pp. 361-372, 2002.
Yasutaka Morikawa is a senior expert in the Electronic Components Laboratory at Dai Nippon Printing Co. Ltd., 2-2-1 Fukuoka, Kamifukuoka-shi, Saitama 356-8507, Japan; ph 81/492-78-1683, e-mail [email protected].