A Study of Closed Loop Application: WLCD CDC for 32nm and beyond reticles

Transcription

1 A Study of Closed Loop Application: WLCD CDC for nm and beyond reticles Arosha Goonesekera a, Ute Buttgereit b, Thomas Thaler b, Erez Graitzer c a Carl Zeiss NTS LLC., SMS Division, Corporation Way, 96 Peabody, MA b Carl Zeiss SMS GmbH, Carl-Zeiss-Promenade, 775 Jena, Germany c Carl Zeiss SMS Ltd., Maale Camon, Karmiel, 6 Israel ABSTRACT Optical lithography stays at 9nm with a numerical aperture of.5 for several more years before moving to EUV lithography. Utilization of 9nm lithography for nm and beyond forces the mask maker to produce complex mask designs and tighter lithography specifications which in turn make process control more important than ever. High yield with regards to chip production requires accurate process control. Critical Dimension Uniformity (CDU) is one of the key parameters necessary to assure good performance and reliable functionality of any integrated circuit. There are different contributors which impact the total wafer CDU: mask CD uniformity, resist process, scanner and lens fingerprint, wafer topography, etc. In this study the newly developed wafer level CD metrology tool WLCD of Carl Zeiss SMS is utilized for CDU measurements in conjunction with the CDC tool from Carl Zeiss SMS which provides CD uniformity correction. The WLCD measures CD based on proven aerial imaging technology. The CDC utilizes an ultrafast femto-second laser to write intra-volume shading elements (Shade-In Elements ) inside the bulk material of the mask. By adjusting the density of the shading elements, the light transmission through the mask is locally changed in a manner that improves wafer CDU when the corrected mask is printed. The objective of this study is to evaluate the usage of these two tools in a closed loop process to optimize CDU of the mask before leaving the mask shop and to ensure improved intra-field CDU at wafer level. Main focus of the study is to investigate the correlation of applied attenuation by CDC and the resulting CD change, the impact of CDC process on CD linearity behavior and the correlation of WLCD data and wafer data. Logic and SRAM cells with features having designed line CD s at wafer level, ranging from 7nm to nm have been used for the study. The investigation provides evidence that the applied attenuation by CDC shows a linear correlation to CD change at wafer level measured with WLCD. Additionally, WLCD data shows that the CDC application does not impact the CD linearity for the tested feature range. The WLCD measurement data in turn show an excellent correlation to wafer print CD data indicating cost effective use case of closed loop WLCD/CDC application. Key words: CD, CDU, CDC, WLCD, reticle, metrology, CDU correction, linearity, aerial image. INTRODUCTION Semiconductor industry is pushing the 9nm lithography to the next technology nodes even down to nm before moving to EUV Lithography. Various resolution enhancement techniques have been implemented to drive resolution to the theoretical limits where k factor comes close to.5. This means that CD errors on mask are getting highly amplified on wafer due to a tremendously increased Mask Error Enhancement Factor (MEEF). In order to maintain a high yield in production, process control becomes an essential element. One key parameter to ensure a high and reliable functionality for any integrated circuit is the critical dimension uniformity (CDU). In this study our main focus is to investigate the correlation of the applied attenuation by CDC and the resulting CD change, the impact of CDC process on CD linearity behavior and the correlation of WLCD data and wafer data. The mask CD uniformity has been measured by WLCD which measures the CD based on proven aerial image technology. Dedicated attenuation changes have been applied to selected areas of the mask using CDC tool. These selected areas contain logic and SRAM cells with designed line CDs from 7nm to nm at wafer level. The post CD uniformity has also been measured with WLCD. The data show evidence that the applied attenuation by CDC shows a linear correlation to CD change at wafer level and also the CDC application does not impact the CD linearity for the tested feature range. In addition, the WLCD measurements show an excellent correlation to wafer print CD data. Photomask Technology, edited by Wilhelm Maurer, Frank E. Abboud, Proc. of SPIE Vol. 866, 8669 SPIE CCC code: X//$8 doi:.7/ Proc. of SPIE Vol

2 . EXPERIMENTAL SET-UP. CD Control - CDC The CDC process utilizes shading elements inside the mask bulk to attenuate the light during the wafer exposure. The CDC process creates small pixels that consist of QZ with a different morphology which create a slightly different refractive index (delta n). This delta n causes a small amount of scattering outside of the scanner objective pupil and hence causes attenuation. The CDC process basic set-up is described in Figure. Figure : CDC Process: At the focal point of the laser beam a pixel is created. Quartz density is altered, and so is the local index of refraction. Each pixel acts as a scattering element. In order to improve intra-field CD uniformity, shading elements of specific attenuation level or pixel density are applied to each specific area in the mask. Figure shows the relevant shading elements: Figure : Applying shading elements to the mask reduces light transmission locally and effectively reduces the local dose. This causes all features to print at a CD closer to target. The utilization of CDC process was thoroughly investigated using wafer CDU data as input,, 5 and in production. In this work we used the tool to apply variable or constant attenuation to the selected regions of the test mask to investigate the correlation of applied attenuation to corresponding CD change and also to investigate how well the corresponding CD change preserves the CD linearity.. Mask Metrology WLCD Aerial Image CD Measurement Zeiss Wafer Level CD metrology system WLCD is based on proven aerial imaging technology. It measures the CD on the reticle in the wafer level plane as it is relevant to printing, (see Fig. ). It captures optical proximity effects and optical MEEF effects induced by the scanner illumination. The use of the WLCD significantly simplifies the CD measurement especially for complex mask designs and complex D features. Proc. of SPIE Vol

3 Figure : WLCD measures the CD on mask as it is relevant for printing, simplifying the CD measurement especially for complex mask design The WLCD is equipped with new Zeiss 9nm imaging and illumination optics. The LITO -grade optics has extremely low aberrations and comes close to the quality of the scanner optics. The variable NA allows measurements up to a scanner equivalent NA of.. A new 9nm laser is used for ultra fast CD measurements of several hundred CD s per hour. The tool is equipped with user defined aperture planes for off-axis illumination in order to illuminate the mask under the same conditions as a scanner. Additionally, the newly developed Free-Form Illumination devices can be used to adopt the illumination not only in geometrical shape but also in intensity distribution. The Free-Form Illumination devices fully support SMO technology. Furthermore, different illumination polarizations (tangential, x, y) are available. Vector effects by high NA imaging can be taken into account by using the Zeiss proprietary scanner mode. For CD measurement the user can define several regions of interest within the field of view, which allows CD measurements on arbitrary features, e.g. logic features, in-cell features.. Test mask description The mask used for the work is a special design containing identical cells shown in Fig. (a) that consist of both logic and SRAM features. The cell structure is shown in Fig. (b), which consists of 9 sub cells. We used two sub-cell rows one each of logic (L) and SRAM (S) consist of 7 sub-cells. S+ S+ S+ S S- S S- S+ S+ S+ S S- S S- S+ S+ S+ S S- S S- L+ L+ L+ L L- L L- L+ L+ L+ L L- L L- L+ L+ L+ L L- L L- (a) (b) Figure : (a) Schematic overview of the mask layout and (b) cell structure consisting of logic (L) and SRAM (S) futures Proc. of SPIE Vol

4 The objective of the study was two fold: () investigation of CD uniformity and correlation to wafer data and () investigation of linearity and impact of CDC application on CD linearity behavior. CD uniformity study: For the investigation of CD uniformity and wafer data correlation study, we chose a SRAM feature as the target across the plate consisting of 8 sites. The sample grid is shown in Fig. 5. A special illumination has been used at immersion lithography conditions. The WLCD data were captured at 8 sites and wafer print data were also captured at identical locations. Same illumination conditions were used for wafer print performed by an immersion scanner. After pre CDC uniformity measurement a dedicated attenuation scheme was applied with the use of CDC application at 6 areas of the mask as shown in Fig. 5, which produces a change in CD at the wafer level. In four regions an attenuation gradient has been applied: -% and % in the direction noted by the arrows in Fig. 5. In the remaining two regions, a constant % attenuation has been applied. After this process, WLCD and wafer print CD data were captured. These pre and post CDC data from WLCD and wafer print CD data have been used for analysis. Figure 5: The 8 point sample plan located within variable attenuation applied areas with CDC application CD linearity study: For the linearity study we chose 6 cells within the plate as marked in black within Fig. 6. A sub-cell row [Fig (b)] was chosen to capture the linearity data, each consisting of 7 logic features and 7 SRAM features. The data were captured before and after the CDC application with variable attenuation levels as described in the previous paragraph. However, no attenuation has been applied to two sites that have been used as reference for this study. In this case too, the same special illumination scheme has been used as for the CD uniformity investigation. The fourteen pre and post CDC data points per each of the 6 sites taken with WLCD tool and wafer print CD measurement tool have been used for analysis. Figure 6: The 6 sites used for the CD linearity study. Four sites are in variable attenuated areas and sites are in no attenuation area and can be used for reference Proc. of SPIE Vol

5 . RESULTS AND DISCUSSION. CD Uniformity: The objective of CD uniformity investigation is to see how well the CD measured with the WLCD tool correlates to the CD measured at the wafer after printing, both before and after the CDC process. Additionally, we investigated how much CD change is caused by % attenuation of the CDC application (CDC Ratio). The targeted feature for the CD uniformity study is a SRAM Line/Space structure of a dedicated design CD. We used the design CD at wafer level as target CD to determine the WLCD threshold using a reference feature and found a threshold of.5. This threshold value has been applied to all 8 points to determine the CD map across the mask before and after CDC application. Figure 7 shows the CDU map from wafer print and WLCD measurements across the plate before the CDC application. The wafer CD data have been measured using optical scatterometry. Pre MTT (nm) <= -. Pre MTT (nm) <= -. 9 <= -.57 <= -. <= -.7 <= -.9 <=. 9 <= -.57 <= -. <= -.7 <= -.9 <=. 8 <=.57 8 <=.57 PosY 7 6 <=. >. MaskPosY 7 6 <=. > PosX MaskPosX Figure 7(a): CDU distribution for SRAM feature determined from wafer print measurements before CDC application Figure 7(b): CDU distribution for SRAM feature determined from WLCD measurements on mask before CDC application As can be seen the wafer print and WLCD CDU signatures are highly correlated and show a radial CD distribution. The CDU statistics show that the measured CDU ranges obtained by wafer print and WLCD are very similar. The difference between wafer print and WLCD data is only.6 nm in terms of CDU range. The CDU -sigma values only differ by. nm. Post MTT (nm) <= -.5 Post MTT (nm) <= <= -.7 <= -. <= -. <=. <=. 9 <= -.7 <= -. <= -. <=. <=. 8 <=.7 8 <=.7 PosY 7 6 <=. >. MaskPosY 7 6 <=. > PosX MaskPosX Figure 8(a): CDU distribution for SRAM feature determined from wafer print measurements after CDC application Figure 8(b): CDU distribution for SRAM feature determined from WLCD measurements on mask after CDC application Proc. of SPIE Vol

6 Post CDU data for wafer print and WLCD are shown in Figure 8. The effect of applied attenuation on CD at the 6 areas is clearly indicated in both contour plots. However, wafer print data show less pronounced effect than WLCD data. A comparison of CD change (delta CD: post - pre) data from WLCD measurement and wafer print data are shown in Fig. 9, where the radius of the circles is proportional to the CD change at each location. WLCD data clearly indicate that the correlation of CD change corresponds to applied attenuation. Although wafer data indicate a correlation to applied attenuation, the plot shows significant measurement noise in the wafer print data. This can be clearly seen by the CD variation within the untreated areas of the test mask. The measurement noise in the wafer print data is also confirmed in Fig. (a) where delta CD (post pre) from wafer print data is plotted against applied attenuation. The CD variation in the untreated area at % attenuation indicates the measurement noise which is in a range of.7 nm at wafer. Scanner conditions, wafer process and OCD tool noise are possible contributing factors for the observed noise in wafer print CD data. The noise of WLCD data is in the range of. nm at wafer level visualized in Fig. (b) at the % attenuation Mask Position Y Mask Position Y Mask Position X Mask Position X Figure 9(a): Delta CD distribution (post pre) for SRAM feature determined from wafer print measurements after CDC application Figure 9(b): Delta CD distribution (post pre) for SRAM feature determined from WLCD measurements on mask after CDC application Figure shows the correlation of applied attenuation and delta CD for wafer print data and WLCD data before and after CDC application. The applied attenuation is well correlated to change of CD in WLCD data demonstrated by an excellent R² of.96. Delta CD vs attenuation plot has been used to determines the CDC ratio Delta CD (nm).. Delta CD (nm) y =.857x +.79 R = y =.696x +.88 R = Attenuation (%) Attenuation (%) Figure (a): Delta CD vs attenuation plot determines the CDC ratio for the wafer print data. The CDC ratio is.9 nm per % attenuation. Figure (b): Delta CD vs attenuation plot determines the CDC ratio for the WLCD data. The CDC ratio is.6 nm per % attenuation. Proc. of SPIE Vol

7 The CDC ratio is.6 nm per % attenuation for WLCD data and.9 nm for wafer print data. The discrepancy in CDC ratio for WLCD data and Wafer print data can be explained by resist effects that are not captured by the WLCD. Additionally, the poor signal to noise ratio which yields to a low R² of.5 only, impacts the quality of the wafer slope value as well.. CDC impact on Linearity The objective of the linearity investigation is to see the impact of applied CDC attenuation on the linearity behavior of the CD change. Also we used our study to investigate how well linearity behavior is preserved after the CDC application. The targeted features for this work are seven SRAM Line/Space structures and seven Logic Line/Space structures as shown in table. SRAM (nm) S- S S- S S+ S+ S+ LOGIC (nm) L- L L- L L+ L+ L+ Table : Feature sizes of logic and SRAM features used for the linearity study The mid designed CD from each category has been used as the target CD to determine the WLCD threshold, which is S for SRAM features and L for Logic features. Corresponding threshold values were found to be.5 and.8 respectively and have been kept constant for the linearity evaluation. The seven CD values for the six sites corresponding to Logic and SRAM features are shown in Fig. before CDC treatment. Fig. shows the Mean to Target deviation (MTT) variation for each feature CD and measurement site. In this case the target value corresponds to the mid CD values (L and S), averaged over the 6 sites. It can be seen that the CD variation between the different sites is very low Logic Site Site Site Site Site5 Site6 - - SRAM Site Site Site Site Site5 Site6.5 L- L L- L L+ L+ L+ - S- S S- S S+ S+ S+ Fig (a) WLCD vs Designed CD data for logic features Fig (b) WLCD vs Designed CD data for SRAM futures In Figure the corresponding averaged CD s over the six sites measured with WLCD are plotted against the average CD on the wafer print measured with optical CD measurement technique. The plots show a very good correlation between wafer print CD values and WLCD values for both logic and SRAM features with R nearly one along with similar slopes of about.85. Delta Wafer Print CD (nm) Logic y =.86x +.58 R = Delta WLCD CD (nm) Fig (a): WLCD (average) vs Wafer print CD (average) data for logic features (pre CDC) Delta Wafer Print CD (nm) - - SRAM y =.856x -.85 R = Delta WLCD CD (nm) Fig (b): WLCD (average) vs Wafer print CD (average) data for SRAM features (pre CDC) Proc. of SPIE Vol

8 The data for CD change for the respective sites as a result of CDC application are shown in Fig., where delta CD measured with WLCD is plotted against the designed CD. As expected, site shows the largest CD change for all feature sizes corresponding to nearly % attenuation by CDC application. Additionally, it can be seen that the CD change per feature size is almost the same over the considered feature size range. Sites, and show reasonably correlated delta CD values corresponding to applied attenuations. Site 5 and 6 were not treated with CDC application and the values shown for delta CD are representing WLCD tool noise level. In the untreated areas we measure delta CD values of.5nm (at wafer) only. WLCD Delta Pre-post (nm) L- L L- L L+ L+ L+ Logic Site Site Site Site Site5 Site6 WLCD Delta Pre-post (nm) S- S S- S S+ S+ S+ SRAM Site Site Site Site Site 5 Site 6 -. Design CD -. Design CD Fig (a): WLCD delta CD (post pre) vs design CD data for logic features for different attenuation (% to %) Fig (b): WLCD delta CD (post pre) vs design CD data for SRAM features for different attenuation (% to %) Fig. (a), (b), (c) and (d) show the CD linearity data before and after the CDC application for both Logic and SRAM features. The data strongly demonstrate how well the CD linearity has been preserved after the CDC application. - - y =.686x -.8 R =.9987 Logic- site ~.5% y =.75x R =.9978 L- L L- L L+ L+ L y =.66x -.87 R =.996 SRAM- site ~.5% y =.85x R =.9965 S- S S- S S+ S+ S+ Fig (a): Linearity behavior of the WLCD measured CD vs design CD before and after CDC application for site where applied attenuation is nearly.5% - - y =.69x -.67 R =.9965 Logic -Site ~.% y =.69x R =.9956 L- L L- L L+ L+ L y =.988x -.65 R =.9985 SRAM - Site (~.%) y =.98x R =.9995 S- S S- S S+ S+ S+ Fig (b): Linearity behavior of the WLCD measured CD vs design CD before and after CDC application for site where applied attenuation is nearly %. Proc. of SPIE Vol

9 - - y =.697x R =.998 Logic- site ~.5% y =.699x -.58 R =.998 L- L L- L L+ L+ L y =.58x -.6 R =.997 SRAM- site ~.5% y =.979x R =.9985 S- S S- S S+ S+ S+ Fig (c): Linearity behavior of the WLCD measured CD vs design CD before and after CDC application for site where applied attenuation is nearly.5%. - - y =.699x R =.9976 Logic- site ~.% y =.68x -.59 R =.9978 L- L L- L L+ L+ L y =.x -.85 R =.996 SRAM- site ~.% y =.x -.87 R =.9978 S- S S- S S+ S+ S+ Fig (d): Linearity behavior of the WLCD measured CD vs design CD before and after CDC application for site where applied attenuation is nearly %. Additionally, WLCD data show a very good linear correlation to design CD with R values of nearly, which are also preserved after CDC application. The linearity behavior has not been affected by CDC application which is clearly demonstrated by the constant slopes of the plots above as summarized in Table (). The data clearly demonstrate that the CDC process does not change the CD linearity. In the evaluated CD range the CDC ratio (CDCR) is independent of the absolute feature CD. This proves that CDC application is equivalent to dose change, a very important fact for using CDC process in production. Attenuation Logic (Slope Pre) Logic (Slope Post) SRAM (Slope Pre) SRAM (Slope Post) Site ~.5% Site ~ % Site ~.5% Site ~ % Table : The slopes of the linear fits for all sites before and after CDC application remain constant. SUMMARY AND CONCLUSIONS In this work we have focused on CD uniformity measurement using WLCD and the correlation of WLCD data to wafer print data. Additionally, we investigated the impact of CDC application on the CD linearity behavior. Dedicated attenuation changes have been applied to selected areas of the mask using CDC tool and CD change before and after the CDC application have been measured with WLCD tool on the mask and with optical CD tool on wafer print. These selected areas contain logic and SRAM cells with designed target CD ranging from 7 nm to nm. The following results have been demonstrated: - WLCD measures the CDU uniformity correctly and shows an excellent correlation to wafer print data - The CDC treatment does not change the CD linearity behavior for the considered CD range demonstrated by constant slopes for logic and SRAM cells Proc. of SPIE Vol

10 REFERENCES [] Sven Martin, Holger Seitz, Wolfgang Degel, Ute Buttgereit, Thomas Scherübl, WLCD: A new System for Wafer Level CD Metrology on Photomasks, Proc. SPIE Vol. 77, pp. 77T-77T-9 (9). [] F.Dufaye, S.Gough, F.Sundermann, V.Farysa, H.Miyashita, L.Sartelli, F. Perissinotti, U.Buttgereit, S.Perlitz, R.Birkner, Mask phase and transmission variation effects on wafer critical dimensions for nodes 65nm and 5nm, Proc. SPIE Vol. 755, 755M () [] Guy Ben-Zvi, Eitan Zait, Vladimir Krugliakov, Vladimir Dmitriev, Gidi Gottlieb, Sergey Oshemkov, " The Effect of Intra-field CD Uniformity Control (CDC) on Mask Birefringence" EMLC 7 [] Pforr et al. Performance comparison of techniques for intra-field CD control improvement, Proc. SPIE Vol. 67, 67 (7) [5] Sz-Huei Wang, Yu-Wan Chen, Chung Ming Kuo,Erez Graitzer, Guy Ben-Zvi, Avi Cohen," Expanding The Lithographygraphy Process Window (PW) With CDC Technology", Proc. SPIE Vol. 788, 788I (9) [6] Ute Buttgereit, Robert Birkner, Mark Joyner, Erez Graitzer, Avi Cohen, Hiroyuki Miyashita, Benedetta Triulzi, Alejandro Fasciszewski Zeballos, and Carmelo Romeo, CD uniformity correction on 5-nm technology nonvolatile memory, Proc. SPIE Vol. 768, 768K () [7] Bertrand Le Gratiet, Frank Sundermann, Jean Massin, Marianne Decaux, Nicolas Thivolle, Fabrice Baron, Alain Ostrovsky, Cedric Monget, Jean Damien Chapon, Yoann Blancquaert, Karen Dabertrand, Lionel Thevenon, Benedicte Bry, Nicolas Cluet, Bertrand Borot, Raphael Bingert, Thierry Devoivre, Pascal Gourard, Laurène Babaud, Ute Buttgereit, Robert Birkner, Mark Joyner, Erez Graitzer, and Avi Cohen; Improved CD control for 5- nm CMOS logic patterning: anticipation for 8 nm, Proc. SPIE Vol. 768, 768A () Proc. of SPIE Vol