Bare wafer metrology challenges in microlithography at 45 nm node and beyond Chunsheng Huang, Ph.D. KLA-Tencor Corporation ADE Division 3470 Universal Way Tucson, AZ 85706 Abstract The shrinking depth of focus (100-150 nm) of high numerical aperture immersion microlithography optics dictates a tight wafer flatness budget. Wafer flatness nanotopography (NT), and edge roll off (ERO) are critical parts of the equation in immersion microlithographic technology at the 45 nm node and beyond. Wafer features at the nanometer level could result not only in focus variation of the litho process, or thin film thickness variation in CMP process, but also in structural defects of the devices. Therefore, the metrology to measure nanometer level features and to control the quality of wafer geometry is a key to the success of IC production at the 45 nm node and beyond. Key words: Bare wafer metrology, interferometer, wafer nanotopography, wafer edge roll off. 1. Introduction Bare wafer surface topography variation is a significant part of focus budget of microlithography. As the IC line-width shrinks, the depth of focus (DOF) is also reduced. The wafer surface flatness variation can take as much as 40% to 50% of total DOF of the microlithographic project optical scanner with a dynamic tilt adjustment on each exposure according to International Technology Roadmap for Semiconductors (ITRS).¹ Thus, as the wafer surface quality becomes increasingly important, the metrology to control the surface quality is increasingly challenged. This paper will present the technological challenges with a focus on nanometer level precision, accuracy, and tool matching. It will also analyze bare wafer flatness, shape, ERO, and NT impact on the immersion lithographic focus budget. The paper will describe and characterize the current state-of-the-art bare wafer metrology tool that has subnanometer precision. Data from multiple tools will be compared to demonstrate the tool-to-tool matching that will overcome emerging microlithographic technology challenges. 2. Wafer flatness impact on Lithographic Focus Budget Immersion technology increases the effective NA of the optical lithography and thus increases the printing solution and makes it possible using ArF laser (193 nm) to resolve a 45 nm line-width. However, because NA increases, the DOF is reduced. The half pitch of optical resolution or line-width L can be expressed² as L= K1 λ0 (1) NA NA = n ⋅ sin(θ ) (2) Quantum Optics, Optical Data Storage, and Advanced Microlithography, edited by Guangcan Guo, Songhao Liu, Guofan Jin, Kees A. Schouhamer Immink, Keiji Shono, Chris A. Mack, Jinfeng Kang, Jun-en Yao, Proc. of SPIE Vol. 6827, 682723, (2007) · 0277-786X/07/$18 · doi: 10.1117/12.759738 Proc. of SPIE Vol. 6827 682723-1 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx Where K 1 is a processing related factor and NA is numerical aperture of the optical system and wavelength in the air and n is the refractive index of immersion media and θ λ0 ⋅ n is the exposing is the marginal angle of the lithographic lens at the wafer plane. The depth of focus (DOF) of the scale equation approximation DOF= K 2 λ0 2 is (3) NA 2 Where K 2 is a processing related factor. As NA increases, L is inversely reduced with NA, and DOF is quadratically inversely reduced. The DOF reduces faster than the line-width does as NA increases in the air. However, if the effective NA increases by refractive index of the immersion media, the L and DOF have the same reduction rate as the refractive index increases or it is proportional to 1 . Figure 1a. is the estimation of DOF calculated based on ITRS n roadmap. Using K1 range specified in ITRS roadmap and equation (2) to calculate NAs shown in Figure 1b. Then using equation (2) and refractive indices of immersion medium obtain the DOF. Of course DOF may vary with the tools and processing factor such as optical proximate correction (OPC), illumination schemes, and double pattern print. Production worthy immersion scanner technology currently uses water as an immersion media which has a refractive index of 1.43 at 193 nm wavelength. The effective NA of 1.2 to 1.35 has been reported. To achieve 45 nm line-width, K1 of 0.28 to 0.31 is needed. A K 2 of 0.7 will result in the DOF of 106 nm to 134 nm range. For high refractive index of the immersion media, the NA could be even higher, and DOF will be further reduced. The process focus window is reduced to 26 to 34 nm after taking consideration of wafer flatness, wafer stage focus accuracy and repeatability, wafer chuck flatness, and others factors, such as thermal stability. The focus accuracy of the state of the art of scanning wafer stage technology has been reported by Mulkens³ to be as small as 25 nm. Chuck 4 ,11 (a Peak-to-Valley within a 26mm flatness can be as small as 10 nm. According to ITRS, the wafer flatness of SFQR x 8mm wafer surface area, after removing the tilt) is 45 nm for 45 nm node design and 32 nm for 32 nm node design. Figure 2 shows a wafer flatness example of SFQR magnitude and distribution over a 300 nm wafer. This wafer has a site with maximum value of 54 nm. In the case, the wafer site (or die) may go out of the process window and cause the critical dimension (CD) out of specification. The wafer flatness can take as much as 32 to 45% DOF for 45 nm and 32 nm design node depending on the scanners and process factor. The primary focus budget allocation is to wafer flatness. To improve wafer flatness, the metrology of wafer flatness is the key technology to be addressed. Proc. of SPIE Vol. 6827 682723-2 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx DOF vs Wafer Flatness 350 300 DOF (nm) 250 200 DOF (Dry) DOF(wet) 150 ITRS Wafer Flatness 100 50 0 150 130 115 100 90 80 70 65 57 50 45 38 32 26 1999 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 ITRS Roadmap 1/2 pitch in nm Figure 1 a. ITRS Roadmap vs. DOF and Wafer Flatness. K2 of 0.7, NAs and refractive indices in Figure 1 b are used to estimate the DOF. K1, NA, and refrative index used for calculate for DOF 2 1.5 K1 NA 1 Refractive Index of medium 0.5 0 150 130 115 100 90 80 70 65 57 50 45 38 32 26 19992001200220032004200520062007200820092010201120122013 ITRS Roadmap 1/2 pitch in nm Figure 1 b. ITRS Roadmap vs. k1, NA, and refractive index of immersion medium that meet the roadmap Proc. of SPIE Vol. 6827 682723-3 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx Figure 2. A 300 mm SFQR map (peak-to-valley value after removal of tilt within the site. The site size is 26mm x 8mm.) There is a site in which SFQR is as large as 54 nm. 3. Edge Roll Off (ERO) effect on the edge of wafer focus. ERO is defined as occurring within a distance from the edge of 1mm to 5 mm, as shown in the Figure 3. If it is not controlled, the amount of topography variation within the edge roll off of a wafer can cause the scanner to go out of focus, thus reducing the area of a wafer that can be used. The ERO can be characterized with variety of parameters such 5 6 as radius of curvatures (ZDD), edge profile, or ESFQR (peak to valley value of a wafer edge radial section after removing tilt) as shown in Figure 4. The ESFQR shown in Figure 4 has 72 sections and has a large maximum value of 197 nm. This amount of edge roll off is out of DOF for 45 nm node design. The wafer manufacturer is clearly challenged to reduce ERO in the bare wafer process. For this surface area to be usable for 45 nm node, ERO should be smaller than 60 nm, or less than 60% of DOF. To accurately measure the edge of wafer as close as 1 mm to the edge is difficult, because it is close the edge of bevel and the edge roll off is very fast. Any sub-pixel data dropped off or added in at the very edge can be a significant measurement variation. The repeatability and accuracy of the metrology is challenged. Proc. of SPIE Vol. 6827 682723-4 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx ERO region Edge Bevel 5 4 2 3 Distance from edge (mm) 1 0 Figure 3. The edge roll off is defined as from 1mm to 5 mm. Figure 4. ESFQR map with 1mm edge exclusion and 5 degrees per section has a maximum value of 197 nm. 4. Wafer nanotopography (NT) impact on litho. Proc. of SPIE Vol. 6827 682723-5 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx 7 The wafer NT represents the nanometer level wafer flatness variation (bumps or trenches) of a bare wafer with the spatial periods of 0.2 to 20mm over a 2x2mm square or a circle, 2 mm in diameter. NT defects at the bare wafer can cause photoresist, thin film or metal layer thickness variations and result in defects of IC devices in the CMP process reported by D. Boning et al. 8 Because of photoresist thickness variations due to NT defects, it is inevitable to have CD 9 variations in the litho-etching process. NT defects has corrected with CD variations by A. Grenville et al at Intel . It is not clear what correlation the defects in nanometer level of a bare wafer with periods over 0.2mm and 20mm range have with immersion litho defects such as bubbles. It is clear that NT impacts not only CD but also thin films such as gate oxide (typical thickness <3 nm) or metal layers. ITRS roadmap 2006 added NT as a part of bare wafer metrics. It recommends that NT be smaller than 11 nm for 45 nm node and 8 nm for 32 nm node. NT requirements pose significant challenges for metrology equipment, to perform with very high precision and accuracy at a nanometer level. In particular for optical metrology, very high quality optical components, insensitivity to vibration and acoustic noise, and higher spatial resolution are a must. NT values are calculated as follows: a) application of high pass filter to the wafer thickness map; b) find PV of the filtered data within 2x2 square mm sliding over an entire wafer for TH2 metrics and 10x10 square mm for TH4; c) 7 calculate PV value distribution over percentage of the area. SEMI standard metrics uses 0.05% of the Full Qualified Area (FQA) (full wafer excluding 2 mm edge) as a wafer NT value. Figure 5 shows an example of a wafer NT map before and after filtering. Figure 6 shows a cumulative defective area versus NT value which indicates at each height, the percentage of FQA present. For this particular wafer, TH2 is 8 nm which meets the ITRS roadmap requirement for 32 nm node. But it may be questionable for TH4. An example of an NT defect of 20 nm is shown in Figure 7 4 um High Pass Spatial Filtering 20 nm -20 nm -4 um Raw Surface Height Map Nanotopography Map Figure 5. A wafer flatness map produces the NT map after applying a high pass spatial filter. Proc. of SPIE Vol. 6827 682723-6 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx 100 TH2(WS-NT) %Defective A 10 TH4(WS-NT) 1 TH2 8 nm 0.1 TH4 16 nm 0.01 0 10 TH2---ITRS Roadmap 45 nm node 20 30 40 50 60 Threshold Height(nm) Figure 6. Cumulative defect area vs height value. The steep one is TH2 and the other is TH4. Proc. of SPIE Vol. 6827 682723-7 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx A-' 2U.U Figure 7. 20 nm NT feature is shown in both 2-d and 3-d plots. 5. The State-of-the Art Metrology to meet the challenges KLA-Tencor’s newly developed WaferSight™ 2 product is designed and developed to meet the bare wafer metrology requirements for 45 nm node and 32 nm node. The state-of the art technology is based on a dual Fizeau interferometer design. Figure 9 shows an optical schematic of the wafer interferometer tool, where two Fizeau interferometer channels are combined to look at the two wafer sides. Light from a single tunable laser diode is fiber-coupled to each interferometer channel. In each channel, a polarization beam splitter (PBS) directs the beam through a quarter-wave plate (λ/4) to the collimator lens. The collimated beam propagates to the reference flat, whose last surface is the reference surface of the interferometer. The wafer is placed between the reference flats of the two interferometer channels, such that both surfaces can be measured simultaneously. The beams reflected at the wafer surface and at the reference surface propagate back to the collimator lens, and then on to the wave plate and beam splitter. The beam splitter acts as an optical valve, now transmitting the light to the relay objective, which images the wafer surface onto the camera, where the interferograms are detected. For the data acquisition, the wavelength of the laser is varied to create a linear phase shift in the interferograms. Several camera frames with phase-shifted fringe patterns are digitized and processed in the computer, where high-performance phase algorithms are employed to convert at each camera pixel the fringe-phase to surface height. Proc. of SPIE Vol. 6827 682723-8 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx Reference Flat λ/4 PBS Relay Camera Wafer Collimator Fibers Channel B Illuminator Illuminato r Computer Channel A Figure 9. Dual channel Fizau interferometer for measuring wafer geometry. Figure 10. Illustration of wafer flatness and shape derivation. According to the Semi-standard M14, wafer flatness f(x,y) is defined as the thickness variation across the wafer, see Fig. 10. The wafer shape s(x,y) is defined as the deviation from a best fitting plane of the median surface between the front and back surface of the wafer in a force-free state. The wafer flatness f ( x, y ) and shape s ( x, y ) can be calculated from the measurements as f ( x , y ) = d A ( x , y ) + d B ( x, y ) − c ( x , y ) s ( x, y ) ≈ (4) 1 (d A ( x, y ) − d B ( x, y )) − tilt 2 (5) Proc. of SPIE Vol. 6827 682723-9 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx Where d A ( x, y ) and d B ( x, y ) are single side distance from the wafer to cavity and c(x,y) is the distance (without wafer) between two reference flat shown in Figure 10. Tilt is the wafer tilt. The details of interferometric computation are described by Freischlad el al. 10 6. Tool performance and characterization 11 Wafer geometry metrics can be characterized by flatness metrics , edge metrics, NT, and shape metrics, industry standards as described in the table 1. The performance is characterized against those metrics. Flatness metrics Peak-to-valley (PV, or range) of the total global flatness. This is equivalent to the total thickness variation over the FQA. GBIR Peak-to-valley of the total site flatness. This is equivalent to the total thickness variation over SBIR the site. Peak-to-valley of the site flatness after removing a best fitting plane over the site. This is equivalent to the thickness variation over the site without any local wedge component. SFQR ERO metrics ESFQR ESBIR ZDD NT TH2 Peak-to-valley of the site flatness after removing a best fitting plane over the site. The site is defined as an angular sector sliced over 360 degrees. The radial length from the edge is from 1 mm to 5 mm. This is equivalent to the thickness variation over the site without any local wedge component. Peak-to-valley of the site flatness without removing a best fitting plane over the site. The site is defined as an angular sector sliced over 360 degrees. The radial length from the edge is from 1 mm to 5 mm. This is equivalent to the thickness variation over the site containing any local wedge component. This radius curvatures as a function of radius over 1mm to 5mm from the edge averaged over the sector which is radially sliced over 360 degrees. Peak-to-valley (PV, or range) over 2x2 mmxmm sliding window after passing the high pass filter over the thickness map. Peak-to-valley (PV, or range) over 10x10 mmxmm sliding window after passing the high pass filter over the thickness map. TH4 Shape metrics Warp Peak-to-valley of the total shape over the FQA. Peak-to-valley of the power term of the shape over the FQA. The power term is determined by a Bow best fit of a quadratic, or be a 4-point sag calculation. Table 1. Bare wafer SEMI standard metrics. 7. Tool repeatability and reproducibility It is important that the tool can repeat and reproduce the measurement results. If the tool has a large variability, the confidence of the measurement results is reduced. The tolerance to precision ratio of a wafer metrology needs to be at 11 lease larger than 3. SEMI M49 specifies the geometry metrology tool performance only to 65nm node. Table 2 summarizes recommended specifications and data examples of the state-of-the art technology performance for 45 nm and possibly 32 nm node in Figure 11 to 17. The data was taken within a 3 day gauge repeatability and reproducibility (GRR) test. There are a total 25 wafers in one cassette. Wafers were measured on WaferSight 2 with one repeat for each cycle and total of 5 cycles for each day. The precision is calculated for each site of each wafer of each day and then averaged over all sites and all wafers and all days. Proc. of SPIE Vol. 6827 682723-10 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx The precisions (1 σ) for SFQR and SBIR are 0.26 and 0.42 nm respectively, and plotted in Figure 11 and Figure 12. ERO precision (1 σ) for ESFQR is 0.32 nm shown in Figure 13. NT precisions for TH2 and TH4 are 0.04 and 0.11 nm, respectively, shown in Figure 14. 65 nm node (Semi M49) The state-of-the art technology Matching Reproducibility tolerance/bias (nm) (1 σ) nm Flatness metrics 4.2 6.3 1.5 2.3 ERO metrics N/A N/A NT metrics 0.3 0.5 1.2 1.8 SBIR SFQR ESFQR TH2 TH4 Reproducibility (1 σ) nm Matching tolerance/bias (nm) 0.42 0.26 0.4 0.2 0.32 0.38 0.04 0.11 0.04 0.078 Table 2. Summary for SEMI M49 specifications for 65 nm node and data examples of the state-of-the art technology performance from Figure 11 to 17. [Variability Chart for SFQR Value (nm) 60 10 40 30 20 10 Site Number Day 4 6 7 8 9 10 11 12 13 14 15 16 17 18 19 21 20 22 23 24 25 Source Slot 2.0 >1.6 1.2 B 0.8 0.4 0.0• ite Number 1 2 3 4 rHrrn I 5 6 1111111111111111111iii 7 8 9 10 11 12 13 14 11 16 17 18 19 20 21 22 23 24 21 Source Slot [Variance Components Component Source Slot Day Ilource Slot1 Site NuniberSource Slot,Day] Within Total Var Component % of Total 20406080 0.738214 0.8782 0 .000000 0.0 83. 2 64112 99.0 0. 061203 0.0776 84. 06 7609 100.0 Sqrt(Var Comp) 0. 8192 0 .0000 9. 1249 0. 2113 σ = 0.26nm 9. 168 8 Figure11. SFQR Precision plot. 1 sigma is 0.26 nm. Proc. of SPIE Vol. 6827 682723-11 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx p C P P σ = 0.42nm Figure 12 SBIR precision plot. 1 sigma precision is 0.42 nm. Variability Chart for ESFOR Value (nm) 110- I 130 110- 90- 70 10 ESFQR Angle (deg) t 4.0 >3.0 Day 10 11 12j1314 15 16 17j9 20 21 22j2324 28 Source Slot 02.0 S10 SFQR Angle (deg) >>>3Day 1 2 3 4 S 67 0 9 10 11 12 13 14 11 le 17 18 19 20 21 22 23 24 21 Source Slot Variance Components Component Source Slot Day[Source SlotS ESFQR Angle (deg)[Source Slot,Day] within Total VarComponent * of Total 20408080 174.82914 0.00000 343.81301 0.08808 118.74184 33.7 0.0 ee.3 0.0191 100.0 oqrt(Var Comp) 13. 22 2 0.000 10. 142 0.3 11 22. 778 σ = 0.32nm Figure 13 ESFQR precision (1 sigma) is 0.32 nm. Proc. of SPIE Vol. 6827 682723-12 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx THA2 Precision VurlubHItyChurt foFront_1HA4mmiu ng.)6 % (nm) 9.0 * 8.0 P8.0 THA4 Precision I Lvarab11' Chart for Front mA (10 mm Square Range) g 0.05% (nm) 18 I Ji I 7.0 1 17 3 18 11 14 I 13 12 oo 678 = 1 1 1 124 8 7 8 2.0 >0.8 0.6 >1.8 1.2 B 0.8 B 0.4 000.2 010.4 o o >__>__>;__>__a.a._-C_a_ >__;0__>__>;_>_;1__>__;0_> 0.0 878 Variance Components Component Var Component * of Total 20406080 Oqrt(Var Comp) Oourte hot 0.0942978 99.9 0.0928 within 0.0016041 0.0407 0.1383 Total 0.0909009 000.0 0.0936 Variance Components Component Var Component * of Total 20408080 Source Slot within Total 3.1224313 0.0118321 3.0342834 σ = 0.04nm 99.7 0.334e 100.0 Sqrt(Var Comp) 1.8788 0.1088 1.8800 σ = 0.11nm Figure 14. NT precisions (1 sigma) are 0.04 and 0.11 nm, THA2 and THA4 respectively. 8. Tool Matching It is very important to tool users to have tools that can match results, so that the measurement is not tool dependent. Tool matching first establishes a baseline by averaging the measurement results of multiple tools and then characterizes the difference of the measurement of the same wafers from the tool against the baseline. Tool matching is a measurement combination of tool accuracy plus tool variability. Figure 15 a) and b) show examples of site flatness SFQR and SBIR as measured by WaferSight 2. The vertical axis is the difference and horizontal axis is the mean value of two tools. The individual point is the corresponding site difference. Within the data, the worst site matching is 6 nm for SFQR and 10 nm for SBIR. Average bias of all sites is 0.2 nm for SFQR and 0.4 nm for SBIR. Figure 16 shows ERO metrics, ESFQR matching. The worst site matching is 8 nm for ESFQR and 20 nm for ESBIR. The mean bias is 0.39 nm for ESFQR and 0.69 nm for ESBOR. Figure 17 a) and b) show TH2 and TH4 of NT metrics matching results. The worst wafer matching is 0.18 nm for TH2 and 0.6 nm for TH4. The mean bias is 0.04 nm for TH2 and 0.078 nm for TH4. The results are summarized in Table 2. 9. Conclusions Wafer site flatness, ERO and NT impact on litho processing window. The site flatness reduces the focus budget for litho tool. Wafer ERO metrics affects the edge die yield. NT defects cause photoresist, thin films thickness to vary and thus impact on etching process and CD variation. As line-width decreases to 45 nm or 32 nm, bare wafer specifications for wafer flatness, ERO, and NT become increasingly tighter. Therefore, the metrology for the bare wafer measurement is critical to achieve those specifications. ITRS roadmap requires the site flatness of SFQR to be less than 45 nm and 32 nm for 45 nm and 32 nm node design and NT of TH2 to be less than 11 nm and 8 nm for 45 nm and 32 nm node design, respectively. The stage-of-the are technology has demonstrated that the wafer site flatness (SFQR) precision is 0.26 nm and worst site matching 6 nm of 100% FQA. NT precision of TH2 is 0.04 nm and worst wafer matching is 0.18 nm. The newly developed technology has demonstrated performance required to enable bare wafer specifications that support shrinking DOF for 45 nm and 32 nodes. Proc. of SPIE Vol. 6827 682723-13 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx SFQR SBIR Loifference: SFQR(beta)BSdb-SFQR(alpha) Loifference: SBIR(beta)BSdb-SBIR(alpha) 30 20 20 cr 10 4 10 _______ 0 -10 10 0 0 10 20 30 40 60 10 70 100 110 200 210 300 80 Mean: (IBIR(beta)8ldb+IBIR(alphaD/2 Mean: (SFQR(beta)8ldb+SFQR(alphaD/2 SFQR(beta)8ldb 08.6074 SFQR(alpha) 18.8207 Mean Difference -0.2032 ltd Error 0.01008 -0.1831 Upperll% LowerRl% -0.223 N Correlation t- Ratio -20. 16R3 0711 DF Prob ItI <.0001 1.0000 <.0001 Prob t Prob < t 1712 0.11718 71.9811 72.3798 t-Ratio Mean Difference -0.3987 Prob < ItI ltd Error 0.03381 Prob t Upperll% Lowerls% -0.3324 -0.4811 Prob < t N 1712 0. 99818 IBIR(beta)8ldb IBIR(alpha) Correlation Figure 15 a) SFQR site flatness matching. DF -11.7791 1711 <.0001 1.0000 <.0001 b) SBIR site flatness matching ESFQR Lofference: ESFQR(beta)85db-ESFQR(WSNT)alpha J 20 16 12 4 -4 12 16 -20 0 10 20 30 40 80 60 70 80 90 100 110 Mean: (E6FQR(beta)8Sdb+ESFQR(WSNT)alpha)/2 48.0948 ESFQR(beta)lldb ESFQR(WSNT)alpha 47.7073 t- Ratio Mean Difference Std Error 9mb Uppercl% Lowemgs% Correlation 0.38741 0.04821 0.48203 0.29286 1139 0.99613 8.037307 1138 DF ItI 9mb t <.0001 <.0001 9mb < t 1.0000 Figure 16 a) ERO: ESFQR matching Proc. of SPIE Vol. 6827 682723-14 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx THA 2 SQMM THA 4 SQMM LOfference: ThA2(beta)88db-THA2(WSNT)alpha 1 LDIffBrBnCB: THA4(Beta)85db-THA4(WSNT)alpha 1.00- z I P 1.6 0.80 z 0.60 1.2 0.40 0.8 0.20 - 1.4 0.0 0.00 - 0.20 0.40 6-0.600.80 9 4.1 1.0 1.1 6.0 6.1 7.0 7.1 8.0 8.1 9.0 11 12 13 14 13 16 17 10 Mean: Mean: (THA4(Beta)8ldb+THA4(WSNT)alpha)/2 (THA2(beta)88db+THA2(WSNT)alpha)/2 6.71706 THA2(beta)88db THA2(WSNT)alpha 6.78937 Mean Difference -0.0423 Std Error 0.01878 -0.0026 Uppercl% Lowergl% -0.0821 N b Correlat:on DF 9mb 9mb 13.3642 THA4(Beta)8ldb THA4(WSNT)alpha 114421 Mean Difference -0.0783 Std Error 0.06301 0.01124 UpperRl% Lowergs% -0.2118 -2.28368 t- Rat:o ItI t 9mb Ct 16 0.0386 0.9807 0.0193 17 0.99869 Figure 17 a) TH2 of NT matching. Correlation -1.24321 t- Ratio DF Prob Prob Prob 16 ItI t t 0.2317 0.8841 0.1188 17 0.88102 b) TH4 of NT matching. 10. References: 1. 2. ITRS 2006 Update “Front End Processes”. B.J. Lin “New λ /NA scaling equations for resolution and depth-of-focus”, Proceedings of SPIE Vol. 4000 (2000). 3. J. Mulkens et al “Defects, Overlay and Focus Performance Improvements with Five Generations of Immersion Exposure Systems,” Optical Microlithography xx, Proc. Of SPIE Vol. 6529, 652005, (2007). 4. SEMI M1-0600, “Specifications for polished monocrystalline silicon wafers”, SEMI 2000. 5. SEMI M68-0307 (Preliminary), “Practice for determining wafer near-edge geometry from a measured height data array using a curvature metric, ZDD”, SEMI 2007. 6. SEMI M67-1106 (Preliminary), “Practice for determining wafer near-edge geometry from a measured thickness data array using the ESFQR and ESFQD metrics”, SEMI 2006. 7. SEMI M43-0301, “Guide for reporting wafer nanotopography”, SEMI 2001. 8. D. Boning et al “Characterization and Modeling of Nanotopography Effects on CMP”, MIT, International CMP Symposium, 2000. 9. A. Grenville et al “Electrical Critical Dimension Metrology for 100-nm Line-widths and Below,” Proc. Of SPIE Vol 4000 (2000). 10. K. Freischlad et al “Interferometry for Wafer Dimensional Metrology,” Proc. of SPIE San Diego (2007). 11. SEMI M49-0307, “Guide for specifying geometry measurement systems for silicon wafers for the 130-nm to 65-nm technology generations”, SEMI 2007. Proc. of SPIE Vol. 6827 682723-15 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx