TW201249598A - Varying coefficients and functions for polishing control - Google Patents
Varying coefficients and functions for polishing control Download PDFInfo
- Publication number
- TW201249598A TW201249598A TW101113982A TW101113982A TW201249598A TW 201249598 A TW201249598 A TW 201249598A TW 101113982 A TW101113982 A TW 101113982A TW 101113982 A TW101113982 A TW 101113982A TW 201249598 A TW201249598 A TW 201249598A
- Authority
- TW
- Taiwan
- Prior art keywords
- layer
- spectrum
- stack
- different
- substrate
- Prior art date
Links
- 230000006870 function Effects 0.000 title claims abstract description 171
- 238000005498 polishing Methods 0.000 title claims description 76
- 238000001228 spectrum Methods 0.000 claims abstract description 390
- 239000000758 substrate Substances 0.000 claims abstract description 242
- 238000000034 method Methods 0.000 claims abstract description 121
- 230000008033 biological extinction Effects 0.000 claims abstract description 95
- 230000003287 optical effect Effects 0.000 claims abstract description 94
- 238000000227 grinding Methods 0.000 claims description 102
- 229910052751 metal Inorganic materials 0.000 claims description 60
- 239000002184 metal Substances 0.000 claims description 60
- 230000003595 spectral effect Effects 0.000 claims description 48
- RYGMFSIKBFXOCR-UHFFFAOYSA-N Copper Chemical group [Cu] RYGMFSIKBFXOCR-UHFFFAOYSA-N 0.000 claims description 22
- 238000002310 reflectometry Methods 0.000 claims description 22
- 229910052802 copper Inorganic materials 0.000 claims description 21
- 239000010949 copper Substances 0.000 claims description 21
- 238000004364 calculation method Methods 0.000 claims description 18
- 238000012360 testing method Methods 0.000 claims description 9
- 239000004575 stone Substances 0.000 claims description 7
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 claims description 6
- 229910052799 carbon Inorganic materials 0.000 claims description 6
- GNPVGFCGXDBREM-UHFFFAOYSA-N germanium atom Chemical group [Ge] GNPVGFCGXDBREM-UHFFFAOYSA-N 0.000 claims description 4
- 229910052732 germanium Inorganic materials 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 230000003647 oxidation Effects 0.000 claims description 2
- 238000007254 oxidation reaction Methods 0.000 claims description 2
- QCLQZCOGUCNIOC-UHFFFAOYSA-N azanylidynelanthanum Chemical compound [La]#N QCLQZCOGUCNIOC-UHFFFAOYSA-N 0.000 claims 1
- 229910052797 bismuth Inorganic materials 0.000 claims 1
- JCXGWMGPZLAOME-UHFFFAOYSA-N bismuth atom Chemical group [Bi] JCXGWMGPZLAOME-UHFFFAOYSA-N 0.000 claims 1
- 238000006243 chemical reaction Methods 0.000 claims 1
- ZKEYULQFFYBZBG-UHFFFAOYSA-N lanthanum carbide Chemical compound [La].[C-]#[C] ZKEYULQFFYBZBG-UHFFFAOYSA-N 0.000 claims 1
- 230000007774 longterm Effects 0.000 claims 1
- 239000010979 ruby Substances 0.000 claims 1
- 229910001750 ruby Inorganic materials 0.000 claims 1
- 229910001925 ruthenium oxide Inorganic materials 0.000 claims 1
- WOCIAKWEIIZHES-UHFFFAOYSA-N ruthenium(iv) oxide Chemical compound O=[Ru]=O WOCIAKWEIIZHES-UHFFFAOYSA-N 0.000 claims 1
- 239000002699 waste material Substances 0.000 claims 1
- 239000010410 layer Substances 0.000 description 477
- 239000000463 material Substances 0.000 description 35
- 238000012544 monitoring process Methods 0.000 description 35
- 238000012886 linear function Methods 0.000 description 26
- 230000004888 barrier function Effects 0.000 description 25
- 230000008569 process Effects 0.000 description 24
- 238000005259 measurement Methods 0.000 description 21
- 239000006185 dispersion Substances 0.000 description 18
- 210000003128 head Anatomy 0.000 description 18
- 230000008859 change Effects 0.000 description 17
- 230000009021 linear effect Effects 0.000 description 14
- 238000011065 in-situ storage Methods 0.000 description 13
- 238000010586 diagram Methods 0.000 description 12
- 235000012431 wafers Nutrition 0.000 description 12
- 239000000203 mixture Substances 0.000 description 10
- 238000004590 computer program Methods 0.000 description 9
- 238000004519 manufacturing process Methods 0.000 description 9
- BOTDANWDWHJENH-UHFFFAOYSA-N Tetraethyl orthosilicate Chemical compound CCO[Si](OCC)(OCC)OCC BOTDANWDWHJENH-UHFFFAOYSA-N 0.000 description 8
- 239000004020 conductor Substances 0.000 description 8
- 239000010408 film Substances 0.000 description 8
- 238000000151 deposition Methods 0.000 description 7
- 238000007517 polishing process Methods 0.000 description 7
- 238000005070 sampling Methods 0.000 description 7
- 239000004065 semiconductor Substances 0.000 description 7
- 239000003989 dielectric material Substances 0.000 description 6
- 230000008901 benefit Effects 0.000 description 5
- 230000008021 deposition Effects 0.000 description 5
- 239000000835 fiber Substances 0.000 description 5
- 239000000945 filler Substances 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 239000002002 slurry Substances 0.000 description 5
- 238000003860 storage Methods 0.000 description 5
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 4
- 238000007667 floating Methods 0.000 description 4
- 210000001747 pupil Anatomy 0.000 description 4
- 239000007787 solid Substances 0.000 description 4
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 4
- CURLTUGMZLYLDI-UHFFFAOYSA-N Carbon dioxide Chemical compound O=C=O CURLTUGMZLYLDI-UHFFFAOYSA-N 0.000 description 3
- 239000011248 coating agent Substances 0.000 description 3
- 238000000576 coating method Methods 0.000 description 3
- 230000007423 decrease Effects 0.000 description 3
- 239000012528 membrane Substances 0.000 description 3
- 238000003801 milling Methods 0.000 description 3
- 150000004767 nitrides Chemical class 0.000 description 3
- 238000002161 passivation Methods 0.000 description 3
- 238000004611 spectroscopical analysis Methods 0.000 description 3
- 239000000126 substance Substances 0.000 description 3
- 229910052715 tantalum Inorganic materials 0.000 description 3
- GUVRBAGPIYLISA-UHFFFAOYSA-N tantalum atom Chemical group [Ta] GUVRBAGPIYLISA-UHFFFAOYSA-N 0.000 description 3
- KJTLSVCANCCWHF-UHFFFAOYSA-N Ruthenium Chemical compound [Ru] KJTLSVCANCCWHF-UHFFFAOYSA-N 0.000 description 2
- 229910000831 Steel Inorganic materials 0.000 description 2
- 239000003082 abrasive agent Substances 0.000 description 2
- 238000010521 absorption reaction Methods 0.000 description 2
- 229910000420 cerium oxide Inorganic materials 0.000 description 2
- 239000010432 diamond Substances 0.000 description 2
- 230000005672 electromagnetic field Effects 0.000 description 2
- 239000007769 metal material Substances 0.000 description 2
- 229910052757 nitrogen Inorganic materials 0.000 description 2
- SIWVEOZUMHYXCS-UHFFFAOYSA-N oxo(oxoyttriooxy)yttrium Chemical compound O=[Y]O[Y]=O SIWVEOZUMHYXCS-UHFFFAOYSA-N 0.000 description 2
- 238000000059 patterning Methods 0.000 description 2
- 239000002689 soil Substances 0.000 description 2
- 239000010959 steel Substances 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 241000238557 Decapoda Species 0.000 description 1
- 241000255925 Diptera Species 0.000 description 1
- 241000238631 Hexapoda Species 0.000 description 1
- UFHFLCQGNIYNRP-UHFFFAOYSA-N Hydrogen Chemical compound [H][H] UFHFLCQGNIYNRP-UHFFFAOYSA-N 0.000 description 1
- 206010021703 Indifference Diseases 0.000 description 1
- MWUXSHHQAYIFBG-UHFFFAOYSA-N Nitric oxide Chemical compound O=[N] MWUXSHHQAYIFBG-UHFFFAOYSA-N 0.000 description 1
- IOVCWXUNBOPUCH-UHFFFAOYSA-M Nitrite anion Chemical compound [O-]N=O IOVCWXUNBOPUCH-UHFFFAOYSA-M 0.000 description 1
- 101150043219 POLD2 gene Proteins 0.000 description 1
- 206010036790 Productive cough Diseases 0.000 description 1
- NRTOMJZYCJJWKI-UHFFFAOYSA-N Titanium nitride Chemical compound [Ti]#N NRTOMJZYCJJWKI-UHFFFAOYSA-N 0.000 description 1
- 239000006061 abrasive grain Substances 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000002238 attenuated effect Effects 0.000 description 1
- 238000010009 beating Methods 0.000 description 1
- 229910002092 carbon dioxide Inorganic materials 0.000 description 1
- 239000001569 carbon dioxide Substances 0.000 description 1
- 229910002090 carbon oxide Inorganic materials 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 239000011231 conductive filler Substances 0.000 description 1
- 210000000877 corpus callosum Anatomy 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000005137 deposition process Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 229910003460 diamond Inorganic materials 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000001804 emulsifying effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 125000001495 ethyl group Chemical group [H]C([H])([H])C([H])([H])* 0.000 description 1
- 238000011049 filling Methods 0.000 description 1
- 210000003811 finger Anatomy 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012625 in-situ measurement Methods 0.000 description 1
- 239000011810 insulating material Substances 0.000 description 1
- 239000012212 insulator Substances 0.000 description 1
- 239000011229 interlayer Substances 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 230000000873 masking effect Effects 0.000 description 1
- 235000012054 meals Nutrition 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 150000002739 metals Chemical class 0.000 description 1
- NFFIWVVINABMKP-UHFFFAOYSA-N methylidynetantalum Chemical compound [Ta]#C NFFIWVVINABMKP-UHFFFAOYSA-N 0.000 description 1
- 231100000862 numbness Toxicity 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 239000012788 optical film Substances 0.000 description 1
- JMANVNJQNLATNU-UHFFFAOYSA-N oxalonitrile Chemical compound N#CC#N JMANVNJQNLATNU-UHFFFAOYSA-N 0.000 description 1
- BMMGVYCKOGBVEV-UHFFFAOYSA-N oxo(oxoceriooxy)cerium Chemical compound [Ce]=O.O=[Ce]=O BMMGVYCKOGBVEV-UHFFFAOYSA-N 0.000 description 1
- 238000000206 photolithography Methods 0.000 description 1
- 238000013138 pruning Methods 0.000 description 1
- 238000000746 purification Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 229910052707 ruthenium Inorganic materials 0.000 description 1
- 239000002356 single layer Substances 0.000 description 1
- 239000007779 soft material Substances 0.000 description 1
- 210000003802 sputum Anatomy 0.000 description 1
- 208000024794 sputum Diseases 0.000 description 1
- 229910003468 tantalcarbide Inorganic materials 0.000 description 1
- MZLGASXMSKOWSE-UHFFFAOYSA-N tantalum nitride Chemical compound [Ta]#N MZLGASXMSKOWSE-UHFFFAOYSA-N 0.000 description 1
- 239000010409 thin film Substances 0.000 description 1
- 210000003813 thumb Anatomy 0.000 description 1
- 239000010936 titanium Substances 0.000 description 1
- 229910052719 titanium Inorganic materials 0.000 description 1
- 230000002463 transducing effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 239000012780 transparent material Substances 0.000 description 1
- 229910052721 tungsten Inorganic materials 0.000 description 1
- 238000001429 visible spectrum Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B24—GRINDING; POLISHING
- B24B—MACHINES, DEVICES, OR PROCESSES FOR GRINDING OR POLISHING; DRESSING OR CONDITIONING OF ABRADING SURFACES; FEEDING OF GRINDING, POLISHING, OR LAPPING AGENTS
- B24B37/00—Lapping machines or devices; Accessories
- B24B37/005—Control means for lapping machines or devices
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B24—GRINDING; POLISHING
- B24B—MACHINES, DEVICES, OR PROCESSES FOR GRINDING OR POLISHING; DRESSING OR CONDITIONING OF ABRADING SURFACES; FEEDING OF GRINDING, POLISHING, OR LAPPING AGENTS
- B24B37/00—Lapping machines or devices; Accessories
- B24B37/04—Lapping machines or devices; Accessories designed for working plane surfaces
- B24B37/042—Lapping machines or devices; Accessories designed for working plane surfaces operating processes therefor
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/27—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands using photo-electric detection ; circuits for computing concentration
- G01N21/274—Calibration, base line adjustment, drift correction
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/41—Refractivity; Phase-affecting properties, e.g. optical path length
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
- H01L21/02—Manufacture or treatment of semiconductor devices or of parts thereof
- H01L21/04—Manufacture or treatment of semiconductor devices or of parts thereof the devices having potential barriers, e.g. a PN junction, depletion layer or carrier concentration layer
- H01L21/18—Manufacture or treatment of semiconductor devices or of parts thereof the devices having potential barriers, e.g. a PN junction, depletion layer or carrier concentration layer the devices having semiconductor bodies comprising elements of Group IV of the Periodic Table or AIIIBV compounds with or without impurities, e.g. doping materials
- H01L21/30—Treatment of semiconductor bodies using processes or apparatus not provided for in groups H01L21/20 - H01L21/26
- H01L21/302—Treatment of semiconductor bodies using processes or apparatus not provided for in groups H01L21/20 - H01L21/26 to change their surface-physical characteristics or shape, e.g. etching, polishing, cutting
- H01L21/304—Mechanical treatment, e.g. grinding, polishing, cutting
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
- H01L21/02—Manufacture or treatment of semiconductor devices or of parts thereof
- H01L21/04—Manufacture or treatment of semiconductor devices or of parts thereof the devices having potential barriers, e.g. a PN junction, depletion layer or carrier concentration layer
- H01L21/18—Manufacture or treatment of semiconductor devices or of parts thereof the devices having potential barriers, e.g. a PN junction, depletion layer or carrier concentration layer the devices having semiconductor bodies comprising elements of Group IV of the Periodic Table or AIIIBV compounds with or without impurities, e.g. doping materials
- H01L21/30—Treatment of semiconductor bodies using processes or apparatus not provided for in groups H01L21/20 - H01L21/26
- H01L21/31—Treatment of semiconductor bodies using processes or apparatus not provided for in groups H01L21/20 - H01L21/26 to form insulating layers thereon, e.g. for masking or by using photolithographic techniques; After treatment of these layers; Selection of materials for these layers
- H01L21/3205—Deposition of non-insulating-, e.g. conductive- or resistive-, layers on insulating layers; After-treatment of these layers
- H01L21/321—After treatment
- H01L21/32115—Planarisation
- H01L21/3212—Planarisation by chemical mechanical polishing [CMP]
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L22/00—Testing or measuring during manufacture or treatment; Reliability measurements, i.e. testing of parts without further processing to modify the parts as such; Structural arrangements therefor
- H01L22/10—Measuring as part of the manufacturing process
- H01L22/12—Measuring as part of the manufacturing process for structural parameters, e.g. thickness, line width, refractive index, temperature, warp, bond strength, defects, optical inspection, electrical measurement of structural dimensions, metallurgic measurement of diffusions
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L22/00—Testing or measuring during manufacture or treatment; Reliability measurements, i.e. testing of parts without further processing to modify the parts as such; Structural arrangements therefor
- H01L22/20—Sequence of activities consisting of a plurality of measurements, corrections, marking or sorting steps
- H01L22/26—Acting in response to an ongoing measurement without interruption of processing, e.g. endpoint detection, in-situ thickness measurement
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Power Engineering (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Computer Hardware Design (AREA)
- Pathology (AREA)
- Mechanical Engineering (AREA)
- Condensed Matter Physics & Semiconductors (AREA)
- Manufacturing & Machinery (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Mechanical Treatment Of Semiconductor (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
Description
201249598 六、發明說明： 【發明所屬之技術領域】 本揭示案係關於例如在基板之化學機械研磨期間的研 磨控制方法。 【先前技術】 積體電路通常藉由矽晶圓上之傳導層、半導體層或絕 緣層之連續沈積形成在基板上。一個製造步驟包括在非 平坦表面上方沈積填料層及將填料層平坦化。對於某些 應用而言，將填料層平坦化直至暴露圖案化之層之頂表 面。傳導填料層（例如）可沈積在圖案化之絕緣層上以 填滿絕緣層中之溝槽或孔。平坦化之後，剩餘在絕緣層 之凸起圖案之間之傳導層部分形成通孔、插塞及線，該 等通孔、插塞及線提供介於基板上之薄膜電路之間的傳 導路徑。對於諸如氧化物研磨之其他應用而言，平坦化 填料層直至在非平坦表面上方留下預先決定之厚度。另 外，通常需要平坦化基板表面以用於光刻法。 化學機械研磨（CMP)為一種公認的平坦化方法。此、， 坦：方法通常要求基板安裝在承載頭上。通常將基板： 暴露表面與旋轉之研磨墊緊靠放置。承載頭在基板θ 供可控制的負載以推動基板緊靠研磨墊。通常將諸2 有磨粒之漿料之研磨液供應至研磨墊之表面。 一 CMP中存在的一個問題為決定研磨製程是否完、 即，是否已經將基板層平坦化至所欲之平坦 、亦 5 201249598 或何時已經移除所欲數量之材料。基板層最初厚度、將 料組成、研磨塾情況、研磨塾與基板之間的相對速声及 K 基板上之負載的變化可弓ί起材料移除速率的變化。該等 變化引起到達研磨終點所需時間之變化。因此，可J無 法僅將研磨、终點決定為研磨時間之函I。 b…、 在些系統中，研磨期間在原位光學監視基板，例如 經由研磨塾中之窗口。《而’現有的光學監視技術可 無法滿足半導體元件製造商日益增加之需求。 【發明内容】 在一些光學監視過程中，將原位量測之光譜（例如， 在CMP之研磨製程期間)與參考光譜庫相比較以發現最 佳匹配參考光譜。建立參考光譜庫之-種技術為基於薄 料疊之光學性質之理論計算參考㈣。料—些基板 而言’在基板上經照射之層堆疊可隨著不同量測改變。 然：，有可能產生對應於層堆疊之各種組合之多個參考 光4。另彳’—些基板（例如，在線後端製程中之基板） 可具有極其複雜之層堆疊，料複雜之層堆疊在計算上 係十刀困難且不可靠的。然而，有可能將複雜層堆疊之 下部分作為單一實體處理。 另外，貫際上沈積層之n值及k值（光學膜性質分別 才曰的疋折射率及消光係數）視膜組成及臈沈積控制而定 隨著不同客戶及不同批次改變。因此，自光學模型產生 之參考光瑨可能為不準確的。解決此問題之技術為產生 201249598 各種η值及k值之參考光譜。舉例而言，對於許多介電 材料，可使用柯西（C_hy)等式模型化可見光譜令之 η及k之分散。對於該等層中之至少—者建立基線柯西 模型’及时在計算理論光譜庫巾柯西模型之係數經由 使用者定義之邊線改變。 在一個態樣中，彦夺史連_ I % 座生翏考先瑨庫之方法包括以下步 驟：儲存具有複數個層之層堆疊之光學模型；接收識別 來自複數個層之第一層之一 一組一或更多個消光係數函 組一或更多個折射率函數及 數之使用者輸入，其中該組 一或更多個折射率函數包括複數個不同折射率函數或該 組-或更多個消光係數函數包括複數個不㈣光係數函 數’以及對於來自該組折射率函數之折射率函數及來自 該組消光係數函數之消光係數函數之每一組合，基於折 射率函數、消光係數函數及第一層之第一厚度使用光學 杈型計算參考光譜，以產生複數個參考光譜。 實施可包括以下特徵中之一或更多者。該組一或更多 個折射率函數可包括複數個不同折射率函數，例如，2 至1〇個函數。接收識別複數個不同折射率函數之使用者 輸入之步驟可包括以下步驟：接收識別折射率函數之第 係數之第一複數個不同第一值（例如，2至1 〇個值） 之使用者輸入。接收識別第一複數個不同第一值之使甩 者輸入之步驟可包括以下步驟：接收下限值、上限值及 值增置或若干值。接收識別複數個不同折射率函數之使 用者輪入之步驟可包括以下步驟：接收識別折射率函數 201249598 之第二係數之第二複數個不同第二值之使用者輸入。對 於來自第—複數個值之第-值及㈣第二複數個值之第 二值之每一組合，可計算指數函數以產生複數個不同指 數函數。計算指數函數之步驟可包括以下步驟：計曾201249598 6. INSTRUCTIONS OF THE INVENTION: TECHNICAL FIELD The present disclosure relates to a grinding control method, for example, during chemical mechanical polishing of a substrate. [Prior Art] An integrated circuit is usually formed on a substrate by continuous deposition of a conductive layer, a semiconductor layer or an insulating layer on a germanium wafer. A fabrication step includes depositing a layer of filler over the non-planar surface and planarizing the layer of filler. For some applications, the filler layer is planarized until the top surface of the patterned layer is exposed. A conductive filler layer, for example, may be deposited over the patterned insulating layer to fill the trenches or holes in the insulating layer. After planarization, portions of the conductive layer remaining between the raised patterns of the insulating layer form vias, plugs, and lines that provide a conductive path between the thin film circuits on the substrate. For other applications such as oxide milling, the filler layer is planarized until a predetermined thickness is left over the non-planar surface. In addition, it is often desirable to planarize the surface of the substrate for photolithography. Chemical mechanical polishing (CMP) is a well-established planarization method. This, Tan: The method usually requires the substrate to be mounted on the carrier head. The substrate: the exposed surface is typically placed against the rotating polishing pad. The carrier head provides a controllable load on the substrate θ to push the substrate against the polishing pad. A slurry of 2 abrasive grains is usually supplied to the surface of the polishing pad. One problem with a CMP is to determine if the polishing process is complete, that is, whether the substrate layer has been flattened to the desired flatness, or when the desired amount of material has been removed. The initial thickness of the substrate layer, the composition of the material, the condition of the polishing crucible, the relative speed of the polishing crucible and the substrate, and the change in the load on the K substrate can vary the rate of material removal. These changes cause a change in the time required to reach the end of the grinding. Therefore, it is impossible to determine only the polishing and the end point as the letter I of the polishing time. b... In these systems, the substrate is optically monitored in situ during grinding, for example via a window in a grinding crucible. "And the existing optical monitoring technology can not meet the increasing demand of semiconductor component manufacturers. SUMMARY OF THE INVENTION In some optical monitoring processes, an in situ measured spectrum (e.g., during a CMP polishing process) is compared to a reference spectral library to find the best matching reference spectrum. The technique for establishing a reference spectral library is a theoretical calculation reference based on the optical properties of thin stacks (4). In the case of substrates - the layer stack that is illuminated on the substrate can vary with different measurements. However, it is possible to generate a plurality of reference lights 4 corresponding to various combinations of layer stacks. Alternatively, substrates (e.g., substrates in an in-line backend process) can have extremely complex layer stacks, which are computationally complex and unreliable. However, it is possible to treat the lower portion of the complex layer stack as a single entity. In addition, the n-value and k-value of the deposited layer on the surface (the refractive index and extinction coefficient of the optical film properties, respectively) vary depending on the film composition and the enthalpy deposition control with different customers and batches. Therefore, the reference pupil generated from the optical model may be inaccurate. The technique to solve this problem is to generate a reference spectrum of various η values and k values of 201249598. For example, for many dielectric materials, the Caxi (C_hy) equation can be used to model the visible spectrum to make the dispersion of η and k. For at least one of the layers, the baseline Cauchy model is established. In time, the coefficients of the theoretical theoretical spectral Kobe model are changed via user-defined edges. In one aspect, the method of 夺 史 _ I I I I I I I 包括 包括 包括 包括 包括 包括 包括 包括 包括 包括 包括 包括 包括 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存 储存a set of one or more extinction coefficient groups of one or more refractive index functions and a user input of the number, wherein the set of one or more refractive index functions comprises a plurality of different refractive index functions or the set - or The plurality of extinction coefficient functions include a plurality of non-fourth optical coefficient functions 'and each combination of a refractive index function from the set of refractive index functions and an extinction coefficient function from the set of extinction coefficient functions, based on a refractive index function, an extinction coefficient function And the first thickness of the first layer is calculated using an optical 杈 type to generate a plurality of reference spectra. Implementations may include one or more of the following features. The set of one or more refractive index functions can include a plurality of different refractive index functions, for example, 2 to 1 函数 functions. The step of receiving a user input identifying a plurality of different refractive index functions can include the step of receiving a user input identifying a first plurality of different first values (eg, 2 to 1 值 values) of a coefficient of the refractive index function . The step of receiving an enabler input identifying the first plurality of different first values may include the steps of: receiving a lower limit value, an upper limit value, and a value addition or a number of values. The step of receiving a user rounding to identify a plurality of different refractive index functions can include the step of receiving a user input identifying a second plurality of different second values of the second coefficient of the refractive index function 201249598. For each combination of the first value from the first plurality of values and the second value of the second plurality of values, an exponential function can be calculated to produce a plurality of different index functions. The step of calculating the exponential function may include the following steps:
— η Λ t B C 其中η(λ)為指數函數’A為第一值，B為第二值且〇 為第一值。5玄組一或更多個消光係數函數可包括複數個 不同消光係、數函數，例如，2至1G個函m或更 多個折射率函數可包括複數個不同折射率函數且該組— 或更多個消光係數函數可包括複數個不同消光係數函 數。可接收識別基板之第—層之複數個不同厚度值之使 用者輸人，該S數個不同厚度值包括第一厚度值。對於 來自該組折射率函數之折射率函數、來自該組消光係數 函數之消光係數函數及來自該複數個不同厚度值之厚声 值之每1合’使Μ學模型計算參考光譜。使用光學 模型計算參考光譜可包括轉換矩陣法。基板可包括阳 層之堆$ ’其中堆蟹包括第—層且其中層。為底層且層 ρ為最外面之第一層。計算參考光譜之步驟可包括以下 步驟：如下計算堆疊反射率：— η Λ t B C where η(λ) is the exponential function 'A is the first value, B is the second value and 〇 is the first value. 5 玄 group one or more extinction coefficient functions may include a plurality of different extinction systems, a number function, for example, 2 to 1G functions or more refractive index functions may include a plurality of different refractive index functions and the group - or More extinction coefficient functions may include a plurality of different extinction coefficient functions. A user may receive a plurality of different thickness values identifying the first layer of the substrate, the S number of different thickness values including the first thickness value. The reference spectrum is calculated for the dropout model for each of the refractive index functions from the set of refractive index functions, the extinction coefficient function from the set of extinction coefficient functions, and the thick acoustic value from the plurality of different thickness values. Calculating the reference spectrum using an optical model can include a transformation matrix method. The substrate may comprise a stack of positive layers $' wherein the stack of crabs comprises a first layer and a middle layer thereof. The bottom layer and the layer ρ are the outermost first layer. The step of calculating the reference spectrum may include the steps of: calculating the stack reflectivity as follows:
R Ερ %, 其中對於每一層4〇，Ej及Hj計算為： [| cos — sin a. 201249598 其中E〇為1且Η〇&μ〇，且其中對於每一層#〇, J ( J lk)) C〇S 且 gj=2?c(n厂ikj)-Vc〇s φ』/λ，其中 nj 為 J之折射率’kj為層j之消光係數，tj為層j之厚度， Φ·)為光至層 之入射角，且λ為波長。計算參考光譜之 步驟可包括以下牛.Α 田 卜步驟·计异堆$反射率RSTACK2: iL· n 一R Ερ %, where for each layer 4〇, Ej and Hj are calculated as: [| cos — sin a. 201249598 where E〇 is 1 and Η〇&μ〇, and where for each layer #〇, J ( J lk )) C〇S and gj=2?c(n factory ikj)-Vc〇s φ』/λ, where nj is the refractive index of J'kj is the extinction coefficient of layer j, tj is the thickness of layer j, Φ· ) is the incident angle of light to the layer, and λ is the wavelength. The steps of calculating the reference spectrum may include the following steps: 步骤 Α Steps · Counting Heap # Reflectance RSTACK2: iL· n
Rstack: = —^ εΡ 「丐1Rstack: = —^ εΡ ”丐1
CQS3j - sin g UJ _ίμ;· sin gj cos 其中E〇為 凡H〇為μ〇，且其中對於每一層】含〇, Ν’’、)).，…广2一广KV〜).Vcos φ」/λ， 其中η』為層j之折射率，kj為層』之消光係數，叫為增 加層j之消光係數之數量，tj為層〗之厚度，屯為光至層 J之入射角，且λ為波長。計算參考光譜之步驟包含以 下步驟：使用光學模型計算第—光譜Rstack及將第一光CQS3j - sin g UJ _ίμ;· sin gj cos where E〇 is where H〇 is μ〇, and for each layer] contains 〇, Ν'',)).,...广2一广KV~).Vcos φ " / λ, where η 』 is the refractive index of layer j, kj is the extinction coefficient of the layer 』, called the number of extinction coefficients of layer j, tj is the thickness of the layer, 屯 is the incident angle of light to layer J, And λ is the wavelength. The step of calculating the reference spectrum comprises the steps of: calculating the first-spectrum Rstack and the first light using an optical model
S 譜rstack及第二光譜組合。計算參考光譜Rlibrary之步 驟可包括以下步驟：計算 9 7 7201249598 其中RStACK1為第一光譜，Rstack2為第二光譜， RREFERENCE為第—堆疊及第二堆疊之底層之光譜，且 為介於〇與！之間的值。底層可為石夕或金屬。； it::碳氧化"化"化"‘碳 在另—態樣中’產生參考光誶 驟：儲存具有複數個、i括以下步 折射玄了叙 層之層堆疊之光學模型；接收識別 者輸^料來自倾個不同值值=吏用 射率函數以產生複數1杯 ，计异折 個折射率函數之每—折射率“數，以及對於來自複數 光係數函數及第_層之第—’基於折射率函數、消 光譜，以產生複數麻考=度使用光學模型計算參考 之第二係數n數㈣=可接收識別折射率函數 對於來自第-複數個不同第弟二值之使用者輸入’及 數個不同第二值 值之第一值及來自第二複 '〜矛—值之jfe 數。 —組合，可計算折射率函 實施可包括以下㈣tn 之步驟可包括以下步驟：計算更多者。計算指數函數S spectrum rstack and second spectral combination. The step of calculating the reference spectrum Rlibrary may include the following steps: calculating 9 7 7201249598 wherein RStACK1 is the first spectrum, Rstack2 is the second spectrum, and RREFERENCE is the spectrum of the bottom layer of the first stack and the second stack, and is between 〇 and ! The value between. The bottom layer can be stone or metal. ; it::Carbon oxidation"化"化" 'Carbon in another-state' generates a reference light step: storing an optical model with a plurality of layers, including the following steps, to refract the stack of layers; The receiver recognizes the input from the different values of the value = 吏 using the radiance function to generate a complex number of cups, the refractive index of each of the refractive index functions, and the function from the complex optical coefficient and the _th layer The first--based on the refractive index function, the extinction spectrum, to generate the complex number of numbness = degree using the optical model to calculate the reference second coefficient n number (four) = the receivable identification refractive index function for the first to plural different second brothers The user inputs 'and a first value of a plurality of different second value values and a jfe number from the second complex' to the spear-value. - Combining, calculating the refractive index function The step of including the following (4) tn may include the following steps: Calculate more. Calculate the exponential function
η(’·）Ί . C “中η(λ)為指數函數，a C為第三值。 馮第—值，B為第二值，且η('·)Ί . C "中η(λ) is an exponential function, a C is a third value. Feng-value, B is the second value, and
S 10 201249598 在另一態樣中，根據先前方法產生參考光譜庫；研磨 基板；在研磨期間量測來自基板之光之一序列光譜；對 於:广序歹!光4中之每—量測之光譜，發現最佳匹配參考 光a、產生序列最佳匹配參考光譜；以及基於該序列 最佳匹配參考光错決^研磨終點或研磨速率之調整中之 至少一者。 在另一態樣中，產味夫去_ 座生參考先瑨庫之方法包括以下步 驟接收代表基板上之第一層堆疊之反射率之第一光 譜，第—堆疊包括第—介電層；接收代表基板上之第二 層堆疊之反射率之筮-土 ％ _ 第一光δθ，弟二堆疊包括第一介電層 及不位於第一堆疊中 笛_人 τ之第一”電層；接收識別基板上之 弟一堆疊或第二堆最φ夕$小 隹且中之至^ 一者之衩數個不同貢獻百 分比之使用者輪入.Β拟认+上 卜 ，乂及對於來自衩數個不同貢獻百分 比之每一貢獻百分士，拍诚# , ^ 曰刀比，根據第一光譜、第二光譜及貢獻 百分比計算參考光譜。 實施可包括以下特微φ ,. s . 将徵中之一或更多者。計算參考光譜 1S 10 201249598 In another aspect, a reference spectral library is generated according to a prior method; the substrate is ground; a sequence spectrum of light from the substrate is measured during the grinding; for: a wide sequence of light! Spectral, finding the best matching reference light a, generating a sequence best matching reference spectrum; and adjusting at least one of the best matching reference optical error based on the sequence. In another aspect, the method for producing a reference reference library includes the following steps of receiving a first spectrum representing a reflectance of a first layer stack on a substrate, the first stack comprising a first dielectric layer; Receiving a 筮-%%_first light δθ representing a reflectivity of the second layer stack on the substrate, the second stack comprising a first dielectric layer and a first “electric layer” not located in the first stack; Receiving a user-involved disc on the identification substrate on the stack or the second stack of the most φ $ 隹 隹 隹 隹 隹 隹 隹 隹 隹 ^ Β 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者Each contribution percentage of each of the different contribution percentages, Pao Cheng #, ^ 曰刀 ratio, calculates the reference spectrum based on the first spectrum, the second spectrum, and the contribution percentage. The implementation may include the following special φ, . s . One or more of them. Calculate the reference spectrum 1
RR
Rlibrary之步驟可包括以下步驟：計算 曰 'LrI3F：.AF. ϊ* […‘ (1 ϋ 嫩 2] 5 其中RSTACK1為第一光譜，Rstack2為第二光譜， Rreference為第一堆疊及第二堆疊之底層之光譜，^ X 為第-堆疊之百分比貢獻。底層可為矽或金屬。可接收 代表基板上之金屬層之反射率鼓. 矛—尤°日，可接收識別 金屬層之複數個不同今屬首勃 小Π玉屬貝獻百分比之使用者輸入丨以 Λ 11 201249598 及對於來自複數個不同貢獻百分比之每一貢獻百分比及 對於來自複數個不同金屬貢獻百分比之每一金屬貢獻百 分比，可根據第一光譜、第二光譜、第三光譜、貢獻百 分比及金屬貢獻百分比計算參考光譜。計算參考光譜 Rlibrary之步驟可包括以下步驟：計算 ^LlSBAr.Y 一 [Yr p > Τ.Λ RRcrE^XCS Λ 為以⑺一（1 — Α — 1〇 :卜 /?订抑」 j 其中RSTACK1為第一光譜，Rstack2為第二光譜，Rmetal 為第二光譜，Rreference為堆疊之底層之光譜，且X為 第-堆疊之百分比貢獻’且γ為金屬之百分比貢獻。底 層可為金屬層之金屬。金屬層可為鋼。接收識別金屬層 之複數個不同金屬貢獻百分比之制者輸人之步驟可包 括以下步驟：接收識別第-堆疊之第—複數個不同貢獻 百分比之使用者輸入；及接收識第二 個不同貢獻百分比之使用者輪入;以及根據第= 不同貢獻百分比及第二複數個又η二 攸双個不冋貢獻百分比計算複數 個不同金屬貢獻百分比。複數個尤门二& i 固不同貢獻百分比可包括 2至10個值。複數個不同金屬言虹、 屬貝獻百分比可包括2至1〇 個值。接收識別複數個不同貢赴 U貝馱百分比之使用者輸入之 步驟可包括以下步驟：接收下PP尤、 叹卜限百分比、上限百分比及 百分比增量0可分別使用第—掩田 ， 乐堆®之光學模型及第二堆 堂之光學权型计异第一光譜及笛-丄4 曰久弟一光譜。計算第一光譜 之步驟包含以下步驟：計算堆疊 八町午KSTACK1 KSTAC：<\ ^ Μβ， 12 201249598 其中對於每一層j>0，Ej及Hj計算為： Γ σ Ί 仏 = cos 1 - -8i u,· J Γ σ Ί 4-1 • j- Ψ< sm g. cos 馬. Η·. L力一1J ， 其中E〇為1且H〇為μ〇，且其中對於每一層j_0, Pj = (nj-ikj)-cos c()j 且 gj = 27i:(nj-ikj).tj.cos (jjj/λ ’ 其中 η』為 層j之折射率，kj為層j之消光係數，tj為層j之厚度， φ』為光至層j之入射角，且λ為波長。計算第二光譜之 步驟可包括以下步驟：計算堆疊反射率R STACK2The Rlibrary step may include the following steps: calculating 曰 'LrI3F: .AF. ϊ* [...' (1 嫩 tender 2) 5 wherein RSTACK1 is the first spectrum, Rstack2 is the second spectrum, Rreference is the first stack and the second stack The spectrum of the bottom layer, ^ X is the percentage contribution of the first-stack. The bottom layer can be tantalum or metal. It can receive the reflectivity drum representing the metal layer on the substrate. The spear - especially the day, can receive multiple different layers of the identification metal layer This is the user input of the percentage of the first 勃 Π 属 贝 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 2012 The reference spectrum is calculated according to the first spectrum, the second spectrum, the third spectrum, the contribution percentage, and the metal contribution percentage. The step of calculating the reference spectrum Rlibrary may include the following steps: calculating ^LlSBAr.Y[Yr p > Τ.Λ RRcrE^ XCS Λ is (7) one (1 - Α - 1 〇: 卜 /?) j where RSTACK1 is the first spectrum, Rstack2 is the second spectrum, Rmetal is the second spectrum, Rreference is the heap The spectrum of the bottom layer of the stack, and X is the percentage contribution of the first-stack and γ is the percentage contribution of the metal. The bottom layer can be the metal of the metal layer. The metal layer can be steel. The percentage of contribution of the plurality of different metals receiving the identification metal layer is The step of inputting the person may include the steps of: receiving a user input identifying the first to the plurality of different contribution percentages of the first stack; and receiving a user rounding of the second different contribution percentage; and different contributions according to the third Percentage and the second plurality of η 攸 攸 冋 冋 百分比 % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % The percentage of the genus can include 2 to 1 。. The step of receiving the user input identifying the plurality of different tributaries to the U 驮 可 can include the following steps: receiving the PP, the percentage of the sigh, the percentage of the upper limit, and the percentage Increment 0 can use the first - masked field, the optical model of Ledun® and the optical weight of the second stack to calculate the first spectrum and flute-丄4 曰A spectrum. The step of calculating the first spectrum comprises the steps of: calculating the stack of Katsucho KSTACK1 KSTAC: <\^ Μβ, 12 201249598 wherein for each layer j > 0, Ej and Hj are calculated as: Γ σ Ί 仏 = cos 1 - -8i u,· J Γ σ Ί 4-1 • j- Ψ< sm g. cos horse. Η·. L force-1J , where E〇 is 1 and H〇 is μ〇, and where j_0 for each layer , Pj = (nj-ikj)-cos c()j and gj = 27i:(nj-ikj).tj.cos (jjj/λ ' where η is the refractive index of layer j and kj is the extinction coefficient of layer j , tj is the thickness of layer j, φ′′ is the incident angle of light to layer j, and λ is the wavelength. The step of calculating the second spectrum may include the step of calculating the stack reflectivity R STACK2
R STACK! Ερ Η^_ ^ιρR STACK! Ερ Η^_ ^ιρ
Γ^-1 k-J 其中對於每一層j>0，Ej及Hj計算為： cos3] fsin^ tt; 1 [liij sin cos 巧 j!_%-!」， 其中E0為1且H〇為μ0，且其中對於每一層j20Γ^-1 kJ where for each layer j>0, Ej and Hj are calculated as: cos3] fsin^ tt; 1 [liij sin cos clever j!_%-!", where E0 is 1 and H〇 is μ0, and Which for each layer j20
Pj = (nj-i(kj + mj))-cos 中j 且 gj = 27i(nj-i(kj + mj)).tj.cos φ』/λ 其中η』為層j之折射率，kj為層j之消光係數，mj為增 加層j之消光係數之數量，tj為層j之厚度，φ」·為光至層 j之入射角，且λ為波長。 13 201249598 在另態樣中，產生參考光譜庫之方法包括以下步 驟：接收代表基板上之第一層堆疊之反射率之第一光 譜’第-堆疊包括第-層；接收代表基板上之第二層堆 疊之反射率之第一光错，第二層堆疊包括不位於第一堆 璧中之第二層；接收代表基板上之第三層堆疊之反射率 之第二光譜，第三層堆疊包括不位於第—堆疊且不位於 第一堆中之第二層；接收識別第一堆疊之第一複數個 不同貢獻百分比之使用者輸入及接收識別第二堆疊之第 二複數個不同貢獻百分比之使用者輸入；以及對於來自 第一複數個不同貢獻百分比之每一第一貢獻百分比及來 自第二複數個不同貢獻百分比之每一第二貢獻百分比， 根據第-光譜、第二光譜、第三光譜、第—貢獻百分比 及第二貢獻百分比計算參考光譜。 只細了包括以下特徵中之一或更多者。第二堆疊可包 括第層。第一堆登可由第一層組成，且第一層可為第 一堆疊之底層。第三堆疊可包括第一層及第二層，第一 層可為第三堆疊之底層，且第二層可介於第一層及第三 層之間。 在另—態樣中’控制研磨之方法包括以下步驟：根據 先刖之方法中之一者’產生參考光譜庫；研磨基板；在 研磨期間量測來自基板之光之一序列光譜；對於該序列 光譜之每一量測之光譜，發現最佳匹配參考光譜以產生 序列教佳匹配參考光谱；以及基於該序列最佳匹配參 考光增決定研磨終點或研磨速率之調整中之至少一者。 14 201249598 某些貫施可包括以下優點中之一或更多者。可迅速計 异參考光譜庫，該參考光譜庫跨過基板上之一或更多層 之折射率或消光係數之可能改變範圍。可迅速計算參考 光譜庫，該參考光譜庫跨過基板上之不同層堆疊之貢獻 之可能改變範圍。當不同批次或不同製造商引起折射率 或消光係數之改變時，產生之參考光譜庫可改良匹配演 算去之可罪性。因此，可改良偵測所欲之研磨終點之終 點系統之可葬'|±，且可減少晶圓内部及晶圓之間厚度的 不均勻性（WIWNU及WTWNU)。 【實施方式】 -種光學監視技術為量測研磨期間自基板反射之光之 光谱’及識別來自參考光譜庫之匹配參考光譜。在一些 實施中，匹配參考光譜提供-系列指數值及一函數（； 如，線）與-系列指數值擬合。函數至目標值之投影可 用於決定終點或改變研磨速率。 士上所豸㈤潛在問題為用於該等模型中之沈積層 之η值及k值視膜組成及膜沈積控制而㈣著不同客戶 及不同批次變化。舉例而言， 名如氧化矽、摻雜碳氧化 石夕、碳㈣、氮切、摻雜碳氮切或多^之一些層 趨向於產生η值及k值之變化。 ^ 评^之，由於沈積製程 中之製程條件’外表上由相同材 」何枓组成之均勻層可具有 變化之η值及k值。因此，即 1之各戶嚴格地控制客戶的 膜性質’在客戶之間亦存“值及k值的變化。 5 15 201249598 為解決此問題，可甚& 產生锼數個參考光譜，其中該複數 個參考光譜包括對 始仏令 、相门層之不同折射率值或消光系數 值所產生之參考井★並。.. 先°曰舉例而言，可儲存一組一或更多 個折射率函數及— 或更夕個消光係數函數。對於來 自該組折射率函數加糾 数之折射率函數及來自該組消光係數函 數之消光係數函數之备—έ 口致之母一組合，可計算參考光譜。 〇 述另問題為一些基板包括具有不同層堆疊 之區域。作為極其簡單之實例，—些區域可包括金屬層 上方之單-介電層’且其他區域可包括金屬層上方之兩 個介電層。當缺，太眘 晶 田…在貫際應用中可能存在更加複雜之層 堆疊。舉例而言，當在線後端製程中研磨基板時，基板 ^一些區域可包括暴露之金屬，其他區域可包括單一層 組，且另其他區域可包括多個垂直佈置之層組。每一： 組可對應於基板之金屬互連結構中之金屬層。舉例而 言，每一層组包括介電層（例如，低k介電）及姓刻終 止層（例如，碳化矽、氮化矽或碳氮化矽（SicN))。 在原位監視製程期間，將光束放置在基板上沒有受到 精確控制。所以光束將有時主要落在具有_個層堆疊之 區域上，及有時光束將主要落在具有不同層堆疊之區域 •^。簡而言之，對於來自基板上之每一不同層堆疊之光 譜之百分比貢獻可隨不同量測改變。然而，有可能產生 多個參考光譜，該多個參考光譜跨過不同層堆疊的貢獻 之可能變化範圍。Pj = (nj-i(kj + mj))-cos in j and gj = 27i(nj-i(kj + mj)).tj.cos φ』/λ where η′′ is the refractive index of layer j, kj is The extinction coefficient of layer j, mj is the number of extinction coefficients of layer j, tj is the thickness of layer j, φ"· is the incident angle of light to layer j, and λ is the wavelength. 13 201249598 In another aspect, a method of generating a reference spectral library includes the steps of: receiving a first spectrum representing a reflectivity of a first layer stack on a substrate; the first stack includes a first layer; and the second substrate is received on a representative substrate a first optical error of the reflectivity of the layer stack, the second layer stack comprising a second layer not located in the first stack; receiving a second spectrum representing the reflectivity of the third layer stack on the substrate, the third layer stack comprising a second layer that is not located in the first stack and not in the first stack; receives user input identifying the first plurality of different contribution percentages of the first stack and receives the second plurality of different contribution percentages identifying the second stack Input; and for each first contribution percentage from the first plurality of different contribution percentages and each second contribution percentage from the second plurality of different contribution percentages, according to the first spectrum, the second spectrum, the third spectrum, The first-contribution percentage and the second contribution percentage are calculated as reference spectra. Only one or more of the following features are included. The second stack can include the first layer. The first stack can be composed of a first layer and the first layer can be the bottom layer of the first stack. The third stack can include a first layer and a second layer, the first layer can be the bottom layer of the third stack, and the second layer can be between the first layer and the third layer. In another aspect, the method of controlling the grinding comprises the steps of: generating a reference spectrum library according to one of the methods of the prior art; grinding the substrate; measuring a sequence spectrum of the light from the substrate during the grinding; for the sequence Each measured spectrum of the spectrum is found to best match the reference spectrum to produce a sequence-matched reference spectrum; and at least one of the adjustment of the polishing endpoint or the polishing rate based on the best matching reference light gain of the sequence. 14 201249598 Some implementations may include one or more of the following advantages. The library of reference spectra can be quickly referenced, which spans the range of possible refractive indices or extinction coefficients of one or more layers on the substrate. The reference spectral library can be quickly calculated, which spans the range of possible variations in the contribution of different layer stacks on the substrate. When different batches or different manufacturers cause changes in the refractive index or extinction coefficient, the resulting reference spectral library can improve the plausibility of the matching algorithm. Therefore, the susceptibility of the end point system for detecting the desired polishing end point can be improved, and the thickness unevenness (WIWNU and WTWNU) between the inside of the wafer and the wafer can be reduced. [Embodiment] An optical monitoring technique is to measure the spectrum of light reflected from a substrate during polishing and to identify a matching reference spectrum from a reference spectral library. In some implementations, the matching reference spectrum provides a -Series index value and a function (; e.g., line) fitted to the -Series index value. The projection of the function to the target value can be used to determine the end point or change the grinding rate. (5) The potential problems are the η value and k value of the sedimentary layer used in these models, and the film composition and film deposition control (4) vary from customer to customer and from batch to batch. For example, layers such as yttrium oxide, doped carbon oxidized stone, carbon (tetra), nitrogen cut, doped carbon nitride cut or many tend to produce changes in η and k values. ^ As a result, the uniform layer composed of the same material in the deposition process, which is composed of the same material, may have varying values of η and k. Therefore, each household has a strict control over the customer's membrane properties. 'There are also changes in the value and k value between the customers. 5 15 201249598 To solve this problem, a number of reference spectra can be generated. The plurality of reference spectra include reference wells generated by different refractive index values or extinction coefficient values of the starting gates, phase gate layers, and.., for example, one set of one or more refractive indices may be stored. Function and - or an extinction coefficient function. For a combination of the refractive index function of the set of refractive index functions plus the correction function of the set of extinction coefficients and the extinction coefficient function of the set of extinction coefficient functions, a reference can be calculated The other problem is that some substrates include regions with different stacks of layers. As an extremely simple example, some regions may include a single-dielectric layer above the metal layer and other regions may include two layers above the metal layer. Electrical layer. When it is missing, Taishen Jingtian... There may be more complicated layer stacks in the continuous application. For example, when the substrate is polished in the online back-end process, some areas of the substrate may include The metal, other regions may comprise a single layer group, and other regions may comprise a plurality of vertically arranged layer groups. Each: group may correspond to a metal layer in a metal interconnect structure of the substrate. For example, each layer group Including a dielectric layer (eg, low-k dielectric) and a surname termination layer (eg, tantalum carbide, tantalum nitride, or tantalum carbonitride (SicN)). During the in-situ monitoring process, the beam is placed on the substrate. Subject to precise control, so the beam will sometimes fall primarily on areas with _ layer stacks, and sometimes the beam will fall primarily in areas with different layer stacks. ^ In short, for each from the substrate The percentage contribution of the spectra of the different layer stacks can vary with different measurements. However, it is possible to generate multiple reference spectra that span the range of possible variations in the contribution of different layer stacks.
S 16 201249598 另一問題在於對於一些基板而言，僅有小部分光能约 穿透基板上之最上層組。另外，來自最上層組之光比來 自第二層组及下層組之純少可能經散射及回到摘測器 以貢獻量敎光譜。@此，在計算理論產生之多堆疊參 考光譜中合理近似法為僅使用頂端兩個層組。 如H/ν所給出’折射率w與材料中之光速成反比， 其中《為折射率，，為真空中之光速且v為材料中之光 逮n意謂光在材料中緩慢行進^肖光係數免與吸 收係數成比例。大的&意謂材料具有強吸收性（入射光 強烈地衰減）^值ο意謂材料為完全透明的。 基板可包括第一層及第二層，第二層置於第一層上 方、。第-層可為介電質。第一層及第二層兩者均為至少 半透明的。第-層及-或更多個附加層（若存在）共同 在第二層下方提供層堆疊。 舉例而言，參閱第丨八圖，基板10可包括基本結構12, 例如，玻璃片或半導體晶圓，該基本結構可能進一步 具有傳導層或絕緣材料層。傳導層14 (例如，諸如銅、 鶴或紹之金屬）置於基本結構12上方。圖案化之下部第 一介電層18置於傳導層14上方，且圖案化之上部第二 介電層22置於下部介電層18上方。下部介電層18及上 部介電層22可為絕緣體，例如，諸如二氧化石夕之氧化物 或諸如摻雜碳二氧化石夕之低k材料（例如，B-k Diamond ( ^ | Applied Materials, Inc. ) ^ C〇ral™( ^ 17 201249598 自N〇vellusSystems，Inc ))。下部介電層i8及上部介電 層22可由相同材料或不同材料組成。 傳導層14及下部介電層18之間可選擇地放置純化層 16 ’例如’氮化石夕。下部介電層18及上部介電層以之 間可選擇地放置㈣終止層2〇,例如，介電材料（例如， 石反化石夕、氮化石夕或碳氮化石夕（SiCN))。阻障層％置於上 部介電層22上方且至少進入上部介電層22中之溝槽 2 ’該阻障層26之組成不同於下部介電層18及上部介 之組成。舉例而言’阻障層26可為金屬或金屬 :匕物，例如，氮化组或氮化鈦。上部介電層22及阻障 ;:6第-層及第二層之間可選擇地放置一或更多個附 口曰4’忒一或更多個附加層24由不同於第二 :之另一介電材料組成，例如，低k敷蓋材料（例如， 22正上7^旨（咖)形成之材料）。放置於上部介電層 枰内）的A且至少在藉由上部介電層22之圖案提供之溝 槽=為傳導材料28’例如，諸如銅、軸之金屬。 可且傳導材料28之間的層，包括阻障層％， =夠低的消光係數及/或為足夠 監㈣統之光。相反地，傳導層_傳導材 肩14及V足夠厚的且具有充足高的消光係數使得傳導 及傳導材料28對於來自光學監視系統的光為不透S 16 201249598 Another problem is that for some substrates, only a small portion of the light energy penetrates the uppermost layer on the substrate. In addition, the light from the uppermost group is less likely to be scattered and returned to the sifter to contribute to the 敎 spectrum than the pure ones from the second and lower groups. @这, A reasonable approximation in the multi-stack reference spectrum produced by the computational theory is to use only the top two layer groups. As H/ν gives, the refractive index w is inversely proportional to the speed of light in the material, where "for the refractive index, the speed of light in the vacuum and v is the light in the material. n means that the light travels slowly in the material. The light coefficient is not proportional to the absorption coefficient. Large & means that the material is strongly absorptive (the incident light is strongly attenuated). The value ο means that the material is completely transparent. The substrate may include a first layer and a second layer, and the second layer is disposed above the first layer. The first layer can be a dielectric. Both the first layer and the second layer are at least translucent. The first layer and/or more additional layers, if present, collectively provide a layer stack below the second layer. For example, referring to Fig. 8, the substrate 10 can include a basic structure 12, such as a glass sheet or a semiconductor wafer, which may further have a conductive layer or a layer of insulating material. A conductive layer 14 (e.g., a metal such as copper, crane, or shovel) is placed over the base structure 12. The patterned lower dielectric layer 18 is placed over the conductive layer 14 and the patterned upper second dielectric layer 22 is placed over the lower dielectric layer 18. The lower dielectric layer 18 and the upper dielectric layer 22 may be an insulator, such as, for example, an oxide of cerium oxide or a low-k material such as doped carbon dioxide (eg, Bk Diamond (^ | Applied Materials, Inc) . ) ^ C〇ralTM ( ^ 17 201249598 from N〇vellusSystems, Inc )). The lower dielectric layer i8 and the upper dielectric layer 22 may be composed of the same material or different materials. A purification layer 16' such as 'Nitrix is optionally placed between the conductive layer 14 and the lower dielectric layer 18. The lower dielectric layer 18 and the upper dielectric layer are optionally interposed between the (4) termination layer 2, for example, a dielectric material (e.g., stone counter-stone, nitrite or carbonitride (SiCN)). The barrier layer % is placed over the upper dielectric layer 22 and at least enters the trench 2' in the upper dielectric layer 22. The composition of the barrier layer 26 is different from the composition of the lower dielectric layer 18 and the upper dielectric layer. For example, the barrier layer 26 can be a metal or a metal: a germanium, such as a nitrided group or titanium nitride. Upper dielectric layer 22 and barrier; 6 optionally interposed between the first layer and the second layer, one or more attachment ports 4', one or more additional layers 24 are different from the second: Another dielectric material composition, for example, a low-k coating material (e.g., a material formed by 22). The A placed in the upper dielectric layer and at least in the pattern provided by the pattern of the upper dielectric layer 22 = a conductive material 28' such as a metal such as copper or a shaft. The layers between the conductive materials 28, including the barrier layer %, = low enough extinction coefficient and/or sufficient light to be supervised. Conversely, the conductive layer _ conductive material shoulders 14 and V are sufficiently thick and have a sufficiently high extinction coefficient such that the conductive and conductive material 28 is impervious to light from the optical monitoring system.
S 18 201249598 在一些實施中，待总廿 ★ s其他層有可能為第一層及第二 層，但上部介電；？。& 曰 棱供第一層且阻障層20提供第二 層。 化學機械研磨可用炉^ - 於將基板平坦化直至暴露第二層。 舉例而言，如第1B岡占 __ 51中所圖示，開始時研磨不透明傳 導材料…暴露非不透明第二層(例如，阻障傳 隨後’參閱第⑴圖’移除剩餘在第—層上方之第^ 之部分及研磨基板直至暴露第一層（例如，上部介電； 22)。另外，有時希望研磨第一層（例如，介電層 直至剩餘目標厚度或已經移除目標數量之材料。 圖至第1C圖之實財，平坦化後，剩餘在上部介電層 2 2之凸起圖案之_問少;兹11 口茱之間之傳導材料28之部分形成通孔 孔之類似物。 種研磨方法為在第—研磨塾上研磨傳導材料Μ至 V直至暴露第二層（例如，阻障声26 )。另冰 丨1早層26)。另外，可移除 層之厚度之部分’例如，在過度研磨步驟期間在第 :磨塾處移除。隨後將基板轉移至第二研磨塾，在此 ^全移除第二層（例％，阻障層26)及亦移除第—層 八例如，諸如低k介電質之上部介電層22)之厚度之部 hi:外，若在第一及第二層之間存在一或更多附加層 移^，敷蓋層）則可在第二研磨墊處以㈣研磨㈣ /于、4 —或更多附加層。 第2圖圖示研磨設備1〇〇之實例。研磨設備⑽包括 旋轉之蝶形平臺120，研磨墊110位於平臺120上。S 18 201249598 In some implementations, the other layers are likely to be the first layer and the second layer, but the upper layer is dielectric; . The < 棱 edge is provided for the first layer and the barrier layer 20 is provided with the second layer. Chemical mechanical polishing can be used to planarize the substrate until the second layer is exposed. For example, as illustrated in 1B Gangzhan__51, initially illuminate the opaque conductive material... expose the non-opaque second layer (eg, block the pass and then 'see figure (1)' to remove the remaining layer The upper portion and the substrate are polished until the first layer is exposed (eg, upper dielectric; 22). Additionally, it is sometimes desirable to polish the first layer (eg, the dielectric layer until the remaining target thickness or the target number has been removed) The material is as shown in Fig. 1C. After flattening, the remaining pattern of the upper dielectric layer 2 2 is less; the portion of the conductive material 28 between the ports 11 is similar to the through hole. The method of grinding is to grind the conductive material Μ to V on the first polishing pad until the second layer is exposed (for example, the barrier sound 26). Another hail 1 early layer 26). In addition, the thickness of the removable layer The portion 'for example, is removed at the:grinding during the overgrinding step. The substrate is then transferred to the second abrasive crucible where the second layer (example %, barrier layer 26) is removed and also removed The first layer VIII, for example, the thickness of the dielectric layer 22 such as the low-k dielectric upper layer 22): In addition, if one or more additional layers are present between the first and second layers, the cover layer may be (4) ground (4) /, 4 - or more additional layers at the second polishing pad. Fig. 2 illustrates an example of a grinding apparatus 1A. The grinding apparatus (10) includes a rotating butterfly platform 120 on which the polishing pad 110 is located.
S 19 201249598 平臺為可操作的以圍繞軸125旋轉。舉例而言，電動機 121可轉動驅動軸124以旋轉平臺12(^研磨墊11〇可為 雙層研磨墊，具有外部研磨層112及較軟的背托層114。 研磨設備100彳包括4 13〇卩分配諸如聚料之研磨液 至研磨墊110上至該墊。研磨設備亦可包括研磨墊 調即益以磨光研磨墊110從而維持研磨墊ιι〇處於連貫 的研磨狀態。 研磨設们00包括一或更多個承載頭14〇。每一承載 頭140為可操作的以固持基板丨〇緊靠研磨墊11 〇。每一 承載頭140可單獨控制與每—各別基板相關聯之研磨參 數（例如，愿力）。 洋。之每一承載頭140可包括擋圈142以保持基板 位於撓性膜144下方。每一承載頭i 4〇亦包括複數個 可獨立控制之由膜界定之可加壓腔室，例如，三個腔室 1 4 6 c該等腔至可獨立施加可控制壓力至撓性膜 144上之關聯區域148a_M8e及由此在基板1()上施加可 控制壓力（參見第3圖參閱第3圖，中心區域⑽ 可為實質上®形的’且其純域刚捕。可為圍繞中 品域8a之同、環形區域。儘管為了便於說明第2 圖及第3圖中僅圖干了-加 . 惶Cj不了二個腔室，但是可存在一個或兩 個腔室，或四個或四個以上腔室（例如，五個腔室）。 回到第2圖，每一承載頭140自支撐結構150(例如， 旋轉料架）懸掛，及藉由驅動車由152連接至承載頭旋轉電 動機154使侍承载頭可圍繞轴155旋轉。每一承載頭 20 201249598 可選擇地橫向擺動 名絲鼓概加 1 π Λ .S 19 201249598 The platform is operable to rotate about the axis 125. For example, the motor 121 can rotate the drive shaft 124 to rotate the platform 12 (the polishing pad 11 can be a two-layer polishing pad having an outer abrasive layer 112 and a softer backing layer 114. The grinding apparatus 100 includes 4 13 〇 The slurry is dispensed, such as a slurry, onto the polishing pad 110 to the pad. The polishing apparatus may also include a polishing pad to smooth the polishing pad 110 to maintain the polishing pad in a continuous state of grinding. One or more carrier heads 14. Each carrier head 140 is operable to hold the substrate against the polishing pad 11. Each carrier head 140 can individually control the grinding parameters associated with each of the individual substrates. Each of the carrier heads 140 can include a retaining ring 142 to hold the substrate under the flexible membrane 144. Each carrier head 4 〇 also includes a plurality of independently controllable membrane-definable Pressurizing chambers, for example, three chambers 1 4 6 c to independently apply controllable pressure to associated regions 148a-M8e on flexible membrane 144 and thereby exert controllable pressure on substrate 1 (see Figure 3, see Figure 3, center area (10) It can be a substantially TM-shaped 'and its pure domain just caught. It can be the same, annular area around the middle product domain 8a. Although for the sake of convenience, only the graphs in Figure 2 and Figure 3 are dry-plus. 惶Cj can't Two chambers, but there may be one or two chambers, or four or more chambers (eg, five chambers). Returning to Figure 2, each carrier head 140 is self-supporting structure 150 (eg The rotating rack is suspended and connected to the carrier head rotating motor 154 by a drive vehicle 152 to rotate the carrier head about the shaft 155. Each carrier head 20 201249598 optionally laterally swings the nominal drum drum plus 1 π Λ .
研磨設備亦包括原位光學監視系統160 (例如，光譜The grinding apparatus also includes an in situ optical monitoring system 160 (eg, spectroscopy)
否調整研磨速率或決定對於研磨速率之調整。藉由包括 118提供穿過研磨墊 豐窗口 118可支撐於 内’但固體窗口 11 8 孔（即’貫穿墊之洞）或固體窗口 u 8 之光學通道。儘管在一些實施中固體窗 平宜12 0上及大出至研磨塾中之孔内， 可緊固至研磨塾1 1 〇，例如，作為填滿研磨塾中之孔之 插塞’例如’經模製或黏附緊固至研磨墊。 光學監視系統1 60可包括光源i 62、光偵測器丨64及 電路166，該電路166用於在遠端控制器19〇 (例如’電 腦）與光源162及光偵測器164之間發送及接收訊號。 —或更多個光纖可用於自光源162傳送光至研磨整中的 光學通道，及傳送自基板1〇反射之光至偵測器164。舉 例而言’分叉光纖170可用於自光源162傳送光至基板 10及將光傳回至偵測器1 6 4。.分又光纖包括定位鄰近於 21 201249598 光子通道之主線172及兩個支線174及17ό ,該兩個支 線1 74及1 76分別連接至光源1 62及偵測器1 64。 在 2只施中，平臺之頂表面可包括凹口 128，該凹 128内安裴光學頭168，該光學頭固持分又光纖 之主線172之一端。光學頭168可包括機構以調整主線 172之頂端及固體窗口 118之間的垂直距離。 電路166之輸出可為數位電子訊號，該數位電子訊號 穿過紅轉耦合器i29 (例如，驅動軸i 24中之滑環）到 達光學監視系統之控制器190。類似地，回應於自控制 器190穿過旋轉轉合器129至光學監視系、统16〇之數位 電子汛號中之控制命令，彳打開或關閉光源。或者，電 路166可藉由無線訊號與控制器19〇通訊。 光源162可為可操作的以發射白光。在一個實施中， 發射之白光包括具有波長為2⑽綱奈卡之光。適當光 源為亂燈或亂采燈。 光偵測器164可為分光計。分光計為用於量測電磁波 譜之部分上光強度之光學儀器。適#分光計為光拇分光 計。分光計之典型輸出為作為波長（或頻率）之函數之 光強度。 如上所述，光源162及光偵測器！ 64可連接至計算裝 置（例如，控制器190 )，該計算裝置為可操作的以控制 光源162及光偵測器164之操作及接收光源162及光偵 測器164之訊號。計算裝置可包括位於研磨設備附近之Do not adjust the polishing rate or determine the adjustment for the polishing rate. The optical channel through which the window 118 can be supported through the polishing pad 118 can be supported by the inner window 118 (or the hole through the pad) or the solid window u 8 . Although in some implementations the solid window is preferably 120 and larger than the hole in the grinding crucible, it can be fastened to the grinding crucible 1 1 , for example, as a plug filling the hole in the grinding crucible 'for example' Molded or adhered to the polishing pad. The optical monitoring system 1 60 can include a light source i 62, a light detector 丨 64, and a circuit 166 for transmitting between the remote controller 19 (eg, 'computer) and the light source 162 and the light detector 164. And receiving signals. - or more fibers can be used to transmit light from source 162 to the optical channel in the polishing process and to the light reflected from substrate 1 to detector 164. For example, the bifurcated fiber 170 can be used to transfer light from the source 162 to the substrate 10 and to pass the light back to the detector 164. The split fiber also includes a main line 172 adjacent to the 21 201249598 photonic channel and two legs 174 and 17A, which are connected to the light source 1 62 and the detector 1 64, respectively. In two applications, the top surface of the platform may include a recess 128 in which the optical head 168 is mounted, the optical head holding one end of the main line 172 of the optical fiber. The optical head 168 can include mechanisms to adjust the vertical distance between the top end of the main line 172 and the solid window 118. The output of circuit 166 can be a digital electronic signal that passes through red coupler i29 (e.g., a slip ring in drive shaft i 24) to controller 190 of the optical monitoring system. Similarly, in response to a control command from the controller 190 passing through the rotary coupler 129 to the digital monitor of the optical monitoring system, the light source is turned on or off. Alternatively, circuit 166 can communicate with controller 19 via a wireless signal. Light source 162 can be operable to emit white light. In one implementation, the emitted white light comprises light having a wavelength of 2 (10). The appropriate light source is a random light or a random light. The photodetector 164 can be a spectrometer. A spectrometer is an optical instrument used to measure the intensity of light on a portion of an electromagnetic spectrum. The #分光计 is an optical thumb spectrometer. The typical output of a spectrometer is the intensity of light as a function of wavelength (or frequency). As mentioned above, the light source 162 and the light detector! 64 can be coupled to a computing device (e.g., controller 190) that is operable to control operation of light source 162 and light detector 164 and to receive signals from light source 162 and light detector 164. The computing device can include a device located adjacent to the grinding device
S 22 201249598 微處理器，例如’可程式化電腦。就控制而言，計算裝 置可（例如）使光源之啟動與平臺12〇之旋轉同步。 在一些實施中，原位監視系統160之光源162及#測 器164=裝在平4120内及隨平4觀走轉。在此情況 下，平臺之移動將引起感測器掃描跨過每—基板。詳言 之’當平臺120旋轉時，控制器19〇可引起光源162就 在光學通道經過基板1G下方之前開始發射—系列閃光 及就在光學通道經過基板1〇下方之後結束發射一系列 閃光。或者，計算裝置可引起光源162就在每—基板1〇 經過光學通道上方之前開始連續發射光及就在每一基板 1〇經過光學通道上方之後結束連續發射光。在任一情況 下，來自偵測器之訊號在取樣週期期間整合從而以取樣 頻率產生光譜量測。 在操作中，控制器190可接收（例如）傳遞資訊之訊 號，該資訊描述對於光源之特定閃光或偵測器之時間 框，藉由光偵測器接收之光之光譜。因此，此光譜為研 磨期間原位量測之光譜。 如第4圖中所圖示，若偵測器安裝在平臺中，由於平 臺之旋轉（藉由箭頭204圖示）’當窗口 108在承載頭下 方行進時，則以取樣頻率產生光譜量測之光學監視系統 將引起在位置2 01處進行光譜量測，位置2 〇 1在橫越夷 板10之弧形中。舉例而言，點201a-201k中之每一者代 表監視系統之光譜量測之位置（點之數量為說明性的； 視取樣頻率而定可進行比所圖示之量測更多或更少之量 23 201249598 測)。可選擇取樣頻率使得窗口 1〇8每次拂掠時收集介於 五個至一十個之間的光譜。舉例而言，取樣週期可在3 毫秒與1 0 0毫秒之間。 如圖所示，在平臺之一個旋轉期間，在基板10上自不 :半徑獲得光譜。亦即，自靠近基板10之中心之位置獲 得一光°θ且自罪近基板10之邊緣之位置獲得一些光 譜。因此’對於跨過基板之光學監視系統之任何給定掃 描基於寺序、電動機編碼器資訊及基板之邊緣及/或擔 圈之光學偵測’控制器19G可自掃描計算每—量測之光 譜之半徑位置（相對於經掃描之基板之巾ο研磨系統 亦可包括旋轉位置感測器’例如，附著於平臺之邊緣之 凸緣，該旋轉位置感測器將經過固定光學斷續器，以提 供額外資料用於決定哪個基板及量測之光譜在該基板上 之位置控制β因此可將各種量測之光譜與基板心及 土板10b上之可控制區域148b]48e (參見第2圖）相 關聯。在-些實施中’光譜之量測之時間可用作代替半 徑位置之準確計算。 在平臺之多個旋轉期間，對於每一區域，可隨時間獲 得-序列光譜。不受限於任何特定理論，由於最外層之 厚度又化胃研磨進仃時（例如，在平臺之多個旋轉期 間，不在跨過基板之單—拂掠期間），自基板iq反射之 光之光…貞變目此產生隨時間改變之一序列光譜。此 外’㈣光譜由層堆疊之特定厚度顯示。 24 201249598 在一些實施中，可程式化控制器（例如，計算裝置） 以將量測之光譜與多個參考光譜比較及決定哪個表考光 譜提供最佳匹配。詳言之，可程式化控制器以將來自每 區域之-序列置測之光譜的每一光譜與多個參考光譜 比較從而產生每一區域之一序列最佳匹配參考光譜。 如本文中所使用，參考光譜為研磨基板之前產生之預 先定義之光譜。假定實際研磨速率遵循預計研磨速率， 則參考光譜可與代表研磨製程中預計出現光譜之時間之 值具有預先定義（亦即’在研磨操作之前定義）之關聯 拴或者或另外，參考光譜可與諸如 板性質之值具有預先定義之關聯性。 I之基 可.5 ^驗產生參考光譜，例如，藉由量測來自測試基 板之光譜（例如，測試基板具有已知之最初層厚度)。舉 例而言’為產生複數個參考光譜，使用將在裝置晶圓研 磨期間所使用之相同研磨參數研磨裝設基板，同時收集 :序列光譜。對於每一光譜，記錄代表研磨製程中收集 光::之時間之值。舉例而言，值可為所經過之時間或若 干平臺旋轉。可過度研磨基板，亦即，基板經研磨超過 所欲之厚度’使得完成目標厚度時可獲得自基板反射之 光之光譜。S 22 201249598 Microprocessors, such as 'programmable computers. In terms of control, the computing device can, for example, synchronize the activation of the light source with the rotation of the platform 12A. In some implementations, the source 162 and # 164 of the in-situ monitoring system 160 are mounted within the flat 4120 and tracked with the flat. In this case, movement of the platform will cause the sensor to scan across each substrate. In detail, when the platform 120 is rotated, the controller 19 can cause the light source 162 to start emitting just before the optical path passes under the substrate 1G - a series of flashes and the end of the series of flashes just after the optical path passes under the substrate 1〇. Alternatively, the computing device can cause the light source 162 to begin to continuously emit light just before each substrate 1 〇 passes over the optical channel and to terminate the continuous emission light just after each substrate passes over the optical channel. In either case, the signal from the detector is integrated during the sampling period to produce a spectral measurement at the sampling frequency. In operation, controller 190 can receive, for example, a signal conveying information describing the spectrum of light received by the photodetector for a particular flash or detector time frame of the light source. Therefore, this spectrum is the spectrum measured in situ during grinding. As illustrated in Figure 4, if the detector is mounted in the platform, due to the rotation of the platform (illustrated by arrow 204), when the window 108 is traveling under the carrier head, the spectral measurement is generated at the sampling frequency. The optical monitoring system will cause spectral measurements at position 201, where position 2 〇1 is in the arc across the panel 10. For example, each of the points 201a-201k represents the location of the spectral measurement of the monitoring system (the number of points is illustrative; depending on the sampling frequency, more or less than the measured measurement may be performed The amount of 23 201249598 measured). The sampling frequency can be selected such that the window 1〇8 collects between five and ten spectra each time it is swept. For example, the sampling period can be between 3 milliseconds and 100 milliseconds. As shown, the spectrum is obtained on the substrate 10 from a radius during one rotation of the platform. That is, a light θ is obtained from a position near the center of the substrate 10 and some spectrum is obtained from the position near the edge of the substrate 10. Therefore, the controller 19G can automatically calculate the spectrum of each measurement for any given scan of the optical monitoring system across the substrate based on the temple sequence, the motor encoder information, and the edge of the substrate and/or the optical detection of the substrate. The radial position (with respect to the scanned substrate, the grinding system can also include a rotational position sensor 'eg, a flange attached to the edge of the platform, the rotational position sensor will pass through the fixed optical interrupter to Additional information is provided for determining which substrate and the position of the measured spectrum are controlled on the substrate. Therefore, the various measured spectra can be combined with the controllable regions 148b] 48e on the substrate core and the soil plate 10b (see Figure 2). Correlation. In some implementations, the time of the measurement of the spectrum can be used as an accurate calculation instead of the radius position. During each rotation of the platform, the sequence spectrum can be obtained over time for each region. Any particular theory, since the thickness of the outermost layer is again etched into the sputum (for example, during multiple rotations of the platform, not during the single-pushing across the substrate), reflected from the substrate iq The light of light... This produces a sequence of spectra that changes over time. Furthermore, the '(four) spectrum is shown by the specific thickness of the layer stack. 24 201249598 In some implementations, a programmable controller (eg, a computing device) will The measured spectrum is compared to a plurality of reference spectra and determines which reference spectrum provides the best match. In particular, the controller can be programmed to map each spectrum of the spectrum from each region-sequence to multiple references Spectral comparisons result in a sequence of best matching reference spectra for each region. As used herein, the reference spectrum is a predefined spectrum generated prior to polishing the substrate. Assuming the actual polishing rate follows the expected polishing rate, the reference spectrum can be represented The value of the time at which the spectrum is expected to appear in the polishing process has a pre-defined (i.e., 'defined before the grinding operation') or, alternatively, the reference spectrum can have a predefined correlation with values such as plate properties. .5 ^ Generate a reference spectrum, for example, by measuring the spectrum from the test substrate (for example, the test substrate has The initial layer thickness). For example, to generate a plurality of reference spectra, the substrate is ground using the same grinding parameters used during the wafer polishing of the device, while collecting: sequence spectra. For each spectrum, the record represents the polishing. The value of the time during which the light is collected in the process: for example, the value may be the time elapsed or the rotation of several platforms. The substrate may be over-polished, that is, the substrate is ground to a desired thickness, so that the target thickness can be achieved. A spectrum of light reflected from the substrate is obtained.
S 為了將每一光譜與基板性質（例如，最外層之厚度） 之值相關聯，可在計量站於研磨前量測最初光譜及與產 基板具有相同圖案之r裝設」基板之性質。亦可使用 相同計量站或不同計量站於研磨後量測最後光譜及性 25 201249598 質。可藉由内插法（例如’基於量測測試基板之光譜所 工過之時間之線性内插）決定介於最初光譜及最後光譜 之間的光譜之性質。 除了憑經驗決定外，參考光错中之—些或全部可根據 :論計算，例如，使用基板層之光學模型計算。舉例而 -’光：模型可用於計算給定外層厚度d之參考光譜。 例如n由假定以均勾研磨速率移除外層，可叶管代表 在研磨製財將收集參考光譜之時間之值。舉例 可藉由假定開始厚度DG及㈣研磨速率 : 定參考光譜之時間T—卿作為另一實= =仃線性内插，該線性内插介於研磨前之厚度m及研 間後^厚度D2(或計量站處量測之其他厚度）之量測時 曰1 ΤΙ、T2之間，該研磨箭夕庖甴八 厚度D1及研磨後之厚度 站處量測之其他厚度）基於用於光學模型之 予度 D(Ts = T2-T1*(D1-D)/(D1-D2)。 因：：：實施中’軟體可用於自動計算多個參考光譜。 因為存在將進入之其祝夕丁麻 土 .之厚度的變化，故製造商 取下層“列如，多個下層）中 圍及厚度增量。對於下層之厚 人者之异度1巳 算參考光譜。對於上層之每 、= ’軟體將計 譜。 母尽度，可計算多個參考光 舉例而言，對於第 學堆疊可依序包括位 14 )、鈍化層、下部低 1B圖中所圖示之結構的研磨，光 於底。卩之金屬層（例如，傳導層 k介電層、蝕刻終止層、上部低kS To correlate the value of each spectrum to the properties of the substrate (e.g., the thickness of the outermost layer), the properties of the initial spectrum and the r-mounted substrate having the same pattern as the substrate can be measured at the metering station prior to polishing. It is also possible to measure the final spectrum and properties after grinding using the same metering station or different metering stations. The nature of the spectrum between the initial spectrum and the final spectrum can be determined by interpolation (e.g., linear interpolation based on the time taken to measure the spectrum of the test substrate). In addition to being determined empirically, some or all of the reference optical errors can be calculated according to the theory, for example, using an optical model of the substrate layer. By way of example - 'light: the model can be used to calculate a reference spectrum for a given outer layer thickness d. For example, n is assumed to remove the outer layer at a uniform grind rate, and the leaf tube represents the value of the time at which the reference spectrum is collected during the grind. For example, by assuming the starting thickness DG and (4) the polishing rate: the time T-qing of the reference spectrum is taken as another real ==仃 linear interpolation, the linear interpolation is between the thickness m before grinding and the thickness D2 after the grinding. (or other thickness measured at the metering station) Measured between 曰1 ΤΙ and T2, the thickness of the grinding arrow D8 and other thicknesses measured at the thickness station after grinding) based on the optical model The degree of D (Ts = T2-T1*(D1-D)/(D1-D2). Because::: The implementation of 'software can be used to automatically calculate multiple reference spectra. Because there will be a wish to enter it. The change in the thickness of the soil. Therefore, the manufacturer takes the middle and thickness increments of the lower layer "column, multiple lower layers." For the difference of the lower layer, the reference spectrum is calculated for the difference of the upper layer. For the upper layer, = 'soft body For example, for example, for the first stack, for example, the polishing of the structure shown in Figure 14), the passivation layer, and the lower low 1B pattern can be performed on the bottom. a metal layer of germanium (eg, conductive layer k dielectric layer, etch stop layer, upper low k
S 26 201249598 介電層、TEOS層、阻障層及水層（代表研磨液，光穿 過該研磨液可到達）。在一個實例中，出於計算參考光譜 之目的，阻障層可以10 A之增量在3〇〇人及35〇人之= 變化，TEOS層可以50 A之增量在480〇人及52〇〇人之 間變化，且上部低k介電頂層可以20A之增量在18〇〇入 及2200 A之間變化。對於層之厚度之每—組合，計瞀象 考光譜。使用該等自由度，將計算9*6*21 = 1丨34個表考 光譜。然而，對於每一層，其他範圍及增量為可能的。 為計算參考光譜’可使用以下光學模型。薄膜堆疊之 頂層p之反射率rstack可計算為： 其中βρ +代表將進入之光束之電磁場強度且Ep-代表將 離開之光束之電磁場強度。 值Ep+及Ep_可計算為： Ερ + = (Ερ+Ηρ/μρ)/2 Ερ- = (Ερ-Ηρ/μΡ)/2 〇 可使用轉換矩陣法根據下層中之場Ε及場Η計算任意 層j中之％ Ε及場Η。因此’在膚〇，1，...，ρ-1，ρ (其中 層〇為底層且層ρ為最外層）之堆疊中，對於給定層」 > 0，S 26 201249598 Dielectric layer, TEOS layer, barrier layer and water layer (representing the polishing liquid through which light can pass). In one example, for the purpose of calculating the reference spectrum, the barrier layer can be varied by 3 A in increments of 3 及 and 35 =, and the TEOS layer can be increased by 50 A in 480 及 and 52 〇. The change between the monks and the upper low-k dielectric top layer can vary between 18 in and 2200 A in increments of 20A. For each combination of the thickness of the layer, the spectrum is measured. Using these degrees of freedom, 9*6*21 = 1丨34 reference spectra will be calculated. However, for each layer, other ranges and increments are possible. The following optical model can be used to calculate the reference spectrum. The reflectivity rstack of the top layer p of the film stack can be calculated as: where βρ + represents the electromagnetic field strength of the beam that will enter and Ep- represents the electromagnetic field strength of the beam that will exit. The values Ep+ and Ep_ can be calculated as: Ερ + = (Ερ+Ηρ/μρ)/2 Ερ- = (Ερ-Ηρ/μΡ)/2 〇 can be calculated using the transformation matrix method according to the field Η and field 下 in the lower layer % Ε and field 层 in layer j. Therefore, in the stack of skin, 1, ..., ρ-1, ρ (where 〇 is the bottom layer and layer ρ is the outermost layer), for a given layer > 0,
Ej及Ej and
Hj可計算為： i 1 -stn 9j 「6叫 cos gj 巧1 = cos8j ΗΛ = sin S： 27 201249598 其中〜 = (n广ikj).c°s Φ』且 gj=2Ti(n广ikj).' c〇s φ』/λ，其 中nj為層j之折 射率，kj為層j之消光係數，^為層』 之厚度，屯為光至層』之入射角 馬波長。對於堆 疊中之底層（亦即，層』 。且 一ik0).cos Φο °可根據科學文獻冰t I a , 于文獻“每—層之折射率n及消光係數 k’且折射率n及消光係數k可為波長之 奈爾（Snell)定律計算入射角φ。 函數。可根據史 層之厚度t可根據藉由使用者輸 厚度增量來計算，例如，對於…入=厚度範圍及，丄，.··，對於 t.^T tj —TMINj+k*TiNCj，其中 Τλ xt rp MINj及1^為層j之厚度範圍 之上限及下限且TlNcj為層j之厚 度增量。可對於層之厚 度值之每一組合重複計算。 此技術之潛在優點為快速產生大量參考光譜庫 量參考光譜庫可對應於基板上之層之厚度之不同二 從而改良發現良好匹配參考光错之可能性及改良光_ 視系統之準確性及可靠性 s 28 201249598 舉例而言，自第ic圖中所圖示之基板反射之光強度 可計算為： Γ£,ι S< cos σ., — sm〇f - Ui ， z sin ^~Hj can be calculated as: i 1 -stn 9j "6 is called cos gj Q1 = cos8j ΗΛ = sin S: 27 201249598 where ~ = (n wide ikj).c°s Φ" and gj=2Ti(n Guangikj). ' c〇s φ 』 /λ, where nj is the refractive index of layer j, kj is the extinction coefficient of layer j, ^ is the thickness of the layer, and 屯 is the incident angle of the light to the layer. For the bottom layer in the stack (ie, layer) and one ik0).cos Φο ° can be based on the scientific literature ice t I a , in the literature "refractive index n and extinction coefficient k' per layer and refractive index n and extinction coefficient k can be wavelength The Snell's law calculates the angle of incidence φ. function. The thickness t of the history layer can be calculated according to the thickness increment by the user, for example, for the range of thickness=, 丄, . . . , for t.^T tj —TMINj+k*TiNCj, where Τλ xt rp MINj and 1^ are the upper and lower limits of the thickness range of layer j and TlNcj is the thickness increment of layer j. The calculation can be repeated for each combination of the thickness values of the layers. A potential advantage of this technique is the rapid generation of a large number of reference spectral libraries. The reference spectral library can correspond to the difference in thickness of the layers on the substrate to improve the likelihood of finding a good matching reference optical error and to improve the accuracy and reliability of the optical system. Sex s 28 201249598 For example, the intensity of light reflected from the substrate illustrated in the ic diagram can be calculated as: ,£, ι S< cos σ., — sm〇f - Ui , z sin ^~
Mz ^in gz cosg2 cos ^3 7- sin c〇sgt .^3 sin 5¾ 50s53 stn _ιμ± sin gt cosg1 其中之值及μ*之值取決於基板10之最外層（例如， 上部介電層22 (例如，低k材料））之厚度、折射率及 消光係數；之值及以之值取決於下層（例如，蝕刻終 止層20 (例如，SiCN))之厚度、折射率及消光係數； g2之值及μ2之值取決於另一下層（例如，下部介電層18) 之厚度、折射率及消光係數；gi之值及μι之值取決於另 一下層（例如，鈍化層（例如，SiN))之厚度、折射率 及消光係數；以及μ0之值取決於底層（例如，傳導層Μ (例如’銅））之折射率及消光係數。 接著反射率rstack可計算為：Mz ^in gz cosg2 cos ^3 7- sin c〇sgt .^3 sin 53⁄4 50s53 stn _ιμ± sin gt cosg1 where the value and the value of μ* depend on the outermost layer of the substrate 10 (for example, the upper dielectric layer 22 ( For example, the thickness, refractive index, and extinction coefficient of the low-k material)) and the value depend on the thickness, refractive index, and extinction coefficient of the lower layer (eg, etch stop layer 20 (eg, SiCN)); value of g2 And the value of μ2 depends on the thickness, refractive index and extinction coefficient of the other lower layer (for example, the lower dielectric layer 18); the value of gi and the value of μ are dependent on another lower layer (for example, a passivation layer (for example, SiN)) The thickness, refractive index, and extinction coefficient; and the value of μ0 depend on the refractive index and extinction coefficient of the underlying layer (e.g., conductive layer Μ (e.g., 'copper)). Then the reflectivity rstack can be calculated as:
Rstack = —^ 儘管未圖示，亦可在光學模型中說明在基板上方存在 水層（代表研磨液，光經由該研磨液可到達）。 上述基板及相關聯之光學堆疊僅為層之一個可能組 合，及許多其他組合為可能的。舉例而言，上述光學堆 疊在光學堆4之底料使用料層，料導層對於線後 端製程中之基板而言將為典型的 '然而，在線前端製程Rstack = —^ Although not shown, it can be stated in the optical model that there is a layer of water above the substrate (representing the slurry through which light can be reached). The above substrate and associated optical stack are only one possible combination of layers, and many other combinations are possible. For example, the optical stack is stacked on the bottom layer of the optical stack 4, and the material guide layer will be typical for the substrate in the post-end process. However, the online front-end process
S 29 201249598 中’或若傳導層At丨 s為透明材料，則光學堆疊之底部可為半 >體晶圓（例如，矽）0作.另本" ^ ^ )作為另一貫例，一些基板可不包 括下部介電層。 除了層厚度的樂化少 化之外’光學模型可包括光學堆疊中 之一或更多個層之拼μ、玄， 祈射率及/或消光係數之變化。該一或 更多個層可包括下層及/或上層。該一或更多個層可包括 乳化石夕層、摻雜碳氧切層、碳切層、氮切層、摻 雜碳氮化矽層及/啖吝曰 、 次夕日日矽層。視基板上之層之組成及沈 積方法而定，可自呈古 目具有較兩折射率或消光係數之層之基 =行-些光譜量測，而其他光譜量測可自具有較低折 射率或消光係數之層之基板進行。 在些實把中，軟體可用於接收使用者輸入以識別— 組—或更多個折射率函數 数及/或一組一或更多個消光係In S 29 201249598, or if the conductive layer At丨s is a transparent material, the bottom of the optical stack may be a half > body wafer (for example, 矽) 0. Another " ^ ^ ) as another example, some The substrate may not include a lower dielectric layer. In addition to the reduced thickness of the layer thickness, the optical model may include variations in the spelling, sharpness, and/or extinction coefficient of one or more of the layers in the optical stack. The one or more layers may include a lower layer and/or an upper layer. The one or more layers may include an emulsifying stone layer, a doped carbon oxide layer, a carbon cut layer, a nitrogen cut layer, a doped carbonitride layer and/or a second day solar layer. Depending on the composition of the layer on the substrate and the deposition method, it can be measured from the basis of the layer with the two refractive index or the extinction coefficient; the other spectral measurements can be derived from the lower refractive index. Or a substrate of a layer of extinction coefficient. In some implementations, the software can be used to receive user input to identify - a set - or more refractive index functions and / or a set of one or more extinction systems
數函數。折射率函數可A思+ U 7為層之材料提供折射率作為波長 之函數。類似地，消氺 先係數函數可為層之材料提供消光 糸作為波長之函數。在基板之間折射率有變化之情況 :，複數個不同折射率函數可用於產生參考光譜。類似 地，在基板之間消光係數有 啕又化之清況下，複數個消光 係數函數可用於產生參考 1尤D曰舉例而言，對於來自該 組折射率函數之折射率函袁 , 数及來自該組消光係數函數之 消光係數函數之每一組合， 私體可计异參考光譜。 不同折射率函數可為登g ,, 見—般折射率函數之變體。舉 例而5 ’ 一般折射率函數 马/皮長之函數及一或更多個Number function. The refractive index function can be as a function of wavelength for the material of the layer. Similarly, the annihilation coefficient function provides the extinction 糸 as a function of wavelength for the material of the layer. In the case where the refractive index varies between substrates: a plurality of different refractive index functions can be used to generate the reference spectrum. Similarly, in the case where the extinction coefficient between the substrates is degraded, a plurality of extinction coefficient functions can be used to generate the reference 1 曰, for example, for the refractive index function from the set of refractive index functions, Each combination of the extinction coefficient functions from the set of extinction coefficient functions can be calculated by the private reference spectrum. Different refractive index functions can be g, see variants of the general refractive index function. For example, 5' general refractive index function, horse/skin length function and one or more
額外係數之函數，且不同拼I 折射率函數可組成一或更多個 s 30 201249598 值 係數之不同值。使用者（例如，半導體製造商）可設定 一或更多係m舉例而言，對於特定係數，使用者 可藉由輪入下限值、上限值及值增量或若干總值來設定 類似地，不同消光係數函數可為常見一般消光係數函 數之變體。舉例而言，—般消光係數函數可為波長之函 數及一或更多個額外係數之函數，且不同消光係數函數 可組成-或更多個係數之不同值。使用者（例如，半導 體製造商）m或更多係數之值。舉例而言，對於 特定係數，使用者可藉由輸人下限值、上限值及值增量 或若干總值來設定值。使用者亦可定義幾組常數之值及 係數之值。因此’使用使用者定義之值可計算不同組消 光係數函數。 在一些實施中，使用柯西等式可將折射率函數及消光 係數函數模型化。柯西模型可用於對於可見光波長範圍 中之透明介電質將η(λ)及kQ)模型化。由下式得出： 11⑵参+ § 其中’ An、Bn、Cn、Ak、Bk由使用者規定，且Ck為常 數（4〇0〇A)。若不存在吸收，則Ak=0。第ls圖為^西產 生折射率函數之實例。柯西模型可用於僅將η(λ)或僅將 k〇)模型化。 為了產生複數個光譜’建立基線柯西模型及隨後料由 使用者定義邊線允許An、Bn、Cn、Ak及Bk中之—咬多 31 201249598 者「浮動亦即，在用於產生函數之係數之多個值上， 可重複計算光譜Rstack ’例如，在An、Bn、cn、Ak及A function of the extra coefficients, and different spelling index functions may form different values for one or more s 30 201249598 value coefficients. A user (eg, a semiconductor manufacturer) can set one or more lines. For example, for a particular coefficient, the user can set a similar value by entering a lower limit, an upper limit, and a value increment or a number of total values. Ground, different extinction coefficient functions can be variants of the common general extinction coefficient function. For example, the general extinction coefficient function can be a function of a function of wavelength and one or more additional coefficients, and different extinction coefficient functions can constitute different values of - or more coefficients. The user (for example, the semiconductor manufacturer) has a value of m or more coefficients. For example, for a particular coefficient, the user can set the value by inputting a lower limit, an upper limit, and a value increment or a number of total values. The user can also define several sets of constant values and values of the coefficients. Therefore, different sets of extinction coefficient functions can be calculated using user-defined values. In some implementations, the refractive index function and the extinction coefficient function can be modeled using the Cauchy equation. The Cauchy model can be used to model η(λ) and kQ) for transparent dielectrics in the visible wavelength range. It is obtained by the following formula: 11(2) 参 + § where 'An, Bn, Cn, Ak, Bk are specified by the user, and Ck is a constant number (4〇0〇A). If there is no absorption, Ak=0. The ls picture is an example of the refractive index function produced by the West. The Cauchy model can be used to model only η(λ) or just k〇). In order to generate a plurality of spectra, the baseline Cauchy model is established and then the user defines the edges to allow the An, Bn, Cn, Ak, and Bk to be bite. 2012-04598 "Floating, that is, the coefficient used to generate the function. On multiple values, the spectrum Rstack can be repeatedly calculated 'for example, in An, Bn, cn, Ak and
Bk之多個值上。舉例而言，An可在丨4〇與丨5〇之間變 化’例如，變化增量為0.02。 此技術之潛在優點為產生參考光譜，該等參考光譜可 對應於基板上之層中之不同折射率或不同消光係數，從 而改良發現良好匹配參考光譜之可能性及改良光學監視 系統之準確性及可靠性。 η(λ)及k(X)可隨不同使用者係數輸入浮動。舉例而 言，可存在η(λι) ’ n(M為使用者以A"、&丨及^規定 之折射率；以及可存在k⑹，收）為使用者以Akl、Bk丨 及Ch規定之消光係數。 第19圖為圖示使用n值浮動模型（藉由浮動&及1) 之厚度追蹤之更好光譜擬合之實例。在此特定實例中，k 膜之參數定義為An=1.435至1.495、Bn=〇 〇〇3至〇 〇〇7， 其中堆疊厚度變化（依序）為：介電質薄膜·入至雇 A、飯刻終止層仍入至525入、介電質薄膜2400 A及 蝕刻終止| 500 A。第19圖圖示介電層厚度與研磨時間 曲線圖。橫轴表示以秒為單位之研磨時間；以及垂直軸 表不以A為單位之厚度最佳匹配。該圖表示最佳匹配之 二個例子。每-例子位置恰好位於其他例子之頂部上。Multiple values of Bk. For example, An can vary between 丨4〇 and 丨5〇', for example, with an increment of 0.02. A potential advantage of this technique is the generation of reference spectra that can correspond to different refractive indices or different extinction coefficients in layers on the substrate, thereby improving the likelihood of finding a well-matched reference spectrum and improving the accuracy of the optical monitoring system and reliability. η(λ) and k(X) can be floated with different user coefficient inputs. For example, there may be η(λι) 'n (M is the refractive index specified by the user with A", & and ^; and k(6) may be present, which is specified by the user in Akl, Bk, and Ch. Extinction coefficient. Figure 19 is a diagram illustrating an example of a better spectral fit using the n-valued floating model (by floating& and 1) thickness tracking. In this particular example, the parameters of the k film are defined as An=1.435 to 1.495, Bn=〇〇〇3 to 〇〇〇7, wherein the stack thickness variation (in order) is: dielectric film·into employment A, The meal stop layer is still in the 525-in, dielectric film 2400 A and etch stop | 500 A. Fig. 19 is a graph showing the thickness of the dielectric layer and the polishing time. The horizontal axis represents the grinding time in seconds; and the vertical axis does not have the best thickness matching in A. This figure shows two examples of the best match. The per-example position is located just on top of the other examples.
在更-般之實施中’產生用於控制研磨之參考光譜庫 =步驟可包括以下步驟：儲存多個層之層堆疊之光學模 ，接丈定義折射率係數之一組不同第一值之使用者輸A 32 201249598 7對於每-第-值，計算折射率函數以產生若干折射 率函數；以及對於每-折射率函數，基於折射率函數、 消光係數函數及第一層之坌 -.+ 矛層之弟—厚度使用光學模型計算參 考光譜。 在更般之態樣中，—組—或更多個折❹@ ϋ & 括複數個不同折射率函數。類似mu㈣ 先係數函數可包括複數個不同消光係數函數。 除了層厚度之變化之外，光學模型可包括金屬層之光 5“獻之變化。亦即，視製造中之晶片上之圖案而定， 可在具有高金屬（例如，來自溝槽中之金屬材料28)濃 度之區域中進行—些光譜量測，而其他光譜量測可在且 有較低金屬濃度之區域中進行。 八 使用者對軟體之輸入可進一步包括基板之第—層之若 干不同厚度值。在該等不同厚度值中，存在至少第一厚 =因二，使用光學模.型，不同折射率函數、消光係 Μ旱度值之每-組組合將產生參考光譜以用於光 譜庫中。 、尤 添加至光5f庫中f光譜RUBRARY可計算為：In a more general implementation, the step of generating a reference spectrum library for controlling the polishing step may include the steps of: storing the optical modes of the layer stack of the plurality of layers, and defining the use of one of the first values of the refractive index coefficients. Transducing A 32 201249598 7 For each-first value, calculate the refractive index function to produce several refractive index functions; and for each-refractive index function, based on the refractive index function, the extinction coefficient function, and the first layer of 坌-.+ spear The brother of the layer—thickness uses an optical model to calculate the reference spectrum. In a more general case, the group—or more than one ❹@ϋ & includes a plurality of different refractive index functions. Similar to the mu (four) pre-coefficient function may include a plurality of different extinction coefficient functions. In addition to variations in layer thickness, the optical model may include variations in the light of the metal layer. That is, depending on the pattern on the wafer in the fabrication, it may have a high metal (eg, metal from the trench) Material 28) some spectral measurements are made in the concentration region, while other spectral measurements can be performed in regions with lower metal concentrations. Eight user input to the software can further include several differences in the first layer of the substrate. Thickness value. Among the different thickness values, there is at least a first thickness = factor 2, using an optical mode, a different refractive index function, and a combination of extinction system dryness values will produce a reference spectrum for the spectrum In the library, especially added to the light 5f library, the f-spectrum RUBRARY can be calculated as:
^^7 ^ " RSTACJ<i ^ (1 - > ^stacS2\ RU3SAS 1 /、中RstACKi為第一光譜，Rstack2為第二光譜， RREFERENCE為第-堆疊及第二堆疊之底層之光譜，且X 為第一堆疊之百分比貢獻。 曰^^7 ^ "RSTACJ<i ^ (1 - > ^stacS2\ RU3SAS 1 /, RstACKi is the first spectrum, Rstack2 is the second spectrum, RREFERENCE is the spectrum of the bottom layer of the first stack and the second stack, and X is the percentage contribution of the first stack.
S 33 201249598 k：；STAl ^STACKt (1 -^ - F) - RSTACK2]S 33 201249598 k:;STAl ^STACKt (1 -^ - F) - RSTACK2]
U3RhF.Y RLIBrary可為多個堆疊模型之組合。舉例而言，可存 在Rstacki及RSTACK2 ’ Rstacki為最高堆疊（包含cAp、 介電、阻障及銅基板）之光譜貢獻，RSTACK2為兩個最高 堆豐（该兩個最高堆疊為來自Rstacki之介電及阻障基 板加上駐留於該介電及阻障基板下方之介電、阻障及銅 土板）之光v曰1貝獻。故Rlibrary之計算看起來類似： 二八中X + Y< 1，RSTACK1為第一光譜，RSTACK2為第二光 。曰，Rmetal為第三光譜，Rreference為堆疊之底層之光 °曰，且x為第一堆疊之百分比貢獻’且Y為金屬之百分 比貢獻。 、在一些實施例中’例如，若金屬層14及金屬材料28 為相同材料（例如，銅）’則RBASELINE及RMe⑷為相同 先t (例如，銅之光譜）。可在X之多個值上重複計算 光譜RLIBRARY。舉例而言，X可以間隔0.2在〇.〇盥i 〇 之間改變。繼續第1B圖中所圖示之堆疊之實例，使用 該專自由度’將計算9,2降刪個參考光譜。此技 ^潛在f點為產生參考光譜，㈣參考光譜可對應於 二：里測之點中之不同金屬濃度’從而改良發現良 Γ'考光譜之可能性及改良光學監視系統之準讀性 及可靠性。 二了：變層厚度之外，光學模型可包括改變金屬層之 貝獻。亦即’視製造中之晶片上之圖案而定，可在 34 201249598 具有南金屬（例如，來自溝槽中之金屬材料28)濃度之 區域申進行一些光譜量測，而其他光譜量測可在具有較 低金屬漠度之區域中進行。由於藉由折射率、消光係數 及厚度定義材料層，故對於給定之材料，存在表徵該材 料之光學性質之折射率及消光係數之每一函數，可量 測、憑經驗決定或模型化該函數。 故對於rlibrary之計算看起來類似： p 〜 1 nLIBI^AF.y ^ ΓΚ R 一 V · D , r R^-SSBNCE L Λ - Rstac：<1 ^ (l ^X_yy,： Rstac^ i 巧其中X+Y< 1，Rstacki為第—光譜，RSTA⑴為第二光 °曰RMETAL為第三光譜，Rreference為堆疊之底層之光 譜’且X為第—堆疊之百分比貢獻，且Y為金屬之百分 〜1 , /日 Η汉贫屬材料2 8 為相同材料（例如，銅）， 、REFER£nce 及 Rmetal 為相 同光譜（例如，銅之来摄） 則之九。曰）。可在X及Y之多個值上重 複計算光譜RLIBRARY。舉例而A v ^ g +例而§ ，X可以間隔〇 j在 〇·〇與1，〇之間改變且γ可 J 乂間隔0.1在0,0與1.0之間 改變。此技術之潛在優點為 座生參考先譜’該等參考光 譜可對應於基板上之量測 ,ώ , 』之點中之不同金屬濃度，從而 改良發現良好匹配參考光级 ..芩尤°曰之可能性及改良光學監視系 、·先之準確性及可靠性。 在一些貫施中，將自單— 饰掠（例如，跨過區域或跨 過整個基板之拂掠）所收 〃之多個量測之光譜均分。由 35 201249598 :自較大區域取樣平均光譜’故平均光譜具有來自各種 1之百分比貢獻之較緊密分佈。此允許使用者將用於 算中之百刀比貝獻限制至更狹小之範圍。舉例而言， 及Υ可以間隔0.02在0,2之範圍内改變。 軟體可接收識別金屬… 屬層之後數個不同金屬貢獻百分比 之使用者輪入，該接必 接收之步驟可包括以下步驟：接收識 別第一堆疊之第—動旦丁 數里不同貢獻百分比之使用者輸入及 接收識別第二堆疊第_ 且又·弟一數®不同貢獻百分比之使用者 輪入。可根據第—旦 ▲办 数里不同貝獻百分比及第二數量不同 貝獻百分比計算溢射_ 谡數個不同金屬貢獻百分比。 在一些實施中， μ 之層及/或可人為^ 十异可忽略弟二層組下方 為3加一些層之消光係數以代表光到達 彼·#層之減少之可能性。 在一些實施中， 弟一光譜之計算可包括計算堆疊參考 汉 STACK1 : ^U3RhF.Y RLIBrary can be a combination of multiple stacked models. For example, there may be Rstacki and RSTACK2 'Rstacki's spectral contribution for the highest stack (including cAp, dielectric, barrier, and copper substrate), and RSTACK2 is the two highest stacks (the two highest stacks are dielectrics from Rstacki) And the barrier substrate plus the dielectric, barrier and copper plate material that resides under the dielectric and barrier substrate. Therefore, the calculation of Rlibrary looks similar: X + Y < 1, RSTACK1 is the first spectrum, and RSTACK2 is the second light.曰, Rmetal is the third spectrum, Rreference is the light of the bottom layer of the stack, and x is the percentage contribution of the first stack' and Y is the percentage contribution of the metal. In some embodiments, for example, if metal layer 14 and metal material 28 are of the same material (e.g., copper), then RBASELINE and RMe(4) are the same first t (e.g., the spectrum of copper). The spectrum RLIBRARY can be repeated over multiple values of X. For example, X can vary between 〇.〇盥i 间隔 at intervals of 0.2. Continuing with the example of the stack illustrated in Figure 1B, the specified degree of freedom' is used to calculate the 9,2 drop reference spectrum. The potential point f is the generation of the reference spectrum, and (4) the reference spectrum can correspond to the difference between the different metal concentrations in the two points: thereby improving the possibility of finding the good spectrum and improving the readability of the optical monitoring system. reliability. Second, in addition to the thickness of the layer, the optical model can include changes to the metal layer. That is, depending on the pattern on the wafer in the fabrication, some spectral measurements can be made in the region of the concentration of the south metal (for example, the metal material 28 from the trench) at 34 201249598, while other spectral measurements can be made at Performed in areas with lower metal indifference. Since the material layer is defined by the refractive index, extinction coefficient, and thickness, for a given material, there is a function of each of the refractive index and extinction coefficient that characterizes the optical properties of the material, which can be measured, empirically determined, or modeled. . Therefore, the calculation for rlibrary looks similar: p ~ 1 nLIBI^AF.y ^ ΓΚ R - V · D , r R^-SSBNCE L Λ - Rstac: <1 ^ (l ^X_yy,: Rstac^ i X+Y< 1, Rstacki is the first spectrum, RSTA(1) is the second light, RMETAL is the third spectrum, Rreference is the spectrum of the bottom layer of the stack' and X is the percentage contribution of the first-stack, and Y is the percentage of metal ~1, / Η Η 贫 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 。 。 。 。 。 。 。 。 。 。 。 。 The spectrum RLIBRARY can be repeatedly calculated over multiple values of X and Y. For example, A v ^ g + and § , X can be changed between 〇·〇 and 1, 〇, and γ can be changed between 0, 0 and 1.0. The potential advantage of this technique is that the reference reference spectrum can correspond to the different metal concentrations in the measurement, ώ, 』 on the substrate, thereby improving the discovery of a well-matched reference light level. The possibility and improvement of the optical monitoring system, the accuracy and reliability of the first. In some implementations, the spectra of multiple measurements taken from a single-pruning (e.g., across a region or across a whole substrate) are equally divided. From 35 201249598: Sampling the average spectrum from a larger area', the average spectrum has a tighter distribution of percentage contributions from various 1's. This allows the user to limit the range of knives used in the calculation to a narrower range. For example, Υ can vary by 0.02 in the range of 0, 2. The software can receive the user wheeling of the plurality of different metal contribution percentages after the genus layer, and the step of receiving the received data can include the following steps: receiving the use of the different contribution percentages in the first-moving number of the first stack The user enters and receives a user rounding that identifies the second stack _ and _ _ _ _ _ _ _. The percentage of different metal contributions can be calculated based on the percentage of different beatings and the second quantity in the number of times. In some implementations, the layer of μ and/or may be artificially negligible. The extinction coefficient of the layer plus 3 layers below the two-layer group is representative of the possibility of a decrease in the light reaching the layer. In some implementations, the calculation of the spectrum of a brother may include calculating a stack reference Han STACK1 : ^
R -ϊηΐΓΛΓΙ — Mr .¾ [^1 UrJ > υ ’ Ε』严Hj計算為： cos&j — sin . % 'fh sm a: c, 其中E〇為i且 ft H〇為μ〇 ,且其中對於每一層】纟〇， Mj = (n；-ikj)-cos φ. Η ^J gj~2?r(n广iki).Vc〇s φ』·/λ，其中 nj 為 s 36 201249598 層j之折射率，岣為層』之消光係 m h為層]之厚度， (^為光至層j之入射角，且人為波長。 類似地，第 譜之計算彳，隹疊反射率 Bp — RSrACK: = --^R -ϊηΐΓΛΓΙ — Mr .3⁄4 [^1 UrJ > υ ' Ε 严 H H H H H H co co co co % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % For each layer 纟〇, Mj = (n; -ikj)-cos φ. Η ^J gj~2?r(n广iki).Vc〇s φ』·/λ, where nj is s 36 201249598 The refractive index of j, the thickness of the extinction system mh of the layer is the thickness of the layer, (^ is the incident angle of light to the layer j, and the artificial wavelength. Similarly, the calculation of the spectrum 彳, the stack reflectance Bp - RSrACK : = --^
R STACK2 其中對於每一層 赝j > Ο，Ej及Hj計算為： 内7一 f ·R STACK2 where for each layer 赝j > Ο, Ej and Hj are calculated as: 内 7一 f ·
COS u； sin3j sin gj cos g,其中 為 & h π i且_H〇為μ〇，且其中 £:]， 對於母一層j会〇， 其中〜為層j之折射率，心為潛j之消光係數，mj為增 加幻之消光係數之數量，tj為幻之厚度，祕至層 J·之入射角，且λ為波長。 A -jt,b j./. v & ，第一堆疊可包括頂端介電層及蝕刻終 止層’例如，ρ & 反夕、氫化矽或碳氮化矽（SiCN)。如第 2 0圖中戶斤κι - t 不可存在頂層組之不同反射貢獻。參閱第 圖圖不光進行至層堆疊中。光1810、光182〇及光 1830代表谁Λ n e 及反射光穿過不同層。光181〇自上層金 屬（M7)反射屮卞 对了出朿，光1820自第一層組（M6上方之層） 37 201249598 反射且光1830自第二層組（河5上方之層）反射。由於 M7、M6及M5中存在金屬線，故存在極低可能性：由 光學監視系統照射之位置201將包括大量自M5下方之 層反射之光。因此，在光學模型中可忽略該等層（例如， 模型將假定金屬層M5為所有堆疊之底層），或Rstack2 可包括所有該等層之影響，從而有效地將副下方之層 作為單-實體處理’以實現決定不同百分比貢獻之目的 (但是可能調整消光係數以代表藉由散射引起的自下層 之減少之反射’如上所述）。當然，帛2()圖僅為示例性 的，可存在不同數量金屬層且阻隔可位於不同金屬層。 為了計算參考光譜，電腦可接收多個個別光譜。舉例 而言，可接收代表基板上之第—層堆疊之反射率之第一 光譜，第一層堆叠句；楚 m i括第一層。可接收代表基板上之第 二層堆疊之反射率之第二光譜，第二層堆疊包括不位於 第-堆疊内之第：層（但是包括第—層）。此外，可接收 代表基板上之第三層堆疊之反射率之第三光譜，第三声 堆疊包括不位於第一堆疊且不位於第二堆疊内之第： 層。使用者（例如，半導體製造操作者）可輸入該等所 收集之堆疊光譜之不同貢獻百分比以產生參考光譜庫， 可根據第一光譜、第-你 弟一忐4及第三光譜，第一貢獻百分 比及第二貢獻百分比計算該參考光譜庫。 ，在。κ施中’光反射组件可經模型化為三個不同模 型°舉例而言’對於鋼貢獻（諸如自頂層銅線反射之幻 可使用理論銅反射光譜。在—些實施中，根據水層取值 38 201249598 之已知折射率值及消光系數值可用於計算銅反射組件。 對於頂層組貝獻，可將光譜向下模型化至第二金屬 層該頂層組貢獻為自正研磨之頂層組反射之光。在— 二只細例中，儿全移除敷蓋層及研磨頂端介電層至給定 厚度。在此情況下堆疊可包括：水、TE〇s敷蓋層、推 雜反氧化矽介電層、碳化矽蝕刻終止層塊及銅（基板）。 α十·^板型可忽、略TE〇S ’因為將完全移除。換雜碳 氧化矽介電層將在模型中具有自最小值至最大值之厚度 範圍’該厚度11圍代表研磨範圍。碳化石夕姓刻終止層通 常將具有標稱厚度及可藉由使用者規定該魏絲刻終 止層以用於預計之下層改變之範圍。 對於多堆疊貢獻，光含有來自剩餘下層（包括頂層） 之反射。因此總反射率為銅反射率、頂層反射率及多堆 s：反射率之線性組合。舉例而言，總反射率等於每一層 、.且反射率之貝獻之百分比總和^使用者能規定標稱銅貢 獻、頂層貢獻及改變範圍’例如，藉由輸入最大值、最 小值及步驟間隔值。COS u; sin3j sin gj cos g, where & h π i and _H〇 is μ〇, and where £:], for the parent layer j will be 〇, where ~ is the refractive index of layer j, the heart is the potential j The extinction coefficient, mj is the number of increasing extinction extinction coefficients, tj is the thickness of the magic, the angle of incidence to the layer J·, and λ is the wavelength. A -jt, b j. /. v & , the first stack may include a top dielectric layer and an etch stop layer 'e.g., ρ & 反, hydrogen hydride or tantalum carbonitride (SiCN). For example, in Figure 20, the household κι - t may not have different reflection contributions of the top group. Refer to the figure to not only proceed to the layer stack. Light 1810, light 182, and light 1830 represent who Λ n e and reflected light pass through the different layers. Light 181 is reflected from the upper metal (M7). The light 1820 is reflected from the first layer (layer above M6) 37 201249598 and the light 1830 is reflected from the second layer (layer above the river 5). Since there are metal lines in M7, M6 and M5, there is a very low probability that the position 201 illuminated by the optical monitoring system will include a large amount of light reflected from the layer below M5. Therefore, the layers can be ignored in the optical model (for example, the model will assume that the metal layer M5 is the bottom layer of all stacks), or Rstack2 can include the effects of all of the layers, effectively making the layer below the sub-layer a single-entity Process 'to achieve the goal of determining different percentage contributions (but it is possible to adjust the extinction coefficient to represent the reflection from the lower layer caused by scattering' as described above). Of course, the 帛2() diagram is merely exemplary, and there may be different numbers of metal layers and the barriers may be located in different metal layers. To calculate the reference spectrum, the computer can receive multiple individual spectra. For example, a first spectrum representing the reflectivity of the first layer stack on the substrate can be received, the first layer stacked sentence; the first layer is included. A second spectrum representative of the reflectivity of the second layer stack on the substrate can be received, the second layer stack including the first layer (but including the first layer) not within the first stack. Additionally, a third spectrum representative of the reflectivity of the third layer stack on the substrate can be received, the third sound stack including a first layer that is not located in the first stack and not in the second stack. A user (eg, a semiconductor manufacturing operator) can input a different contribution percentage of the collected stacked spectra to generate a reference spectral library, which can be based on the first spectrum, the first-fourth and the third spectrum, the first contribution The reference spectrum library is calculated as a percentage and a second contribution percentage. ,in. The κ Shizhong 'light-reflecting component can be modeled into three different models. For example, 'for steel contribution (such as the theoretical copper reflection spectrum from the reflection of the top copper wire. In some implementations, according to the water layer The known refractive index value and extinction coefficient value of value 38 201249598 can be used to calculate the copper reflective component. For the top layer group, the spectrum can be modeled down to the second metal layer. The top layer group contributes to the top group reflection from the positive grinding. In the second example, the cover layer and the top dielectric layer are completely removed to a given thickness. In this case, the stack may include: water, TE〇s coating, push-and-reverse oxidation矽 Dielectric layer, ruthenium carbide etch stop layer block and copper (substrate). α 十·^ plate type can be neglected, slightly TE〇S 'because it will be completely removed. The carbon-doped yttrium oxide dielectric layer will have in the model The thickness range from the minimum to the maximum 'this thickness 11' represents the grinding range. The carbonized stone etched layer will typically have a nominal thickness and can be specified by the user to predict the underlying layer. The scope of change. For multi-stacking The light contains reflections from the remaining lower layers (including the top layer). Thus the total reflectivity is a linear combination of copper reflectivity, top layer reflectivity, and multi-stack s: reflectance. For example, the total reflectance is equal to each layer, and the reflection The sum of the percentages of the rate ^ user can specify the nominal copper contribution, top contribution and range of change 'for example, by entering the maximum, minimum and step interval values.
S 另外，可需要方法來說明Γ散射」。當光在堆疊中進一 步向下行進時，由於下層t之散射，故較少光會反射回 來。因此下部低k介電層及阻障層將對光譜具有較少影 響，僅僅因為該等層更加向下且該等層中存在之銅線將 阻隔-些反射之光返回。可使用允許將額外消光係數添 加至彼層之使用中之消光系數值之經驗模型。額外消光 係數可為使用者規定之等式，該等式有效增加τ層之消 39 201249598 光0 在什异模型巾，只要當分別處理頂層時將頂層模型 化，那麼.出現模型化誤差之空間將更少。若模型化整個 ^層堆疊’則計算結果將更複雜及更易出現誤差。因此 藉由分別及不同地處理堆疊，可得到更好的計算結果用 於產生基於模型的光譜庫。舉例而言，最後光譜可為多 堆疊光譜之下層部分、頂層光譜之頂層部分及銅光譜之 頂層鋼部分之總和’該銅光譜之頂層銅部分等於自整個 部分中減去前兩個部分之剩餘部分。 對於#類型之基板（例如，一些層結構及晶片圖 案），上述用於基於光學模型產生參考光譜庫之技術可為 充分的。然而’對於一些類型之基板，基於此光學模型 之參考光譜不對應於憑經驗所量測之光譜。不受限於任 何特定理論，當將附加層添加至基板上之堆疊時，增加 光的政射（例如’自基板上之不同圖案化金屬層散射 簡而言之，當金屬層數量增加時，來自基板上之下層的 光變得不太可能會被反射進人光纖及到達偵測器。 在一些實施中，為模擬藉由增加金屬層數量引起之散 射，可在光學模型中使用改變之消光係數用於計算參考 光譜。改變之消光係數比層之材料之自然消光係數更 大。靠近晶圓之層^肖光係數所增加之數量可更大。 舉例而5 ’在以上等式中’ μ’』及g、可分別替代術語… 及gj，其中μ」及g，」·可計算為：S In addition, methods may be needed to illustrate Γ scattering. As the light travels further down the stack, less light is reflected back due to the scattering of the lower layer t. Therefore, the lower low-k dielectric layer and barrier layer will have less impact on the spectrum, simply because the layers are more downward and the copper lines present in the layers will block some of the reflected light back. An empirical model that allows the addition of additional extinction coefficients to the values of the extinction coefficients used in the other layer can be used. The extra extinction coefficient can be defined by the user, and the equation effectively increases the τ layer. 39 201249598 Light 0 In a different model towel, as long as the top layer is modeled when the top layer is processed separately, then there is room for modeling error. Will be less. If the entire stack of layers is modeled, the calculation results will be more complicated and more error-prone. Therefore, by processing the stack separately and differently, better calculation results can be obtained for generating a model-based spectral library. For example, the final spectrum can be the sum of the layer portion below the multi-stack spectrum, the top portion of the top spectrum, and the top portion of the copper spectrum. The top copper portion of the copper spectrum is equal to the remainder of the first two portions subtracted from the entire portion. section. For a #type substrate (e.g., some layer structures and wafer patterns), the techniques described above for generating a reference spectral library based on an optical model may be sufficient. However, for some types of substrates, the reference spectrum based on this optical model does not correspond to the spectrum measured empirically. Without being bound by any particular theory, when additional layers are added to the stack on the substrate, increasing the political policy of light (eg 'scattering from different patterned metal layers on the substrate, in brief, when the number of metal layers increases, Light from the lower layer on the substrate becomes less likely to be reflected into the human fiber and into the detector. In some implementations, to simulate the scattering caused by increasing the number of metal layers, a modified extinction can be used in the optical model. The coefficient is used to calculate the reference spectrum. The extinction coefficient of the change is larger than the natural extinction coefficient of the material of the layer. The amount of the optical coefficient closer to the layer of the wafer can be increased. For example, 5 'in the above equation ' μ '』 and g, can replace the term... and gj, respectively, where μ" and g," can be calculated as:
S 40 201249598 从 j=(nj—i(kj + mj))-cos <J)j g」 Φ]/λ 其中叫為增加層j之消光係數之數量。大體而言，叫 等於或大於0, 可到彡i。對於接近堆疊之頂端之層， 叫可為小的（例如’0)6對於更深之層，叫可為較大的 (例如’ 0.2、0.4或〇‘6)。當』減少時，數量％可單調 地增加。數量m」·可為波長之函數，例如，對於特定；， 在較長之波長處叫可更大或在較短之波長處叫可更曰大。 參閲第5圖及第6圖’量測之光譜3〇〇 (參見第$圖） 可與來自-或更多個庫31〇之參考光譜32〇(參見第6 圖）相比較。如本文所使用，參考光譜庫為代表享有共 同性質之基板之參考光譜之集合。然而，在單一光譜庫 中共同享有之性質可跨過多個參考光譜庫改變。㈣而 兩個㈣庫可包括參考光譜，料參考光譜代表且 有兩個不同下層厚度之基板1於給定參考光譜庫，1 層厚度改變，而不是其他因素（諸如晶圓圖案之差異、 度之差異或層組成之差異)’可主要地對光 之差異負責。 不同庫3 10之參考光譜320可藉由研磨且 其 性質（例如，下層厚度…成）之I、有不_ X飞層··且成；之多個「裝設」基板 201249598 及收集上述光瑨產生；來自一個裝設基板之光譜可提供 第一光譜庫且來自具有不同下層厚度之另一基板之光譜 可提供第一光譜庫。或者或另外，不同庫之參考光譜可 =據理論計算，例如，可使用具有下層之光學模型計算 第-光譜庫之光譜，該下層具有第一厚度；及可使用具 有下層之光學模型計算第二光譜庫之光譜，該下層具有 個不同厚度。舉例而言，此揭示案使用銅基板用於產 生光S普庫及隨後用於光譜量測。 在一些實施中，每一參考光譜320分配有指數值330。 大體而言，每一光譜庫310可包括許多參考光譜320(例 如，預計基板研磨時間期間之每—平臺旋轉之—或更多 個（例如’正好一個）參考光譜）。此指數別可為代表 研：製程中預計觀察參考光譜32〇之時間之值（例如， 光4可經索引使得每—光譜在特^光譜庫具有唯 才日數值。可實施索引使得指數值以量測測試基板之 序。當研磨進行時，可選擇指數值以單調地 改……，增加或減少)。特定而言，可選擇參考光譜 之扣數值使得指數值形成平臺旋轉之時n H θ 函數（假㈣㈣料❹M 或數1之線性 所愿逑羊遵循用於在光譜庫中產生參考光雄 之楨型或測試基板之研磨速率 曰 若干平臺旋轉成比例(例如相 -, 相荨）’在该平臺旋轉處量 、^丨忒基板之參考光譜或參 — /号尤》曰將出現在光學模型 * m值可為整數。指數可代表預計之平S 40 201249598 From j=(nj−i(kj + mj))-cos <J)j g” Φ]/λ where is called the number of extinction coefficients of layer j. In general, called equal to or greater than 0, you can go to 彡i. For layers near the top of the stack, it can be small (e.g., '0) 6 for deeper layers, which can be larger (e.g., '0.2, 0.4, or 〇 '6). When 』 decreases, the quantity % can increase monotonically. The number m" can be a function of wavelength, for example, for a particular wavelength; it can be said to be larger at longer wavelengths or more powerful at shorter wavelengths. Refer to Figure 5 and Figure 6 for the measured spectrum 3〇〇 (see Figure $). It can be compared to the reference spectrum 32〇 from the - or more banks 31〇 (see Figure 6). As used herein, a reference spectral library is a collection of reference spectra representing substrates that share common properties. However, the properties shared in a single spectral library can vary across multiple reference spectral libraries. (4) The two (four) banks may include a reference spectrum, and the substrate reference spectrum represents two substrates of different underlying thicknesses. In a given reference spectrum library, the thickness of one layer changes, rather than other factors (such as differences in wafer patterns, degrees). The difference or the difference in layer composition) can be primarily responsible for the difference in light. The reference spectrum 320 of the different banks 3 10 can be ground by grinding and having the properties (for example, the thickness of the lower layer...), and the plurality of "mounting" substrates 201249598 and collecting the light. The spectrum from one of the mounting substrates provides a first library of spectra and the spectrum from another substrate having a different underlying layer provides a first library of spectra. Alternatively or additionally, the reference spectra of the different libraries may be calculated according to theory, for example, the spectrum of the first-spectrum library may be calculated using an optical model having a lower layer having a first thickness; and the second model may be used to calculate a second The spectrum of the spectral library, the lower layer has a different thickness. For example, this disclosure uses a copper substrate for the production of a light S-base and subsequently for spectrometry. In some implementations, each reference spectrum 320 is assigned an index value of 330. In general, each spectral library 310 can include a plurality of reference spectra 320 (e.g., each time a substrate is rotated during the substrate polishing time - or more (e.g., exactly one) reference spectrum). This index can be representative of the value of the time during which the reference spectrum is expected to be observed in the process (for example, light 4 can be indexed such that each spectrum has a unique date value in the spectral library. The index can be implemented such that the index value is The order of the test substrate is measured. When the grinding is performed, the index value can be selected to monotonically change, increase or decrease. In particular, the deduction value of the reference spectrum can be selected such that the exponent value forms the n H θ function when the platform rotates (false (four) (four) material ❹ M or the linearity of the number 1 is desired to follow the type used to generate the reference glory in the spectral library. Or the polishing rate of the test substrate is proportional to the rotation of several platforms (eg phase-phase, phase 荨) 'the amount of rotation at the platform, the reference spectrum of the substrate or the reference - / 尤 曰 曰 will appear in the optical model * m The value can be an integer. The index can represent the expected flat
S -疋轉，在該平臺旋轉處將出現相關聯之光譜。 42 201249598 參考光譜及與參考光譜關聯之指數值可儲存在參考庫 中。舉例而言，每一參考光譜320及與每一參考光譜32〇 關聯之指數值330可儲存在資料庫35〇之記錄34〇 _。 可在研磨設備之計算裝置之記憶體中實施參考光譜之參 考庫之資料庫350。 如上所述，對於每一基板之每一區域，基於該序列量 測之光譜或彼區域及基板，可程式化控制器19〇以產生 序列最佳匹配光譜。可藉由將量測之光譜與來自特定 庫之參考光譜相比較來決定最佳匹配參考光譜。 在一些實施中，可藉由對於每一參考光譜計算量測之 光w與參考光譜之平方差之和來決定最佳匹配參考光 譜。具f最低平方差之和之參考光譜具有最佳擬合。用 於^現最锃匹配參考光譜之其他技術（例如，最低絕對 差之和）為可能的。 在二κ把中，可藉由使用平方差之和以外的匹配技 術來決定最佳匹配參考光譜。在一個實施中，對於每一 參考光譜，計算量測之光譜及參考光譜之間的互相關， 及认擇/、有最大相關之參考光譜作為匹配參考光譜。互 相關之潛在優點為對於光譜之橫向位移不太敏感，且因 此可打層厚度變化不太敏感。為了執行互相關，當參 考光譜相對於量測之光譜移動時，可使用「零」填充量 '則之光°曰之4端及尾端以提供資料從而與參考光譜對 比或者，可使用與量測之光譜之前沿處之值相等之值 真充里測之光譜之前端，及可使用與量測之光譜之後沿 43 201249598 處之值相等之值填充量測之光譜之後端。快速傅立葉 (Fourier)變換可用於增加匹配技術之實時應用之互相 關之計算速度。 在另—實施中’歐氏向量距離之和，例如， ϋ-1/(λα-λΙ))·[Σ 至 λ1?|ΐΜ(λ)2_Ικ(λ) 2丨]，其中 ^ 至乩為 計算求和之波長’ ΐΜ(λ)為量測之光譜，且。（”為參考光 譜。在另一實施中’對於每一參考光譜，將導數差求和 (例如，ϋ=1/(λ&·λΐ3)·[Σ卜“至⑴/叫])， 及選擇具有最低和之參考光譜作為匹配參考光譜。 現在參閱第7圖’該圖圖示僅單一基板之單—區域: 結果，可決定序列中之最佳匹配光譜十之每一者之指』 值以產生一序列隨時間改變之指數值212。此序列指3 值可稱為指數跡線21〇β在一些實施中，藉由將每一 ^ ’則之光谱與來自正好一個庫之參考光譜相比較來產生4 數跡線。Α體而言，指數跡線210可包括光學監視系、 在基板下方之每次拂掠之—個（例如，正好一個）指· 值。 對於給疋指數跡線2 i 〇，其中存在光學監視系統之單 :拂掠中之特定區域之量測之多個光譜（稱為「當前光 曰」），可在當前光譜中之每一者及一或更多個（例如， 好個）庫之參考光譜中之每一者之間決定最佳匹 配在一些實施中，將每—選擇之當前光譜與一或更多 2擇之庫之每-參考光譜相比較。給定當前光譜e、當 則光瑨f及當前光譜g及參考光譜E、參考光譜F及參 44 201249598 考光譜G，（例如）可計算當 合中之每一者之匹配係數： 匹配係數指示最佳匹配（例 配參考光譜及因此決定指數值。或者 前光譜及參考光譜之以下組 e與E、e與1?、6與〇、1^與 g與F及g與g。任何一個 如）為最小的，決定最佳匹 可組合當前光譜（例如，經平均 與參考光譜相比較以決定最佳匹 在一些實施中， )及將產生之組合光譜 配及因此決定指數值。 ，一些實施中，對於一些基板之至少一些區域，可產 生複數個指數跡線。對於給定基板之給定區域，可產生 感興趣之每一參考庫之指數跡線。 亦即，對於給定基板 :給定區域所感興趣之每一參考庫，將一序列量測之光 譜中之每一量測之光譜與來自給定庫之參考光譜相比 較，決定一序列最佳匹配參考光譜，及該序列最佳匹配 參考光譜之指數值提供給定庫之指數跡線。 總之，每一指數跡線包括一序列210指數值212，其 中藉由選擇來自給定庫之參考光譜之指數產生該序列之 母一特定指數值212，指數值212與量測之光譜最近擬 σ。指數跡線2丨〇之每一指數之時間值可為與量測經量 測之光譜之時間相同。 原位監視技術用於偵測第二層之清除及下層或層結構 之暴路。舉例而言，如下更詳細之描述，可藉由電動機 轉矩或自基板反射之光之總強度之突然改變或自收集之 光4之分散來偵測在時間TC處之第一層之暴露。 45 201249598 如第8圖所圖示，函數（例如，已知階之多項式函數 (例如’第一階函數（例如，、線214)))與時間代之後 收集之光譜之該序列指數值擬合（例如，使用穩健的線 性擬合）。將函數與該序列指數值擬合時，忽略時間代 之前收集之光譜之指數值。可使用其他函數（例如，二 階多項式函數），但是線提供容易之計算。可在終點時間 =處停止研磨，該終點時間丁£為線214與目標指數汀 交又之時。 第9圖圖示製造及研磨產品基板之方法之流程圖。當 測試基板用於產生庫之參考光譜時，產品基板可具有至 少相同的層結構及相同圖案。 開始時，在基板上沈積第一層及將第一層圖案化（步 驟902 )。如上所述，第一層可為介電質（例如，諸如摻 雜石反二氧化矽之低k材料（例如，BUck Diam〇ndTM (來 自 Applied Materials，Inc )或 c〇ralTM (來自 N〇veUus Systems，Inc. ) ) ) 〇 視需要’視第—材料之組成而定，將不同於第一材料 之另一介電材料（例如，低k敷蓋材料（例如，正矽酸 乙酯(TEOS)))《—或更多個附加層沈積在產品基板上 之第一層上方（步驟9〇3)。第一層及—或更多個附加層 共同提供層堆疊。視需要，可在沈積—或更多個附加層 之後進行圖案化（使得一或更多個附加層不會延伸至第 一層中之溝槽内，如第1A圖所圖示）。 46 201249598S-twist, the associated spectrum will appear at the rotation of the platform. 42 201249598 The reference spectrum and the index values associated with the reference spectrum can be stored in the reference library. For example, each reference spectrum 320 and the index value 330 associated with each reference spectrum 32 可 can be stored in the record 35 〇 _ of the database 35 。. A library 350 of reference libraries of reference spectra can be implemented in the memory of the computing device of the polishing apparatus. As described above, for each region of each substrate, based on the sequence-measured spectrum or region and substrate, the controller 19 can be programmed to produce a sequence-best matching spectrum. The best matching reference spectrum can be determined by comparing the measured spectrum to a reference spectrum from a particular library. In some implementations, the best matching reference spectrum can be determined by calculating the sum of the squared differences of the measured light w and the reference spectrum for each reference spectrum. The reference spectrum with the sum of the lowest squared differences of f has the best fit. Other techniques for matching the reference spectrum (e.g., the sum of the lowest absolute differences) are possible. In the two-kappa, the best matching reference spectrum can be determined by using a matching technique other than the sum of the squared differences. In one implementation, for each reference spectrum, the cross-correlation between the measured spectrum and the reference spectrum is calculated, and the reference spectrum with the largest correlation is selected as the matching reference spectrum. The potential advantage of cross-correlation is that it is less sensitive to lateral shifts in the spectrum and therefore less sensitive to layer thickness variations. In order to perform the cross-correlation, when the reference spectrum moves relative to the measured spectrum, the "zero" fill amount can be used to provide the data to the reference spectrum or to the reference spectrum. The value at the leading edge of the measured spectrum is equal to the front end of the spectrum, and the measured end of the spectrum can be filled with a value equal to the value at 43 201249598 after the measured spectrum. The Fourier transform can be used to increase the computational speed of the real-time application of the matching technique. In another implementation, the sum of the Euclidean vector distances, for example, ϋ-1/(λα-λΙ)··[Σ to λ1?|ΐΜ(λ)2_Ικ(λ) 2丨], where ^ to 乩 is calculated The wavelength of the summation ΐΜ(λ) is the measured spectrum, and. (" is the reference spectrum. In another implementation", for each reference spectrum, the derivative differences are summed (eg, ϋ = 1 / (λ & · λ ΐ 3) · [Σ卜" to (1) / called]), and selection The reference spectrum with the lowest sum is used as the matching reference spectrum. Now refer to Figure 7, which shows a single-area of a single substrate: the result, which determines the value of each of the best matching spectra in the sequence. Generating a sequence of index values 212 that change over time. This sequence refers to a value of 3 that can be referred to as an exponential trace 21 〇 β. In some implementations, by comparing each of the spectra with a reference spectrum from exactly one bank To generate 4 number of traces. For the corpus callosum, the exponential trace 210 can include an optical monitoring system, one for each plunder under the substrate (for example, exactly one). For the given index trace 2 i 〇 where there is a single optical monitoring system: multiple spectra measured in a particular area of the smash (called “current pupils”), each of the current spectrum and one or more ( For example, a good one) determines the best match between each of the library's reference spectra. In some implementations, the current spectrum of each selection is compared to each of the reference spectra of one or more selected libraries. Given the current spectrum e, then the pupil f and the current spectrum g and the reference spectrum E, reference Spectral F and reference 44 201249598 Test spectrum G, for example, can calculate the matching coefficient of each of the combinations: The matching coefficient indicates the best match (except the reference spectrum and hence the index value or the front spectrum and the reference spectrum) The following groups e and E, e and 1?, 6 and 〇, 1^ and g and F, and g and g. Any one of which is the smallest, determines the best possible combination of the current spectrum (for example, the average and reference spectra) The comparison is to determine the best match in some implementations, and the resulting combined spectrum is assigned to determine the index value. In some implementations, for at least some regions of some of the substrates, a plurality of exponential traces may be generated. A given region of the substrate can produce an exponential trace for each reference library of interest. That is, for a given substrate: each reference library of interest for a given region, each of the spectra of a sequence of measurements Measured spectrum and Comparing the reference spectra of a given library, determining a sequence of best matching reference spectra, and the index values of the best matching reference spectra of the sequence provide an exponential trace for a given bank. In summary, each index trace includes a sequence of 210 indices. A value 212, wherein a particular index value 212 of the sequence is generated by selecting an index from a reference spectrum of a given library, the index value 212 being closest to the measured spectrum σ. Each index of the exponential trace 2丨〇 The time value can be the same as the time measured by the measured spectrum. The in-situ monitoring technique is used to detect the clearing of the second layer and the storm of the lower layer or layer structure. For example, as described in more detail below, The exposure of the first layer at time TC is detected by a sudden change in the total torque of the motor torque or light reflected from the substrate or from the dispersion of the collected light 4. 45 201249598 As illustrated in Figure 8, a function (eg, a polynomial function of known order (eg, 'first order function (eg, line 214))) is fitted to the sequence index value of the spectrum collected after the time generation (For example, use a robust linear fit). When fitting a function to the sequence index value, the index value of the spectrum collected before the time generation is ignored. Other functions (for example, second-order polynomial functions) can be used, but the lines provide easy calculations. Grinding can be stopped at the end time = the end time is the line 214 and the target exponent. Figure 9 is a flow chart showing a method of manufacturing and polishing a product substrate. When the test substrate is used to generate a reference spectrum of the library, the product substrate can have at least the same layer structure and the same pattern. Initially, a first layer is deposited on the substrate and the first layer is patterned (step 902). As noted above, the first layer can be a dielectric (eg, a low k material such as doped anti-cerium oxide (eg, BUck Diam〇ndTM (from Applied Materials, Inc) or c〇ralTM (from N〇veUus) Systems, Inc. ) ) ) Depending on the composition of the material, another dielectric material that is different from the first material (eg, a low-k coating material (eg, ethyl ortho-decanoate (TEOS) ))) - or more additional layers are deposited over the first layer on the product substrate (steps 9〇3). The first layer and - or more additional layers together provide a layer stack. If desired, the patterning may be performed after deposition - or more of the additional layers (so that one or more additional layers do not extend into the trenches in the first layer, as illustrated in Figure 1A). 46 201249598
Ik後將不同材料之第二層（例如，阻障層（例如， 氮化物（例如，一氮化叙或一氮化欽川，沈積在產品基 曰或層堆髮上方（步驟9〇4 )。另外，傳導層（例 如’金屬層（例如’銅））’可沈積在產品基板之第二層 上方（及在第一層之圖案所提供之溝槽内）（步驟9〇6)。 視需要’第一層之圖案化可發生在沈積第二層之後（在 此情況下’第二層不會延伸至第—層_之溝槽内 研磨產品基板（步驟9〇8)。舉例而言，可使用第一研 磨墊在第一研磨站處研磨及移除傳導層及第二層之部分 (步驟908a)。隨後可使用第二研磨墊在第二研磨站處 研磨及移除第二層及第一層之部分（步驟908b )。然而， 應注意，對於一些實施例而言，不存在傳導層，例如， 當研磨開始時第二層為最外層。當然，可在別處執行步 驟902-906,使得研磨設備之特定操作者可從步驟9〇8 開始製程。 原位監視技術用於偵測第二層之清除及第一層之暴露 (步驟9 10 )。舉例而言，如下更詳細之描述，可藉由電 動機轉矩或自基板反射之光之總強度之突然改變或自收 集之光譜之分散來偵測在時間TC處（參見第8圖）之 第一層之暴露。 至少從偵測到第二層之清除開始（及可能較早地，例 如’從使用第二研磨墊研磨產品基板開始），（例如，如 上所述使用原位監視系統）在研磨期間獲得一序列量測 之光譜（步驟912 )。 47 201249598 分析量測之光譜以產生一序列指數值，且—函數” 序列指數值擬合。詳言之，對於該序列量測之光摄中之 每-量測之光譜，衫參考光譜之指數值以產生^序列 指數值（步驟914)，該參考光譜為最佳擬合。函數（例 如’線性函數）肖時間TC之後收集之光譜之該序列指 數值擬合，在該時間TC處偵測到第二層之清除（步驟 叫。換言之，在時間TC之前收集之光譜之指數值不 用於計算函數，在時間TC處偵測到第二層n -旦指數值（例如，自線性函數產生之計算之指數值， 該線性函數與新序列指數值擬合）到達目標指數，則可 停止研磨（步驟918)。在研磨操作之前可由使用者設定 目標厚度π及儲存目標厚度IT。或者，可由使用者設 定待移除之目標數量，及可根據待移除之目標數量計算 目標指數IT。舉例而言’可根據待移除之目標數量（例 如’根據憑經驗決定之已移除數量與指數（例如，研磨 速率）之比率）來計算折射率差m及將折射率差⑴增 加至時間TC處之指數值IC，在時間Tc處偵測到上層 之清除（參見第8圖）。 亦可能使用與來自偵測到第二層之清除之後收集之光 譜之指數值擬合之函數來調整研磨參數，例如，調整基 板上之一或更多個區域之研磨速率，以改良研磨均勻性。 參閱第10圖，圖示複數個指數跡線。如上所述，可產 生每一區域之指數跡線。舉例而言，可產生第一區域之 第一序列210之指數值212 (藉由空心圓圈圖示），可產 48 201249598 生第二區域之第二序列220之指數值222 (藉由空心方 塊圖示）’及產生第三區域之第三序列23〇之指數值232 (藉由空心三角形圖示）。儘管圖示了三個區域，但可能 存在兩個區域或四個或四個以上區域。所有區域可位於 相同基板上’或區域中之一些可來自在相同平臺上同時 研磨之不同基板。 如上所述，原位監視技術用於偵測到清除第二層及暴 路下層或層結構。舉例而言，如下更詳細之描述，可藉 由電動機轉矩或自基板反射之光之總強度之突然改變； 自收集之光譜之分散來偵測在時間丁c處之第一層之暴 路 ° 珂從母一暴板指 / 少、吗双I 讲， -階函數（例如’線)）與關聯區域之時間TC之後收 =光譜之該序列減錢合，例如，使㈣健的線性 舉例而言，第—線214可與第-區域之指數值⑴ ^:第二線224可與第二區域之指數值如擬合，及 值：1:4可與第三區域之指數值232擬合。線與指數 :之擬合可包括計算線之斜率…轴交又時間τ，在 W Τ處線與開始指數值（例如，0)交又。可以 I⑴=S，（t-T)之形式表示函數，並 時，"可具有負值，指干美板：％間。x轴交又 ，丨、。4 ^ 才曰不基板層之開始厚度比預計的 六…’第一線214可具有第-斜率S1及第一 X軸 乂又％間Ti ’第二線224可具有第二斜率以及第二χAfter Ik, a second layer of different materials (eg, a barrier layer (eg, a nitride (eg, a nitride or a nitride) is deposited over the product substrate or layer stack (step 9.4) Additionally, a conductive layer (eg, a 'metal layer (eg, 'copper))' can be deposited over the second layer of the product substrate (and within the trench provided by the pattern of the first layer) (steps 9-6). It is desirable that the patterning of the first layer can occur after deposition of the second layer (in this case the second layer does not extend into the trenches of the first layer to ground the product substrate (steps 9-8). For example The conductive layer and portions of the second layer may be ground and removed at the first polishing station using a first polishing pad (step 908a). The second polishing pad may then be used to polish and remove the second layer at the second polishing station. And a portion of the first layer (step 908b). However, it should be noted that for some embodiments, there is no conductive layer, for example, the second layer is the outermost layer when the polishing begins. Of course, step 902 can be performed elsewhere. 906, such that a particular operator of the grinding apparatus can begin the process from step 9-8. The in-situ monitoring technique is used to detect the removal of the second layer and the exposure of the first layer (step 9 10 ). For example, as described in more detail below, the total intensity of the light that can be reflected by the motor or from the substrate A sudden change or dispersion of the collected spectrum to detect exposure of the first layer at time TC (see Figure 8). At least from detection of the second layer of cleaning (and possibly earlier, eg ' From the use of the second polishing pad to grind the product substrate, (eg, using an in situ monitoring system as described above) a sequence of measured spectra is obtained during milling (step 912). 47 201249598 Analyzing the measured spectra to produce a sequence The index value, and the - function, sequence index value fit. In detail, for each of the spectra measured in the sequence of measurements, the index value of the shirt reference spectrum is used to generate a sequence index value (step 914). The reference spectrum is the best fit. The function (eg, 'linear function') fits the sequence index value of the spectrum collected after the time TC, and the second layer is detected at that time TC (step call. In other words ,in The index value of the spectrum collected before the TC is not used for the calculation function, and the second layer n-denier index value is detected at time TC (for example, the calculated index value generated from the linear function, the linear function and the new sequence index value) If the target index is reached, the grinding may be stopped (step 918). The target thickness π and the storage target thickness IT may be set by the user before the grinding operation. Alternatively, the target number to be removed may be set by the user, and may be determined according to The target quantity to be removed is calculated as the target index IT. For example, 'the refractive index difference can be calculated according to the number of targets to be removed (for example, 'based on empirically determined ratio of removed quantity to index (eg, grinding rate)) m and increasing the refractive index difference (1) to the index value IC at time TC, and detecting the removal of the upper layer at time Tc (see Fig. 8). It is also possible to adjust the grinding parameters using a function fitted to the index value of the spectrum collected after the detection of the second layer is removed, for example, adjusting the polishing rate of one or more regions on the substrate to improve the polishing uniformity. . Referring to Figure 10, a plurality of exponential traces are illustrated. As mentioned above, the index traces for each region can be generated. For example, an index value 212 of the first sequence 210 of the first region can be generated (illustrated by an open circle), and an index value 222 of the second sequence 220 of the second region can be generated 48 201249598 (by means of a hollow block diagram And the index value 232 of the third sequence 23 of the third region (shown by a hollow triangle). Although three regions are illustrated, there may be two regions or four or more regions. All of the regions may be on the same substrate or some of the regions may be from different substrates that are simultaneously ground on the same platform. As described above, the in-situ monitoring technique is used to detect the removal of the second layer and the underlying layer or layer structure. For example, as described in more detail below, a sudden change in the total intensity of the motor torque or light reflected from the substrate; the dispersion of the collected spectrum to detect the first layer of the storm at time c ° 珂 From the mother to the storm board finger / less, double I speak, - the order function (such as 'line") and the associated area time TC after the = spectrum of the series of money, for example, to make (four) health linear example In other words, the first line 214 can be compared with the index value of the first region (1) ^: the second line 224 can be compared with the index value of the second region, and the value: 1:4 can be compared with the index value 232 of the third region. Hehe. Line and Exponential: The fit can include calculating the slope of the line...the axis intersection and the time τ, where the line at W 与 intersects with the start index value (for example, 0). The function can be expressed in the form of I(1)=S, (t-T), and " can have a negative value, referring to the dry board: %. The x-axis is again, 丨,. 4 ^ The initial thickness of the substrate layer is less than the expected six...' The first line 214 may have a first slope S1 and a first X axis 乂 and a % Ti ′ second line 224 may have a second slope and a second χ
S 49 201249598 軸交叉時間T2,及第三線234可具有第三斜率s3及第 三X軸交叉時間T3。 在研磨製程期間的一些時間處（例如，在時間τ〇處》 調整至少-個區域之研磨參數以調签基板之區域之研磨 速率，使得在研磨終點時間處，複數個區域比沒有此調 整時更靠近該複數個區域之目標厚度。在一些實施例 中’每一區域在終點時間處可具有大約相同厚度。 參閱第η圖’在一些實施中，選擇—個區域作為參考 區域’及決定預測終點時間ΤΕ，在該預測終點時間π 處參考區域將到達目標指數ΙΤ。舉例而言，如第U 中所圖示，儘管可選擇不同區域及/或不同基板，但㈣ 第一區域作為參考區域。在研磨操作之前可由使用者設 定目標厚度及儲存目標厚度…或者， 設定待移除之目標數量TR，月·占 TR&數里TR，及可根據待移除之目標數量 什异目標指數IT。舉例而言， 了根據待移除之目標 例如，根據憑經驗決定之已移除之數量與指數 如’研磨速率）之比 ΨΆ ID 5 * °折射率差ID及將折射 手差iD增加至時間tc處之指數 測到上層之清除。 值W，在時間TC處镇 為了决疋預測時間，在該預測時 _數，可計算參考區域之線(例如，達 偏二 定貫穿其餘研磨製程研磨速率沒有 預4之研磨速率，那麼該序 線性級數。因此 , ^值應保持實質上 預計之終點時間巧可作為線對目標 50 201249598 指數IT之簡單線性内插計算，例如，IT=S*(TE—丁）因S 49 201249598 The axis crossing time T2, and the third line 234 may have a third slope s3 and a third X-axis crossing time T3. At some time during the polishing process (eg, at time τ〇), the polishing parameters of at least one region are adjusted to adjust the polishing rate of the region of the substrate such that at the polishing endpoint time, the plurality of regions are less than the adjustment Closer to the target thickness of the plurality of regions. In some embodiments 'each region may have approximately the same thickness at the end time. See Figure η 'in some implementations, select one region as a reference region' and determine the prediction At the end time ΤΕ, the reference area will reach the target index π at the predicted end time π. For example, as illustrated in the U, although different regions and/or different substrates may be selected, (4) the first region serves as a reference region The target thickness and the storage target thickness can be set by the user before the grinding operation... or, the target number TR to be removed, the month, the TR& number of TRs, and the target number index according to the target to be removed. For example, depending on the target to be removed, for example, based on empirically determined quantities and indices such as 'grinding rate' ΨΆ ID 5 * ° than the refractive index difference between the ID and the refractive index increased to hand iD difference at the measured time tc remove the upper layer. The value W, at time TC, in order to determine the prediction time, at the time of the prediction, the line of the reference area can be calculated (for example, the polishing rate of the remaining grinding process does not exceed the polishing rate of the pre-4, then the order The linear progression. Therefore, the ^ value should remain at the substantially predicted end time and can be used as a simple linear interpolation calculation for the line-to-target 50 201249598 index IT, for example, IT=S*(TE-D)
此，在使用關聯之第一線214選擇第一區域作為參考區 域之第11圖之實例中’ I丁=S1.(TE ·* 1 J ，亦即， TE=IT/S1-T1 〇 ;又夕徊區域，例如，除了參考區域之外的全部區 域（包括其他基板上的區域），可定義為可調整區域。當 可調整區域之線到達預計終點時間TE時，定義可調整 區域之預測終點。因此，每一可調整區域之線性函數（例 如，第11圖中之線224及線234)可用於外推將在關聯 區域之預計終點時Γθ1 ET處到達之指數（例如，M2及 ΕΠ)。舉例而言，第二線224可用於外推在第二區域之 預計終點時間ET處之預計指數Eu，及第三線Μ#可用 於外推在第三區域之預計終點㈣Ετ處之預計 ΕΙ3。 如第11圖所圖示，若沒有調整時間T〇後之區域中之 任一者之研磨速率，隨❹同時對所有區域強制終點， 則每-區域可具有不同厚度（不同厚度並非所欲的，因 為不同厚度會導致缺陷及產量損失）。 若將在不同區域之不同睥問卢 u守間處到達之目標指數（或相 备於，可調整區域將在表 /号&域之預測終點時間處具有 不同預計指數），則可向上或 4向下調整研磨速率，使得區 域將比沒有此調整時更靠近 k N 4 (例如，大約同時）到 達目標指數（及因此到達目桿 ^ , 知与·度）’或區域將比沒有此 調整時在目標時間處更靠近相 、相问指數值（及因此更靠近 51 201249598 相同厚度），例如，更靠近大約相同指數值（及因此更靠 近大約相同厚度^ 因此’在第11圖之實例中’在時間τ〇處開始，改變 品域之至；>' 個研磨參數，使得區域之研磨速率增 加（及因此增加指數跡線22〇之斜率）。同樣地，在此^ 例中’改變第三區域之至少—個研磨參數，使得第三區 域之研磨速率減少（及因此減少指數跡線230之斜率）。 =此，區域將大約同時到達目標指數（及因此到達目標 厚度)(或若同時停止對區域之壓力，則區域將以大約： 同厚度停止）。 匕在些只施中，若在預計終點時間ετ處之預測指數 ^ τ基板之區域位於目標厚度之預先決定範圍内，則彼 區域不f要調整。範圍可為目標指數之2%(例如，不超 過 1 % )。 可調整可調整區域之研磨速率使得所有區域比沒有此 調整時在預計終點時間處更靠近目標指數β舉例而言， 可選擇參考基板之參考區域及調整所有其他區域之：理 ’數使侍所有區域將在參考基板之大約預測時間處停 止參考區域可為（例如）預先決定區域（例如，中心 區域_或緊靠地圍繞中心區域之區域該區域 具有基板中之任—去u &域中之任一者之最早及最晚預 測終點時間，或基板之區域具有所欲之·終點。若同 時停亡研磨，則最早時間等同於最薄基板。同樣地，若 同時停止研磨’則最晚時間等同於最厚基板。參考基板 52 201249598 α ’’、、（例如）預先決定基板，基板具有區域，該區域具 =板之最早或最晚預測終點時間。若同時停止研磨， ' 157早時間等同於最薄區域。同樣地，若同時停止研磨， 則最晚時間等同於最厚區域。 …:於可調整區域中之每一者’可計算指數跡線之所欲 / 使得可5周整區域與參考區域同時到達目標指數。舉 例而5，可根據（it-i)=sd*(te_to)計算所欲斜率SD，其 為將改艾研磨參數之時間το處之指數值（根據與該 序歹Η曰數值擬合之線性函數計算Ιτ為目標指數，且 =_為广算之預計終點時間。在第11圖之實例中，對於 弟-Q域’可根據（it_I2) = SD2j(!(te_tg)計算所欲之斜率 SD2，及對於第三區域可根據叩工3卜sd3*(te_t〇)計算 所欲之斜率SD3。 只施中，不存在參考區域，及預 時間可為預先決定時間，例如，在研磨製程之前由使用 者設定，或可根據來自-或更多個基板之二個或二個以 上區域（當藉^預測各㈣域到達目標指數之線計算 之預計終料間之平均值或其他組合計算。在此^ 中，儘官亦必須計算第一基板之第一區域之所欲斜率 如’可根據（iT-.sr^dTo)計算所欲之 SD1)，但實質上如上所述計算所欲之斜率。 gs. 或在。實施中，存在不同區域之不同目標指數。 此可允許在基板上建立預先考慮但可控制的不均句厚戶 輪廓。使用者（例如，使用控制器上的輸入設備）可ς 53 201249598 入目標指數。舉例而言，第-基板之第-區域可具有笫 1㈣’第—基板之第二區域可具 基板之第一區域可具有第三目標指數，及第二 基板之第二區域可具有第四目標指數。 一 對於上述方法中之任一者，調整研磨速率以使得指數 二線之斜率更靠近所欲之斜率。例如，可藉由增加或減 少承载頭之相應腔室中之壓力來調整研磨速率。可假定 研磨速率之變化與壓力之變化直接成正比（例如，簡單 的P⑽onian模型）。舉例而t，對於每·一基板之每二區 域，其中在時間T0之前使用壓力_研磨區域，在時 門το之後將施加之新壓力pnew可計算為·' pnew=P〇ld*(SD/s)，其+ s為時間τ〇之前的:的斜率 且SD為所欲之斜率。 舉例而言’假定將壓力Poldl施加至第一基板之第_ 區域，將壓力Pold2施加至第一基板之第二區域，將壓 力Pold3施加至第二基板之第一區域，及將壓力 施加至第二基板之第二區域，則第—基板之第一區域之 新壓力 Pnewl 可計算為：pnewl=P〇ldl*(SDl/Sl)，第一 基板之弟二區域之新壓力 pnevv2可計算為. Pnew2=P〇ld2*(SD2/S2) ’第二基板之第一區域之新壓力 Pnew3 可計算為：pnew3=P〇ld3*(SD3/S3)，第二基板之 弟二區域之新壓力 Pnew4 可計算為 · Pnew4=Pold4*(SD4/S4) ° 決定基板將到達目標厚度之預測時間及調整研磨速率 S' 54 201249598 之製程僅可在研磨制 …知 衣心期間執行-次（例如，在給定時 = (=，經過預計研磨時間娜至60%) 在研磨製程期間多次勃并^ 3(1 丁 a 執仃（例如，每三十至六十秒執行 、=研磨製程期間在後續時間處，可再次調整速率（如 適：:。又研磨製程期㈤，僅可幾次改變研磨速率（諸如 人兩'人或僅一次）。可在接近研磨製程之開始 處、中間處或接近結束處進行調整。 在已、i凋整研磨速率之後（例如’時間μ之後）繼 續研磨，光學監視系統繼續收集至少參考區域之光譜及 2參考區域之指數值。在—些實施中’光學監視系統 繼續收集每-區域之光譜及決定每—區域之指數值。一 旦參考區域之指數跡線達到目標指數，則洲終點及終 止研磨操作。 例而。’如第！ 2圖中所圖示，在時間之後，光 ，監視系統繼續收集參考區域之光譜及蚊參考區域之 心數值3 1 2。右參考區域上之壓力沒有改變（例如，如 在第U圖之實施中）’則可使用來自τ〇之前（但不是 月J )及Τ0之後的資料點計算線性函數以提供更新 之線性函冑314,及線性函數314到達目標指數Ιτ之時 間扣不研磨終點時間。另一方面，若參考區域上之壓力 在Τ0處改變’則可根據時間τ〇之後之該序列指數值 計算具有斜率S，之新線性函數3丨4 ’及新線性函數314 到達目標指數IT之時間指示研磨終點時間。用於決定終 點之參考區域可為上述使用之相同參考區域以計算預計 201249598 終點時間或不同區域(或如參閱第u圖所述若調整所有 區域，則可選擇參考區域以實現決定終點之目的）。若新 線性函數川比根據原始線性函數214計算之預測時間 賴遲（如第12圖所圖示）或稍早到達目標指數it，則 區域中之一或更多者可分別略微過度研磨或研磨不足。 然而，由於預計終點時間與實際研磨時 小於幾秒鐘，故此舉不會嚴重影響研磨均句^差異應 在一些實施中（例如’對於銅研磨），在偵測到基板之 終點後，基板將立即經歷過度研磨製程（例如，以移除 銅剩餘物）。可在基板之所有區域之均勻i力下進行過度 研磨製程，例如’ 1至！ 5 psi。過度研磨製程可具有預 置持續時間（例如’ 1 0至1 5秒）。 若產生特疋區域之多個指數跡線（例如，特定區域感 興趣之每一庫之一個指數跡線），則可選擇指數跡線中之 一者用於特定區域之終點或壓力控制演算法。舉例而 言，相同區域產生之每一指數跡線，控制器190可將線 性函數與彼指數跡線之指數值擬合，及決定彼線性函數 與該序列指數值之擬合良好性。可選擇產生之指數跡線 作為特定區域及基板之指數跡線，該指數跡線具有與該 指數跡線自身的指數值有著最佳擬合良好性之線。舉例 而言’當決定如何調整可調整區域之研磨速率時（例如， 在時間T0處），具有最佳擬合良好性之線性函數可用於 計算。如另一實例’當具有最佳擬合良好性之線之計算 之指數（當根據與該序列指數值擬合之線性函數計算） 56 201249598 匹配或超過目標指數時，可調用終點。同樣地，並非根 據線性函數計算指數值，指數值本身可與目標指數相比 較以決定終點。 決定與光譜庫相關聯之指數跡線是否具有與庫相關聯 之線性函數之最佳擬合良好性之步驟可包括以下步驟： 相對地，相較於與關聯之強健線及與另一庫相關聯之指 數，線之差異（例如，最低標準差、最大相關性或其他方 差量測），決定關聯之光譜庫之指數跡線是否具有與關聯 之強健線之最少差異量。在一個實施中，藉由計算指數 資料點與線性函數之間之平方差之和決定擬合良好性； 具有最小平方差之和之庫具有最佳擬合。 參閱第13圖’圖示概述性流程圖13〇〇。如上所述， 使用相同研磨塾在研磨設備中同時研磨基板之複數個區 域（步驟1 302 )。在此研磨操作期間，藉由獨立可變的 研磨參數（例如，由特定區域上方之承載頭中的腔室施 力的壓力）每一區域具有自身之獨立於其他基板之可控 制的研磨速率。在研磨操作期間，如上所述監視基板（步 驟13⑷’例如’使用自每—區域獲得之—序列量測光 譜。對於該序列巾之每-量測之光譜，決定參考光譜， :參考光譜為最佳匹配（步驟13〇6)。決定最佳擬合之 每-參考光譜之指數值以產生一序列指數值（步驟 1308 ) 〇 偵測到第二層之清除（步驟131〇)。對於每一區域， 線函數與偵測到第二層之清除之後所收集光譜之該序 57 201249598 列指數值擬合（步驟1312)。在—個實施中，（例如，藉 由線性函數之線性内插）決定參考區域之線性函數將^ 達目標指數值之預計終點時間（步驟1314)。在其他實 施中，預先決定預計終點時間或作為多個區域之預計終 點時間之組合計算預計終點時間。若有需要，調整其他 區域之研磨m調整彼基板之研磨速率，<吏得複數個 區域大約同％到達目標厚度或使得複數個區域在目標時 間具有大約相同厚度（或目標厚度）（步驟1316)。調整 參數之後繼續研磨，及對於每一區域：量測光譜；決定 來自庫之最佳匹配參考光譜；決定最佳匹配光譜之指數 值以在調整研磨參數之後的時間週期期間產生一序列新 指數值；以及將線性函數與指數值擬合（步驟13丨8)。 一旦參考區域之指數值（例如，自線性函數產生之計算 之指數值，該線性函數與該序列新指數值擬合）到達目 標指數’則可停止研磨（步驟13 3 〇 )。 在一些實施中，該序列指數值用於調整基板之一或更 多個區域之研磨速率’但是另一原位監視系統或技術用 於偵測研磨終點。 如上所述，對於—些技術及一些層堆疊，偵測到上層 之清除及下層之暴露可為困難的。在一些實施中，收集 一序列光谱組，及計算每一組光譜之分散參數之值以產 生一序列分散值。可根據該序列分散值偵測到上層之清 除。此技術可用於（例如，在上述研磨操作之步驟910 或13 10中）偵測第二層之清除及第一層之暴露。 58 201249598 第14圖圖不用於偵測第二層之清除及第一層之暴露 之方法1400。當正在研磨基板時（步驟14〇2)，收集一 序列光譜組（步驟1404)。如第4圖中所圖示，若光學 監視系統緊固至旋轉平臺，則在光學監視系統跨過基板 之單拂掠中，可自基板上之多個不同位置20 lb-20 lj 收集光譜。自單一拂掠收集之光譜提供一組光譜。當研 磨進行時，光學監視系統之多個拂掠提供一序列光譜 組。每一平臺旋轉時可收集一組光譜，例如，可以與平 臺旋轉速率等同之頻率收集光譜組。通常，每一組將包 括五至二十個光譜。可使用相同光學監視系統收集光 瑨，該光學監視系統用於收集上述峰值追蹤技術之光譜。 第15A圖提供研磨開始時（例如，當下層上方剩餘上 層之明顯厚度時）自基板10反射之光之一組量測之光譜 15〇0a之實例。光譜組i鳥可包括在光學監視系統跨 過基板之第一拂@中在基板上之不同&置處&集之光譜 2仏-咖。第15B圖提供在上層之清除處或附近自基= 10反射之光之一組量測之光15_ t實例。光譜组 15_可包括在光學監㈣統跨過基板之不同第二拂掠 中在基板上之不同位置處收集之光譜2〇2b_2〇4b (可自 基板上不同於收集光譜1500b之位置的位置處收集光譜 1500a)。 θ 開始時，如第i5A圖中所圖示，光譜15_為極相似 的。然而，如第圖中所圖*，當清除上層（例如， 阻障層）及暴露下層（例如’低k層或敷蓋層）時，來 59 201249598 自基板上之不同位置之光譜1500b之間的差異逐漸變得 明顯。 對於每一組光譜，計算該組中之光譜之分散參數之值 (步驟1406)。此舉產生一序列分散值。 在-個實施中，A了計算一組光譜之分散參數，將強 度值（作為波長之函數）一起平均化以提供平均光譜。 亦即WWSi/nhi： i=1至ν Ιί(λ)]，其中N為該組中光 譜之數量且Ιί(λ)為光譜。對於該組中之每一光譜，可隨 後計算光譜及平均光譜之間的全差，例如，使用平方差 之和或絕對值差之和（例如，㈣1/(λ^).[Σ⑼至心 (λ)-ΙΑνΕ(λ)]2]],/2 ^ Dfn/aaAb).!^ λ=“ 至 ，其中^至為求和之波長範圍）。 -旦已經計算該組光譜中之每一光譜之差值，則可根 據差值計算該組之分散參數之值。各種分散參數為可能 的，諸如標準差、四分位差、全距（最大值減最小值）、 平均差、絕對令位差及平均絕對差。可分析該序列分散 值及將該序列分散值用於偵測上層之清除（步驟14〇8)。 第16圖圖示作為研磨時間之函數之光譜之標準差之 圖1600 (其中根據一組光譜之差值計算每一標準差）。 因此，對於在光學監視系統之給定拂掠處所收集之該組 光譜之差值，圖中每—標出豸16G2為標準差。如圖所 示’在第-時間週期1610期間，標準差值保持相當低。 然而，在時間週期1610之後，標準差值變大且更為分 散。不受限於任何特定理論’厚的阻障層趨向於支配反 60 201249598 射之光譜、阻障層本身之厚度中之掩蔽差及任何下層。 當研磨進行時，阻障層變得更薄或為完全地移除，及反 射之光譜對下層厚度之變化變得更為敏感。因此，當清 除阻障層時，光譜之分散將可能增加。 田'月 當清除上層時，可使用各種演算法偵測分散值之性質 ，變化。舉例而言’該序列分散值可與閾值相比較，及 若分散值超過閾值，則產生訊號指示已清除上層。作為 另-貫例’可計算移動窗口内部之該序列分散值之部分 之斜率’及若斜率超過閾值，則產生訊號指示已清除上 層0 作為摘測分散中之增加的演算法之部分，為了移除高 頻雜訊’該序列分散值可受m (例如，低通或頻帶 濾波器）支配。低通濾波器之實例包括流動平均數及巴 特沃斯（Butterworth)濾波器。 儘管以上論述集中在偵測阻障層之清除，但在直他上 下文中該㈣可mm之清除，例如，在使用介 電層堆g (例如’層間介雷「IT pj〗）+ σ 包)之另一類型半導體製 程中之上層之清除，或介電屛μ 、 电層上方之溽金屬層之清除。 除了如上所述用作觸發器啟叙 段勁特徵追蹤之外，用於谓 測上層之清除之此技術可為立 苟再他目的用於研磨操作中， 例如，用作終點訊號自身，以錨 ^以觸發計時器使得在暴露下 層之后在預先決定之時間期問讲沿τ w 4間研磨下層，或用作觸發器 以改變研磨參數（例如，在晨♦ 恭路下層之後改變承載頭壓 力或漿料組成）》Thus, in the example of FIG. 11 in which the first region 214 is selected as the reference region using the associated first line 214, 'I ·=S1. (TE ·* 1 J , that is, TE=IT/S1-T1 〇; The evening region, for example, all regions except the reference region (including regions on other substrates) may be defined as an adjustable region. When the line of the adjustable region reaches the expected end time TE, the predicted end point of the adjustable region is defined. Thus, a linear function of each adjustable region (eg, line 224 and line 234 in Figure 11) can be used to extrapolate the index that will arrive at Γ θ1 ET at the predicted end point of the associated region (eg, M2 and ΕΠ) For example, the second line 224 can be used to extrapolate the predicted index Eu at the predicted end time ET of the second region, and the third line Μ# can be used to extrapolate the expected ΕΙ3 at the predicted end point (four) Ετ of the third region. As illustrated in Fig. 11, if the polishing rate of any of the regions after the time T〇 is not adjusted, each region may have a different thickness (the thickness is not desirable as it is forcibly terminating all regions simultaneously). Because different thicknesses will lead Defects and production losses). If the target index of the arrival of Luu custodian is different in different regions (or prepared, the adjustable region will have different prediction indices at the predicted end time of the table/number & field) ), the polishing rate can be adjusted upwards or downwards so that the region will reach the target index (and thus reach the target, knowing and degree) closer to k N 4 (eg, approximately simultaneously) than without this adjustment. The area will be closer to the phase, the interrogation index value (and therefore closer to the same thickness of 51 201249598) at the target time than without this adjustment, for example, closer to approximately the same index value (and therefore closer to approximately the same thickness ^ thus 'in In the example of Fig. 11, 'starting at time τ〇, changing the range of the product; >' grinding parameters increase the polishing rate of the region (and thus increase the slope of the exponential trace 22〇). In this example, 'change at least one grinding parameter of the third region, such that the polishing rate of the third region is reduced (and thus the slope of the exponential trace 230). = This, the region will be approximately simultaneously When the target index is reached (and therefore reaches the target thickness) (or if the pressure on the area is stopped at the same time, the area will stop at approximately: the same thickness). 匕 In some applications, if the predicted index at the expected end time ετ ^ τ If the area of the substrate is within a predetermined range of the target thickness, then the area is not adjusted. The range may be 2% of the target index (for example, no more than 1%). The polishing rate of the adjustable area can be adjusted so that all areas are less than This adjustment is closer to the target index β at the estimated end time. For example, the reference area of the reference substrate can be selected and all other areas can be adjusted: the number is such that all areas will stop the reference area at approximately the predicted time of the reference substrate. The earliest and latest predicted end time of any one of the regions in the substrate, for example, the central region _ or the area immediately surrounding the central region, may be, for example, a predetermined region (eg, a central region _ or a region immediately adjacent to the central region), Or the area of the substrate has the desired end point. If the grinding is stopped at the same time, the earliest time is equivalent to the thinnest substrate. Similarly, if the grinding is stopped at the same time, the latest time is equivalent to the thickest substrate. The reference substrate 52 201249598 α ′′, for example, determines the substrate in advance, and the substrate has a region having the earliest or latest predicted end time of the plate. If the grinding is stopped at the same time, '157 early is equivalent to the thinnest area. Similarly, if the grinding is stopped at the same time, the latest time is equivalent to the thickest area. ...: Each of the adjustable areas can be calculated as desired by the exponential trace / so that the entire area can reach the target index at the same time as the reference area. For example, 5, the desired slope SD can be calculated according to (it-i)=sd*(te_to), which is the index value at the time το of the grinding parameter (according to the linearity fitted to the sequence value) The function calculates Ιτ as the target index, and =_ is the estimated end time of the general calculation. In the example of Fig. 11, for the brother-Q domain, the desired slope SD2 can be calculated according to (it_I2) = SD2j(!(te_tg) And for the third region, the desired slope SD3 can be calculated according to the completion 3 sd3*(te_t〇). Only the reference region is present, and the pre-time can be a predetermined time, for example, before the polishing process The setting may be based on an average or other combination between two or more regions from - or more substrates (calculated by predicting each (four) domain to reach the target index). ^, the official must also calculate the desired slope of the first region of the first substrate, such as 'can calculate the desired SD1 according to (iT-.sr^dTo)), but calculate the desired slope substantially as described above. Gs. or in the implementation, there are different target indices for different regions. This allows A pre-considered but controllable uneven-thickness profile is established on the board. The user (eg, using an input device on the controller) can enter the target index 。 53 201249598. For example, the first-region of the first substrate can have The first region of the first substrate of the first substrate may have a third target index, and the second region of the second substrate may have a fourth target index. For any of the above methods, the adjustment The grinding rate is such that the slope of the exponential two lines is closer to the desired slope. For example, the grinding rate can be adjusted by increasing or decreasing the pressure in the corresponding chamber of the carrier head. It can be assumed that the change in the grinding rate is directly related to the change in pressure. Proportional (for example, a simple P(10) onian model). For example, t, for every two regions of a substrate, where the pressure_grinding region is used before time T0, the new pressure pnew to be applied after the time gate το can be calculated as 'pnew=P〇ld*(SD/s), where + s is the slope of the time before the time τ〇 and SD is the desired slope. For example, 'assuming that the pressure Poltl is applied to the first _ region of the first substrate a region, the pressure Pold2 is applied to the second region of the first substrate, the pressure Pold3 is applied to the first region of the second substrate, and the pressure is applied to the second region of the second substrate, the first region of the first substrate The new pressure Pnewl can be calculated as: pnewl=P〇ldl*(SDl/Sl), the new pressure pnevv2 of the second substrate of the first substrate can be calculated as. Pnew2=P〇ld2*(SD2/S2) 'The second substrate The new pressure Pnew3 of the first region can be calculated as: pnew3=P〇ld3*(SD3/S3), and the new pressure Pnew4 of the second substrate of the second substrate can be calculated as · Pnew4=Pold4*(SD4/S4) ° The predicted time to reach the target thickness and the process to adjust the grinding rate S' 54 201249598 can only be performed during the grinding process... (for example, at the given time = (=, after the expected grinding time is 60%) During the grinding process, a number of tens of times (for example, every 30 to 60 seconds of execution, = during the grinding process at a subsequent time, the rate can be adjusted again (if appropriate::. In addition to the grinding process (five), the grinding rate can only be changed several times (such as two people or only once). Adjustments can be made near, at or near the beginning of the grinding process. After the grinding has continued, i.e., after 'time μ, the optical monitoring system continues to collect index values for at least the reference region and the reference region. In some implementations, the optical monitoring system continues to collect the spectrum for each region and determines the index value for each region. Once the index trace of the reference area reaches the target index, the end point of the continent and the grinding operation are terminated. For example. ‘如第! As shown in Fig. 2, after time, the light monitoring system continues to collect the spectrum of the reference area and the heart value of the mosquito reference area 3 1 2 . The pressure on the right reference area has not changed (for example, as in the implementation of Figure U), then the linear function can be calculated using data points from before (but not after month J) and after 〇0 to provide an updated linear function. 314, and the time when the linear function 314 reaches the target index Ιτ is not the grinding end time. On the other hand, if the pressure on the reference region changes at Τ0, the new linear function 3丨4' having the slope S, and the new linear function 314 reaching the target index IT can be calculated from the sequence index value after the time τ〇. The time indicates the grinding end time. The reference area used to determine the end point can be the same reference area used above to calculate the expected 201249598 end time or different areas (or if the area is adjusted as described in Figure u, the reference area can be selected for the purpose of determining the end point) . If the new linear function is delayed by the predicted time calculated according to the original linear function 214 (as illustrated in Fig. 12) or earlier reaches the target index it, one or more of the regions may be slightly overgrinded or ground, respectively. insufficient. However, since the expected end time is less than a few seconds from the actual grinding time, this will not seriously affect the grinding. The difference should be in some implementations (eg 'for copper grinding'), after the end of the substrate is detected, the substrate will Immediately undergo an over-grinding process (eg, to remove copper residue). Excessive grinding processes can be performed at uniform i-forces in all areas of the substrate, such as '1 to! 5 psi. The overgrinding process can have a preset duration (e.g., '10 to 15 seconds). If multiple exponential traces of a particular region are generated (eg, an exponential trace for each library of interest for a particular region), one of the exponential traces may be selected for the end of a particular region or a pressure control algorithm . For example, for each of the exponential traces generated by the same region, the controller 190 can fit the linear function to the exponential value of the exponential trace and determine the fit of the linear function to the index value of the sequence. The resulting exponential trace can be selected as an exponential trace for a particular region and substrate that has a line that best fits the index value of the exponential trace itself. For example, when deciding how to adjust the grinding rate of the adjustable region (for example, at time T0), a linear function with the best fit goodness can be used for the calculation. As another example, the endpoint can be called when the calculated index of the line with the best fit goodness (calculated based on a linear function fitted to the sequence index value) 56 201249598 matches or exceeds the target index. Similarly, instead of calculating an index value based on a linear function, the index value itself can be compared to the target index to determine the end point. The step of determining whether the exponential trace associated with the spectral library has a best fit goodness to the linear function associated with the library can include the following steps: Relatively, compared to the associated robust line and associated with another library The index of the joint, the difference in the line (for example, the minimum standard deviation, the maximum correlation, or other variance measurements) determines whether the index trace of the associated spectral library has the least amount of difference from the associated strong line. In one implementation, the fit is determined by calculating the sum of the squared differences between the index data points and the linear function; the library with the sum of the least squares differences has the best fit. See Figure 13 for an overview flow chart 13A. As described above, a plurality of regions of the substrate are simultaneously polished in the polishing apparatus using the same abrasive crucible (step 1 302). During this grinding operation, each zone has its own controllable polishing rate independent of the other substrate by independently variable grinding parameters (e.g., pressure applied by the chamber in the carrier head above a particular zone). During the grinding operation, the substrate is monitored as described above (step 13(4)', eg, using the sequence measurement spectrum obtained from each region. For each spectrum of the sequence towel, the reference spectrum is determined, the reference spectrum is the most Good match (steps 13〇6). Determine the best fit per index value of the reference spectrum to produce a sequence of index values (step 1308) 〇 detect the second layer of clearing (step 131〇). For each The region, the line function is fitted to the sequence of the sequence acquired by the second layer after the second layer is detected, and the index value is fitted (step 1312). In one implementation, (for example, linear interpolation by a linear function) Determining the linear function of the reference region will reach the predicted end time of the target index value (step 1314). In other implementations, the expected end point time is determined in advance or as a combination of expected end times for multiple regions to calculate the expected end time. Adjusting the grinding m of the other areas to adjust the polishing rate of the substrate, < obtaining a plurality of regions reaching the target thickness by about the same % or making the plurality of regions have the target time at the target time Approximately the same thickness (or target thickness) (step 1316). Continue grinding after adjusting the parameters, and for each region: measure the spectrum; determine the best matching reference spectrum from the library; determine the index value of the best matching spectrum to adjust Generating a sequence of new index values during the time period after the grinding parameter; and fitting the linear function to the index value (steps 13丨8). Once the index value of the reference region (eg, the calculated index value from the linear function, The linear function is fitted to the new index value of the sequence) to the target index' to stop grinding (step 13 3 〇). In some implementations, the sequence index value is used to adjust the polishing rate of one or more regions of the substrate' However, another in-situ monitoring system or technique is used to detect the polishing endpoint. As noted above, for some techniques and some layer stacks, it may be difficult to detect the removal of the upper layer and the exposure of the lower layer. In some implementations, collection a sequence of spectra, and calculating the values of the dispersion parameters of each set of spectra to produce a sequence of discrete values. The detected values can be detected based on the dispersion values of the sequence. This technique can be used (for example, in step 910 or 13 10 of the above-described grinding operation) to detect the removal of the second layer and the exposure of the first layer. 58 201249598 Figure 14 is not used to detect the second layer. Method 1400 for removing and exposing the first layer. When the substrate is being polished (step 14〇2), a sequence of spectra is collected (step 1404). As illustrated in Figure 4, if the optical monitoring system is tightened to rotation The platform collects spectra from a plurality of different locations on the substrate from 20 lb to 20 lj in a single sweep of the optical monitoring system across the substrate. The spectrum collected from a single sweep provides a set of spectra. Multiple sweeps of the optical monitoring system provide a sequence of spectra. Each platform can be rotated to collect a set of spectra, for example, a set of spectra can be collected at a frequency equal to the rate of rotation of the platform. Typically, each group will consist of five to twenty spectra. The pupil can be collected using the same optical monitoring system used to collect the spectrum of the peak tracking technique described above. Figure 15A provides an example of a spectrum 15 〇 0a of a group of light reflected from the substrate 10 at the beginning of the grinding (e.g., when the apparent thickness of the upper layer above the lower layer is present). The spectral group i bird can be included in the first 拂@ of the optical monitoring system across the substrate. The different & set & Figure 15B provides an example of light 15_t measured at a group of light reflected from the base at or near the upper layer. The spectral group 15_ may include a spectrum collected at different positions on the substrate in different second sweeps of the optical monitor across the substrate. 2〇2b_2〇4b (a position different from the position on the substrate different from the collected spectrum 1500b) The spectrum is collected at 1500a). At the beginning of θ, as illustrated in Figure i5A, the spectrum 15_ is very similar. However, as shown in the figure in the figure, when the upper layer (for example, the barrier layer) is removed and the lower layer (for example, the 'low-k layer or the cap layer) is exposed, 59 201249598 is derived from the spectrum 1500b at different positions on the substrate. The difference has gradually become apparent. For each set of spectra, the values of the dispersion parameters of the spectra in the set are calculated (step 1406). This produces a sequence of discrete values. In one implementation, A calculates the dispersion parameters for a set of spectra, and averages the intensity values (as a function of wavelength) to provide an average spectrum. That is, WWSi/nhi: i=1 to ν Ιί(λ)], where N is the number of spectra in the group and Ιί(λ) is the spectrum. For each spectrum in the set, the total difference between the spectrum and the average spectrum can then be calculated, for example, using the sum of the squared differences or the sum of the absolute differences (eg, (4) 1/(λ^). [Σ(9) to heart ( λ)-ΙΑνΕ(λ)]2]],/2 ^ Dfn/aaAb).!^ λ=“ to , where ^ is the summed wavelength range. — Once each spectrum in the set of spectra has been calculated The difference can be used to calculate the value of the dispersion parameter of the group. Various dispersion parameters are possible, such as standard deviation, interquartile range, full range (maximum minus minimum), average difference, absolute order Difference and mean absolute difference. The sequence dispersion value can be analyzed and the sequence dispersion value used to detect the removal of the upper layer (steps 14-8). Figure 16 is a diagram 1600 of the standard deviation of the spectrum as a function of the milling time. (where each standard deviation is calculated from the difference between a set of spectra.) Thus, for the difference in the set of spectra collected at a given swept position of the optical monitoring system, each of the plotted 豸16G2 is the standard deviation. As shown in the figure, during the first-time period 1610, the standard deviation remains fairly low. However, at the time After the period 1610, the standard deviation becomes larger and more dispersed. It is not limited to any particular theory. The thick barrier layer tends to dominate the spectrum of the reflection 60 201249598, the masking difference in the thickness of the barrier layer itself, and any underlying layers. As the grinding progresses, the barrier layer becomes thinner or completely removed, and the reflected spectrum becomes more sensitive to changes in the underlying thickness. Therefore, when the barrier layer is removed, the spectral dispersion may increase. When Tian's month clears the upper layer, various algorithms can be used to detect the nature and variation of the dispersion value. For example, 'the sequence dispersion value can be compared with the threshold value, and if the dispersion value exceeds the threshold value, the signal indication is cleared. The upper layer. As a further example, the slope of the portion of the sequence of discrete values within the moving window can be calculated and if the slope exceeds the threshold, a signal is generated indicating that the upper layer 0 has been cleared as part of the algorithm in the increase in the spread. In order to remove high frequency noise, the sequence dispersion value can be dominated by m (eg, low pass or band filter). Examples of low pass filters include flow average and Butterworth (Bu Ttterworth) filter. Although the above discussion focuses on detecting the removal of the barrier layer, in the context of straightening, the (4) can be removed, for example, in the use of dielectric layer stack g (eg 'interlayer dielectric mine' IT pj〗 ) + σ package) removal of the upper layer in another type of semiconductor process, or dielectric 屛μ, removal of the ruthenium metal layer above the electrical layer. In addition to being used as a trigger for the feature tracking as described above, This technique for predicting the removal of the upper layer can be used for the grinding operation, for example, as the end point signal itself, to anchor the timer so that after the exposure of the lower layer, the predetermined time period is asked. Speak the ground layer along τ w 4 or use it as a trigger to change the grinding parameters (for example, changing the load head pressure or slurry composition after the lower layer of the road)
S 61 201249598 另外儘官以上討論假定存在具有安裝在平臺中之光 學終點監視器之旋轉平臺，但是“可適用於監視系統 與基板之間的其他類型相對運動。舉例而言，在一些實 細中（例如，軌道運動），光源橫越基板上之不同位置， +疋不跨過基板之邊，緣。在此等情況下，仍可將所收集 之光》曰刀、.且，例如，可在某一頻率處收集光譜及在時間 週期内收集之光譜視為一組之部分。時間週期應當足夠 長使彳于對於每一組可收集五至二十個光譜。 如在本說明書中所使帛’術語基板可包括，例如，產 基板（例如，產品基板包括多個記憶體或處理器晶 片）、測試基板、裸基板及閘控基板。基板可處於積體電 路製造之各種階段，例如，基板可為裸晶圓，或基板可 包括一或更多個沈積層及/或圖案化之層。術語基板可包 括圓盤及矩形片材。 可在數位電子電路中，或在計算機軟體、勒體或硬體 (包括此說明書中所揭示之結構性構件及該等結構性構 件之結構性等同物）中’或在上述者之組合中執行此說 明書中所I之本發明之實施例及所有功能操作。本發明 之實施例可執行為-或更多個電腦程式產品，亦即，藉 由資料處理設備（例如’可程式化處理器、電腦或多個 處理器或多個電腦）執行之機器可讀取儲存媒體中切實 包括之-或更多個電腦程式或用於控制資料處理設備之 操作之機器可讀取儲存媒體中切實包括之—或更多個電 腦程式。可以包括編譯或解釋語言在内之任何形式之程 62 201249598 式設計語言編寫電腦程式（亦稱a 、、 柄馮裎式、軟體、軟體應 用程式或碼），及可以任何形式部署 程式，包括作為 獨立程式或作為模組、組件、次堂 人⑦式或適用於計算環境 中之其他單元。電腦程式不必要對岸 了應於檔案。程式可儲 存在保持其他程式或資料之檔荦 系<σΡ分中、可儲存在用 於所述程式之單一檔案中或可儲在 a J储存在多個坐標檔案（例 如’儲存-或更多個模組、次程式或碼部分之槽案）中。 可部署電腦程式以在位於一個地點處執行該電腦程式或 跨過多個地點分散該電腦程式及藉由通訊網路互連該電 腦程式。此說明書中所述之製程及邏輯流程可藉由執行 一或更多個電腦程式之一或更多個可程式化處理器執行 以藉由在輸入資料上操作及產生輸出來執行函數。製程 及邏輯流程可藉由專用邏輯電路（例如，fpga(現場可 程式化閘陣列）或ASIC (特定應用積體電路））執行， 及設備亦可執行為專用邏輯電路（例如，FPGA(現場可 程式化閘陣列）或ASIC (特定應用積體電路））。 第17圖圖示對於具有TE〇s層之不同厚度之基板，使 用互相關性之方法匹配光譜之指數跡線与使用平方差之 和之方法匹配光譜之指數跡線（最佳匹配參考光譜之指 數作為平臺旋轉之數量之函數）之比較。產生具有15〇〇 入厚度之Black Diamond層、130 A厚度之Blok層及厚 度為5200 A、5100 A或5000 A之TEOS層之堆疊之產 品基板之資料。產生具有厚度為52〇〇 A之te〇S層之參 考基板之參考庫。如藉由跡線17〇2所圖示，在產品基板S 61 201249598 In addition, the above discussion assumes that there is a rotating platform with an optical endpoint monitor installed in the platform, but "can be applied to monitor other types of relative motion between the system and the substrate. For example, in some real details (for example, orbital motion), the light source traverses different positions on the substrate, +疋 does not cross the edge of the substrate, the edge. In these cases, the collected light can still be found, and, for example, The spectrum collected at a certain frequency and the spectrum collected over a period of time are considered as part of a group. The time period should be long enough to allow for the collection of five to twenty spectra for each group. The term substrate may include, for example, a substrate (eg, a product substrate including a plurality of memory or processor wafers), a test substrate, a bare substrate, and a gated substrate. The substrate may be at various stages of integrated circuit fabrication, for example, The substrate may be a bare wafer, or the substrate may include one or more deposited layers and/or patterned layers. The term substrate may include a disk and a rectangular sheet. Or in computer software, orthography or hardware (including structural members disclosed in this specification and structural equivalents of such structural members) or in the combination of the above, Embodiments of the present invention and all functional operations. Embodiments of the present invention may be implemented as - or more computer program products, that is, by a data processing device (eg, 'programmable processor, computer, or multiple processing Or a computer executing the machine readable storage medium - or more computer programs or machine readable storage medium for controlling the operation of the data processing device - or more Computer program. Any form of programming that can be compiled or interpreted. 62 201249598 Design language programming computer program (also known as a, 裎 裎, software, software application or code), and can be deployed in any form , including as a stand-alone program or as a module, component, sub-clan 7 or other unit in the computing environment. In the file, the program can be stored in a file system that maintains other programs or data, can be stored in a single file for the program, or can be stored in a J coordinate file (for example, 'storage - or a plurality of modules, sub-programs or code parts). A computer program can be deployed to execute the computer program at a location or to distribute the computer program across multiple locations and interconnect the network via a communication network Computer program. The process and logic flow described in this specification can be performed by executing one or more computer programs or one or more programmable processors to perform functions by operating on input data and generating output. Process and logic flow can be performed by dedicated logic circuits (eg, fpga (field programmable gate array) or ASIC (application-specific integrated circuit)), and devices can also be implemented as dedicated logic circuits (eg, FPGA (on-site) Programmable gate array) or ASIC (application-specific integrated circuit). Figure 17 is a diagram showing the matching of the exponential trace of the matching spectrum with the sum of the squared differences using the method of cross-correlation for the substrate with different thicknesses of the TE〇s layer (the best matching reference spectrum) The comparison of the index as a function of the number of platform rotations. A data sheet of a product substrate having a Black Diamond layer of 15 inches in thickness, a Blok layer of 130 A thickness, and a TEOS layer having a thickness of 5200 A, 5100 A, or 5000 A was produced. A reference library of a reference substrate having a layer of te 〇 S having a thickness of 52 Å is produced. As shown by trace 17〇2, on the product substrate
S 63 201249598 及參考基板具有相同厚度（亦即，5200 A )之TEOS層 之情況下’兩個指數跡線重疊而無明顯差異。然而，在 產品基板具有厚度為5100人之TEOS層及參考基板具有 厚度為5200 A之TEOS層之情況下，使用平方差之和產 生之指數跡線1 704與線性性質有一些偏離。相反，使用 互相關性產生之指數跡線與指數跡線1 7〇2重疊（及因此 在圖中不明顯）。最後，在產品基板具有厚度為5〇0〇 A 之TEOS層及參考基板具有厚度為52〇〇 A之teOS層之 情況下’使用平方差之和產生之指數跡線1 7〇6與線性性 質及跡線1 702有著明顯偏離，而使用互相關性產生之指 數跡線1 708通常保持線性及更靠近跡線丨7〇2。總之， 由此可見當下層厚度發生改變時使用互相關性決定最佳 匹配光谱產生跡線’該跡線更好地匹配理想者。 可經應用以減少電腦處理之方法將限制經搜索用於匹 配光譜之庫之部分。庫通常包括比研磨基板時將獲得之 光譜祀圍更寬廣之光譜範圍。在研磨基板期間，庫搜索 夂限於預先決定之庫光譜範圍。在一些實施例中，決定 正在研磨之基板之當前旋轉指數Νβ舉例而言，在最初 平里鉍轉中，可藉由搜索庫之所有參考光譜來決定 對於後續旋轉期間獲得的光譜，在Ν之自由度範圍内搜 索庫。亦即，若在一個旋轉期間發現指數為Ν，在後來 有X個旋轉之後續旋轉期間，其中自由度為γ，則將搜 索之範圍為（Ν+Χ)·~Υ至（Ν+χ) + γ。In the case where the S 63 201249598 and the reference substrate have TEOS layers of the same thickness (i.e., 5200 A), the two index traces overlap without significant difference. However, in the case where the product substrate has a TEOS layer having a thickness of 5,100 and the reference substrate has a TEOS layer having a thickness of 5,200 Å, the exponential trace 1 704 produced using the sum of the squared differences has some deviation from the linear property. In contrast, the exponential traces produced using cross-correlation overlap with the exponential traces 1 7〇2 (and therefore are not apparent in the figure). Finally, in the case where the product substrate has a TEOS layer having a thickness of 5 〇 0 〇 A and the reference substrate has a te OS layer having a thickness of 52 〇〇 A, the exponential trace generated by the sum of the squared differences 1 7 〇 6 and linear properties Trace 1 702 has a significant deviation, while exponential trace 1 708 generated using cross-correlation typically remains linear and closer to trace 丨7〇2. In summary, it can be seen that the cross-correlation is used to determine the best matching spectral generation trace when the thickness of the underlying layer changes. This trace better matches the ideal. The method that can be applied to reduce computer processing will limit the portion of the library that is searched for matching spectra. The library typically includes a broader spectral range than the spectral range that would be obtained when the substrate was polished. During the polishing of the substrate, the library search is limited to a predetermined library spectral range. In some embodiments, determining the current rotation index Νβ of the substrate being grounded, for example, in the initial flat rotation, the spectrum obtained during the subsequent rotation can be determined by searching all the reference spectra of the library, Search the library within the range of degrees of freedom. That is, if the index is found to be Ν during one rotation, and then there is a subsequent rotation of X rotations, wherein the degree of freedom is γ, the range of the search is (Ν+Χ)·~Υ to (Ν+χ) + γ.
S 64 201249598 上述研磨設備及方法可靡 用於各種研磨系統中。研磨 墊或承载頭或兩者均可S 64 201249598 The above grinding apparatus and method can be used in various grinding systems. Grinding pad or carrier head or both
移動M提供研磨表面与基板之間 的相對運動。舉例而言’ A 至可，凡軌道移動而不是旋轉p 研磨墊可為緊固至平臺 ^ 堂之®形（或其他形狀）墊。終點 γ貞測系統之^—此態槐可座田γ 1 々 "' 僳了應用至線性研磨系統’例如，在 該等線性研磨系統中研磨墊為線性移動之連續或捲軸至 捲㈣。研磨層可為標準（例如，具有或不具有填料 之聚氨基τ ^ )研磨材料'軟材料或固定研磨材料。 使用相對定位術語；應理解，可在垂直方向或其他方向 固持研磨表面及基板。 已經描述本發明之特定實施例。其他實施例落入以下 申凊專利範圍之范畴内。 圖式簡單說明】 第1Α圖至第1C圖為研磨之前、研磨期間及研磨之後 基板之概要性剖視圖。 第2圖圖示研磨設備之實例之概要性剖視圖。 第3圖圖示具有多個區域之基板之概要性俯視圖。 第4圖圖示研磨墊之俯視圖及圖示基板上進行原位量 測之位置。 第5圖圖示來自原位光學監視系統之量測之光雄。 弟6圖圖不參考光譜庫〇 苐7圖圖不指數跡線。 第8圖圖示指數跡線，該指數跡線具有偵測到上層之 65 201249598 清除後與所收集之指數值擬合之線性函數。 第9圖為用於製造基板及偵測研磨終點之實例製程之 流程圖。 第1 〇圖圖示複數個指數跡線。 第11圖圖示基於參考區域之指數跡線到達目標指數 之時間計算複數個可調整區域之複數個所欲斜率。 第12圖圖示基於參考區域之指數跡線到達目標指數 之時間計算終點。 第13圖為貫例製程之流程圖’該實例製程用於調整複 數個基板中之複數㈣域之研磨速率錢得複數個區域 在目標時間具有大約相同厚度。 第14圖圖示m貞測上層之清除之流程圖。 第15A圖圖示研磨開始時在單一拂掠期間所收集之光 譜圖。 第15B圖圖示在靠近阻障清除之單—拂掠期間所收集 之光譜圖。 第1 6圖圖示作為研磨丰^ ~ 5雄于間之函數之光譜之標準差圖。 第 1 7圖為圖不用 用於决疋最佳匹配參考光譜之不同技 術之比較圖。 第18圖為圖示柯西產生之折射率分散模型之實例。 第19圖為圖不厚度追縱與11值浮動模型之更好光譜擬 弟2 0圖圖示古· p 圆口下先仃進至層堆疊中之示意圖。 在各個圖式中相同元件含Movement M provides relative motion between the abrasive surface and the substrate. For example, 'A to OK, where the track moves rather than the rotating p-grinding pad can be fastened to the platform's (or other shape) pad. The end point of the gamma detection system is that the field can be applied to a linear grinding system. For example, in such linear grinding systems, the polishing pad is linearly moving continuous or reel to roll (4). The abrasive layer can be a standard (e.g., polyamino τ^ with or without filler) abrasive material 'soft material or fixed abrasive material. Relative positioning terms are used; it should be understood that the abrasive surface and substrate can be held in a vertical or other orientation. Specific embodiments of the invention have been described. Other embodiments fall within the scope of the following claims. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 to Fig. 1C are schematic cross-sectional views of a substrate before, during, and after polishing. Fig. 2 is a schematic cross-sectional view showing an example of a grinding apparatus. Figure 3 illustrates a schematic top view of a substrate having a plurality of regions. Fig. 4 is a plan view showing the polishing pad and the position where the in-situ measurement is performed on the substrate. Figure 5 illustrates the measurement from the in-situ optical monitoring system. Brother 6 is not referring to the spectral library 〇 图 7 graph is not an exponential trace. Figure 8 illustrates an exponential trace with a linear function that detects the upper layer's 65 201249598 clearance and fits the collected index value. Figure 9 is a flow diagram of an exemplary process for fabricating a substrate and detecting the end of the polishing. Figure 1 illustrates a number of exponential traces. Figure 11 illustrates the calculation of a plurality of desired slopes for a plurality of adjustable regions based on the time at which the exponential trace of the reference region reaches the target index. Figure 12 illustrates the calculation of the end point based on the time at which the exponential trace of the reference region reaches the target index. Figure 13 is a flow chart of a conventional process. The example process is used to adjust the polishing rate of the complex (four) domain in a plurality of substrates. The plurality of regions have approximately the same thickness at the target time. Figure 14 illustrates a flow chart for the removal of the upper layer. Figure 15A is a diagram showing the spectra collected during a single sweep at the beginning of the grinding. Figure 15B illustrates a spectrum of spectra collected during a single-sweep near the barrier clearing. Figure 16 shows a standard deviation diagram of the spectrum as a function of the polishing. Figure 17 is a comparison of the different techniques that are not used to determine the best matching reference spectrum. Figure 18 is a diagram showing an example of a refractive index dispersion model produced by Cauchy. Figure 19 is a diagram showing the better spectral spectroscopy of the thickness-free and 11-value floating model. Figure 20 shows the schematic diagram of the ancient p-round into the layer stack. The same components are included in each drawing
件符唬及名稱指示相F 66 201249598 【主要元件符號說明】 10 基板 10a 基板 12 基本結構 14 傳導層 16 鈍化層 18 下部介電層 20 I虫刻終止層 22 上部介電層 24 附加層 26 阻障層 28 傳導材料 1.00 研磨設備 108 窗口 110 研磨墊 112 外部研磨層 114 背托層 118 固體窗口 120 平臺 121 電動機 124 驅動轴 125 中心轴 128 凹口 129 旋轉耦合器 130 埠 132 研磨液 140 承載頭 142 擋圈 144 撓性薄膜 146a 腔室 146b 腔室 146c 腔室 148a 區域 148b 區域 148c 區域 150 支撐結構/旋轉料架 152 驅動軸 154 承載頭旋轉電動機 155 承載頭中心軸 160 原位光學監視系統 162 光源 164 光偵測器 166 電路 168 光學頭 170 分叉光纖F 2012 名称 and name indication phase F 66 201249598 [Main component symbol description] 10 substrate 10a substrate 12 basic structure 14 conductive layer 16 passivation layer 18 lower dielectric layer 20 I insect stop layer 22 upper dielectric layer 24 additional layer 26 resistance Barrier 28 Conductive Material 1.00 Abrasive Device 108 Window 110 Abrasive Pad 112 External Abrasive Layer 114 Backing Layer 118 Solid Window 120 Platform 121 Motor 124 Drive Shaft 125 Center Shaft 128 Notch 129 Rotary Coupler 130 埠 132 Grinding Fluid 140 Carrier Head 142 Retaining ring 144 flexible film 146a chamber 146b chamber 146c chamber 148a region 148b region 148c region 150 support structure / rotating rack 152 drive shaft 154 carrier head rotating motor 155 carrier head central axis 160 in situ optical monitoring system 162 light source 164 Photodetector 166 circuit 168 optical head 170 bifurcated fiber
S 67 908b 201249598 172 主線 176 支線 201 位置 201b 點 " 20 Id 點 20 If 點 20 lh 點 20 lj 點 202a 光譜 203 a 光譜 204 光譜 204b 光譜 212 指數值 220 序列 224 線 232 指數值 300 光譜 312 指數值 3 20 參考光譜 340 記錄 902 步驟 904 步驟 908 步驟 步驟 174 支線 190 控制器 201a 點 201c 點 201e 點 201g 點 201i 點 201k 點 202b 光譜 203b 光譜 204a 光譜 210 指數跡線/序列 214 線/線性函數 222 指數值 230 序列 234 線 3 10 庫 3 14 線性函數 330 指數值 350 資料庫 903 步驟 906 步驟 908a 步驟 910 步驟 68 201249598 912 步驟 914 步驟 916 步驟 918 步驟 1300 流程圖 1302 步驟 1304 步驟 1306 步驟 1308 步驟 13 10 步驟 1312 步驟 1314 步驟 1316 步驟 1318 步驟 1330 步驟 1400 方法 1402 步驟 1404 步驟 1406 步驟 1408 步驟 1500a 光譜 1500b 光譜 1600 圖 1602 點 1610 時間週期 1702 指數跡線 1704 指數跡線 1706 指數跡線 1708 指數跡線 1810 光 1820 光 1830 光 EI2 指數 EI3 指數 S' 斜率 SD2 斜率 SD3 斜率 SD4 斜率 69S 67 908b 201249598 172 Main line 176 Branch line 201 Position 201b Point " 20 Id point 20 If point 20 lh Point 20 lj Point 202a Spectrum 203 a Spectrum 204 Spectrum 204b Spectrum 212 Index value 220 Sequence 224 Line 232 Index value 300 Spectrum 312 Index value 3 20 Reference Spectrum 340 Record 902 Step 904 Step 908 Step Step 174 Branch Line 190 Controller 201a Point 201c Point 201e Point 201g Point 201i Point 201k Point 202b Spectrum 203b Spectrum 204a Spectrum 210 Index Trace/Sequence 214 Line/Linear Function 222 Index Value 230 Sequence 234 Line 3 10 Library 3 14 Linear Function 330 Index Value 350 Library 903 Step 906 Step 908a Step 910 Step 68 201249598 912 Step 914 Step 916 Step 918 Step 1300 Flowchart 1302 Step 1304 Step 1306 Step 1308 Step 13 10 Step 1312 Step 1314 Step 1316 Step 1318 Step 1330 Step 1400 Method 1402 Step 1404 Step 1406 Step 1408 Step 1500a Spectrum 1500b Spectrum 1600 Figure 1602 Point 1610 Time Period 1702 Index Trace 1704 Index Trace 1706 Index Trace 170 8 Index Trace 1810 Light 1820 Light 1830 Light EI2 Index EI3 Index S' Slope SD2 Slope SD3 Slope SD4 Slope 69
Claims (1)
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US13/096,777 US20120278028A1 (en) | 2011-04-28 | 2011-04-28 | Generating model based spectra library for polishing |
| US13/096,714 US8942842B2 (en) | 2011-04-28 | 2011-04-28 | Varying optical coefficients to generate spectra for polishing control |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| TW201249598A true TW201249598A (en) | 2012-12-16 |
| TWI574787B TWI574787B (en) | 2017-03-21 |
Family
ID=47072999
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| TW101113982A TWI574787B (en) | 2011-04-28 | 2012-04-19 | Varying coefficients and functions for polishing control |
Country Status (4)
| Country | Link |
|---|---|
| JP (1) | JP6030636B2 (en) |
| KR (1) | KR101981814B1 (en) |
| TW (1) | TWI574787B (en) |
| WO (1) | WO2012148716A2 (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI626121B (en) * | 2013-11-27 | 2018-06-11 | 美商應用材料股份有限公司 | Computer program product, computer-implemented method, and polishing system for adjusting polishing rates during substrate polishing with predictive filters |
| CN111886686A (en) * | 2018-09-26 | 2020-11-03 | 应用材料公司 | Compensation of substrate doping in edge reconstruction for in-situ electromagnetic induction monitoring |
| TWI784719B (en) * | 2016-08-26 | 2022-11-21 | 美商應用材料股份有限公司 | Method of obtaining measurement representative of thickness of layer on substrate, and metrology system and computer program product |
| TWI809389B (en) * | 2020-06-08 | 2023-07-21 | 美商應用材料股份有限公司 | System, method and computer porgram product for profile control during polishing of a stack of adjacent conductive layers |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9352440B2 (en) * | 2014-04-30 | 2016-05-31 | Applied Materials, Inc. | Serial feature tracking for endpoint detection |
| US9573243B2 (en) * | 2014-11-04 | 2017-02-21 | Headway Technologies, Inc. | Method for adaptive feedback controlled polishing |
| US11951587B2 (en) * | 2018-09-26 | 2024-04-09 | Taiwan Semiconductor Manufacturing Co., Ltd. | Zone-based CMP target control |
Family Cites Families (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW374050B (en) * | 1997-10-31 | 1999-11-11 | Applied Materials Inc | Method and apparatus for modeling substrate reflectivity during chemical mechanical polishing |
| US6242136B1 (en) * | 1999-02-12 | 2001-06-05 | Corning Incorporated | Vacuum ultraviolet transmitting silicon oxyfluoride lithography glass |
| US6491569B2 (en) * | 2001-04-19 | 2002-12-10 | Speedfam-Ipec Corporation | Method and apparatus for using optical reflection data to obtain a continuous predictive signal during CMP |
| IL145699A (en) * | 2001-09-30 | 2006-12-10 | Nova Measuring Instr Ltd | Method of thin film characterization |
| JP3865637B2 (en) * | 2002-01-15 | 2007-01-10 | 株式会社ルネサステクノロジ | Manufacturing method of semiconductor device |
| KR101593927B1 (en) * | 2005-08-22 | 2016-02-15 | 어플라이드 머티어리얼스, 인코포레이티드 | Apparatus and methods for spectrum based monitoring of chemical mechanical polishing |
| JP5534672B2 (en) * | 2005-08-22 | 2014-07-02 | アプライド マテリアルズ インコーポレイテッド | Apparatus and method for spectrum-based monitoring of chemical mechanical polishing |
| US7495781B2 (en) * | 2006-07-10 | 2009-02-24 | Tokyo Electron Limited | Optimizing selected variables of an optical metrology model |
| US20090275265A1 (en) * | 2008-05-02 | 2009-11-05 | Applied Materials, Inc. | Endpoint detection in chemical mechanical polishing using multiple spectra |
| JP5774482B2 (en) * | 2008-10-27 | 2015-09-09 | アプライド マテリアルズ インコーポレイテッドＡｐｐｌｉｅｄ Ｍａｔｅｒｉａｌｓ，Ｉｎｃｏｒｐｏｒａｔｅｄ | Goodness of fit in spectral monitoring of substrates during processing |
| US8295967B2 (en) * | 2008-11-07 | 2012-10-23 | Applied Materials, Inc. | Endpoint control of multiple-wafer chemical mechanical polishing |
-
2012
- 2012-04-13 WO PCT/US2012/033677 patent/WO2012148716A2/en active Application Filing
- 2012-04-13 JP JP2014508401A patent/JP6030636B2/en active Active
- 2012-04-13 KR KR1020137031251A patent/KR101981814B1/en active IP Right Grant
- 2012-04-19 TW TW101113982A patent/TWI574787B/en active
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI626121B (en) * | 2013-11-27 | 2018-06-11 | 美商應用材料股份有限公司 | Computer program product, computer-implemented method, and polishing system for adjusting polishing rates during substrate polishing with predictive filters |
| TWI784719B (en) * | 2016-08-26 | 2022-11-21 | 美商應用材料股份有限公司 | Method of obtaining measurement representative of thickness of layer on substrate, and metrology system and computer program product |
| US11682114B2 (en) | 2016-08-26 | 2023-06-20 | Applied Materials, Inc. | Thickness measurement of substrate using color metrology |
| CN111886686A (en) * | 2018-09-26 | 2020-11-03 | 应用材料公司 | Compensation of substrate doping in edge reconstruction for in-situ electromagnetic induction monitoring |
| TWI809389B (en) * | 2020-06-08 | 2023-07-21 | 美商應用材料股份有限公司 | System, method and computer porgram product for profile control during polishing of a stack of adjacent conductive layers |
| TWI810069B (en) * | 2020-06-08 | 2023-07-21 | 美商應用材料股份有限公司 | System, method and computer porgram product for profile control during polishing of a stack of adjacent conductive layers |
| US11850699B2 (en) | 2020-06-08 | 2023-12-26 | Applied Materials, Inc. | Switching control algorithms on detection of exposure of underlying layer during polishing |
| US11865664B2 (en) | 2020-06-08 | 2024-01-09 | Applied Materials, Inc. | Profile control with multiple instances of contol algorithm during polishing |
Also Published As
| Publication number | Publication date |
|---|---|
| KR101981814B1 (en) | 2019-05-23 |
| KR20140028028A (en) | 2014-03-07 |
| WO2012148716A2 (en) | 2012-11-01 |
| TWI574787B (en) | 2017-03-21 |
| WO2012148716A3 (en) | 2012-12-20 |
| JP2014512704A (en) | 2014-05-22 |
| JP6030636B2 (en) | 2016-11-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| TW201249598A (en) | Varying coefficients and functions for polishing control | |
| KR101892914B1 (en) | Fitting of optical model to measured spectrum | |
| US9573242B2 (en) | Computer program product and method of controlling polishing of a substrate | |
| TWI521625B (en) | Detection of layer clearing using spectral monitoring | |
| KR101867385B1 (en) | Building a library of spectra for optical monitoring | |
| US20120278028A1 (en) | Generating model based spectra library for polishing | |
| TW200849416A (en) | Using spectra to determine polishing endpoints | |
| TWI465314B (en) | Construction of reference spectra with variations in environmental effects | |
| TW201223702A (en) | Techniques for matching measured spectra to reference spectra for in-situ optical monitoring | |
| TWI546154B (en) | Fitting of optical model with diffraction effects to measured spectrum | |
| US8942842B2 (en) | Varying optical coefficients to generate spectra for polishing control | |
| TW201205703A (en) | Dynamically or adaptively tracking spectrum features for endpoint detection | |
| TW201205704A (en) | Control of overpolishing of multiple substrates on the same platen in chemical mechanical polishing | |
| KR20130093099A (en) | Tracking spectrum features in two dimensions for endpoint detection | |
| TW201213050A (en) | Spectrographic monitoring using index tracking after detection of layer clearing | |
| JP5774482B2 (en) | Goodness of fit in spectral monitoring of substrates during processing | |
| TW201208812A (en) | Endpoint control of multiple substrates with multiple zones on the same platen in chemical mechanical polishing | |
| TWI223366B (en) | Method of judging residual film by optical measurement | |
| TW201141662A (en) | Automatic generation of reference spectra for optical monitoring | |
| JP5376293B2 (en) | End point detection device and polishing device | |
| JP2013252613A (en) | End point detection device and polishing apparatus |