United States Patent 106477432B1 a (10) Patent No: US 6,477,432 BL Chen et al. (45) Date of Patent: Nov. 5, 2002 (58) STATISTICAL IN-PROCESS QUALITY A490 at a oa CONTROL SAMPLING BASED ON ose Miter vin PRODUCT STABILITY THROUGH A A+ i Laetal soor0 SYSTEMATIC OPERATION SYSTEM AND A ioe avin METHOD + cited by examiner (75) toventors: Shun-An Chen, Kaoshiong (HW): Fang Hsu Taipei (IW), Vinge Hsu, Hsin-Chu (TW); Constance Y. Primary Examiner—Thomas Black Ch Hn (2) Assists Examiner—Ciysal 3. Baroes (78) Atom, Agen or Firm George Oeil; Stephen B (73) Assignee: ‘Taiwan Semiconductor ‘Ackerman; Graham S. Jones, II Manufacturing Company, tio-on (rw) 6 ABSTRACT (+) Novee: Subject aay disclaimer, he emt ths Aste for managing quit cont in « manutieing pate is extended adjusted under 33. plant for processing les of workin provess (WIP) fra les USC. 134) yO cays tne produc, comprises « manufhcuring proces which tntoes 2 manufacturing exeetive syste (MES) which (21) Appl. No. 09/480,269 provides inspection data to a statistical process control (SPC) datibee,andan SPC analyaer for analysing the (22) Filed: Jan. 11, 2000 inspection data and providing a sampling rate rule output to 6) mc! Gosn 1vo2 a sroling rate dautaseAerver sapling ate ©) USC. Fons1; Tose nde as tle the MES. The MES tens conion aso weber scr sewrcn ra 00 WIP sould be sampled I the conden so met hen LOIS0-S15 100, gid an iapet contol ial for aapecton othe lant Fevinspecting tn 1 Ith condiion snotmet then branch 60) References Cited away rom ie conta sigh to provide a alratve Snot Suna pane tthe et cess pi th pt US. PATENT DOCUMENTS 5180289 A 91002 Padus ro 20 Claims, § Drawing Sheets a es ¥ ‘Lo fiSase JH sec | T 2] seo —_)_ f ‘SAMPLING | 1 Ly RATE MANUFACTURING oataase || | extddre‘Sverew a2] 3 tor [PROCESS } haxcrine_| [prostoure RATE 34 | st NOT NsL GRAPHICAL \ {SAMPLED SAMPLED USER LoTs: Lors J sa[PROGESE] [PROCESS], son See | [Sees [PROCESS] [PROCESS] 1 STEP STEP, ‘4 TeECTION acKerouna| 17-f SEEH] | arancn |f-r9 ‘SERVER: PROCESS| [PROCESS | 20 step step [20 7} 221"Step | "Stee | By =U.S. Patent Nov. 5, 2002 Sheet 1 of 5 US 6,477,432 B1 10~GTARD 12~4, PROCESS STEP 14~4 PROCESS STEP 17~j1PQC INSPECTION STEP 20~;, PROCESS STEP 22~—, PROCESS STEP 24 ENO. FIG. 7 PROCESS. INSPECTION STABILITY sAMereNe Low | ZZ | ZZACZAWZA|_ 1.100%) MIDDLE ZZ ZA 1/2 (50%) HIGH ZZ f 1/3(33%) MIDDLE Z 1/2 (50%) FIG. 2U.S. Patent Nov. 5, 2002 Sheet 2 of 5 US 6,477,432 BL SPC ¢ [DATABASE fF SPC ( ANALYZER EW b—FD 29 SRD M SAMPL ING ( RATE MANUF ACTUR I NG [| DATABASE EXECUTIVE SYSTEM 324 31 LOT PROCESS ) TRACKING | |PROCEDURE SAMPL ING RATE eA SL NOT NSL GRAPHICAL SAMPLED SAMPLED USER LOTS LOTS INTERFACE 12] PROCESS PROCESS |_,, ow STEP STEP f- PROCESS PROCESS |_, 14 sTeP step [- INSPECTION L BACKGROUND \} 17] step eranci || "9 SERVER PROCESS] [PROCESS |. 20 step STEP PROCESS PROCESS BS S| —y 4 224, ster step [22 36 FIG. 8US 6,477,432 B1 Sheet 3 of 5 Nov. 5, 2002 bv COTA [esoig [Addy dIM }UasIND pud yO7 MENC 60 80 40 90 sO *O sO 720 1000 Aug 107 eNO uDds }98jap YIy Peau YO!YM P| JO] JO JaqUNN | 101 ay] 399195 OL ainy men Addy 0999N1 [95] 71N058Q vOVdXV d3ad SSvd {Zt eC9SW 1 | SC 6°S'L PWENXY TWATNIL LL LCOSWL | 7¢} yINoYEg VvSVNXV 1 4V~PWN OL LOQOWL | Se 71NOsEqQ vAavNxy ISV~PN| 6 WVYS WNSt 0} 6 LySOW {Zo} y1NB,EqQ VSCWNXY 14V-EW] 8 91907 WNOS O}| 3g BSPrOW1 | LS] y1NB,aqQ vIACNxV ISV7EN] Z WVYS WNOS 0} Z CVPONL | OF} yINoFeG| FSZNXV Tiv-2n| 9| [01907 wnse OF 9 91¥9NI [6c y1INDYEQ VACANXV ISV~ZN| S WVYYS WNSE O} S| LECOWL |8Z yINoFeG| PSINXV Tav-tw|_+| | 91907 WNOS O| 4 vEcOni [Ze| y1NOJEQ vIlWxyv ISV~LW]| ¢ 919071 WNSZ Oj ¢ O9COW1L |9Z| yINoyeg| VdELXV Sid@ ail | ¢ wvyd WNrz O| Z| OvCOWL |SZI 6's't | vNwOxXv | INNVT3LVS] 1! wvas 19 WNSZ O| 1 SSCON! |r| gina | aiaoay BILL A9010NHO 3) fp 301A30 ad1oey K60 | 0uydeL Bo1Aag a1ny 307 3sog Sin 707 Ag sun, a1ny ebuouD Buiuny 8yoy bu! duos Win U.S. PatentGS “Old dim 1UaIIND PUD yOT MONO [esoig [Kiddy A\ug 307 maNO 60 80 40 9050 +70 ©0 20 '!000 US 6,477,432 B1 Sheet 4 of 5 Nov. 5, 2002 U.S. Patent OL ajny man Kiddy upos 199)3P VI¥ P9aU YO!4M P| 307 JO JAQUAN | 101 a4, 399195 6'8°L'9'S'¥'e'7'1'0 | BNZdXV IdvV~Zd |¢l LESZW 1 | 64] 6dldxXV Tav-td[2t FOSZM1 [8 6'8'L'9'S'¥'E'7'1'0 | GALAXY ISWTLd] EL serZWl |Z7 y1NBJaQ 6dOWXV| !0V~ZOLWIOL Z9OCLWL | 94) 6°8°L'9'S"¥'E'2'1'0 | BASWXV ISV~CW] 6 WHS WNSt O} 6 cOCLWL |S7| 6dZNXV tav~zw|] ef 91907 wnos OT 8] T9SZnI [ov SRLISH ECO | GICNXV isw-en| Z Wes Wnos O| Z Z¥EZNL [EP 31N049qQ Bd LWXV lav LW] 9 91901 WASS O} YQ OS LLWL | ZY eBLISHE C10 | 6ILNXV Isv-2n] S| wus Wnse O| S| TIS9Ni [14 y1NOJEQ 6BWLLXV] IdV¥~LOd! | 4] 91901 WNOE O} F SOZSWL | 04] y1NDJaQ 6dZOxv !dv Zo} ¢ 91901 WAS O} ¢| L699WL | 6S) y1NoyEq 6d LOxv lav7tol Z| WvYG WNYZ O}| Z| 9999W1 | BE) 6'8'Z'9'S*¥'e'Z'L'0 | GGNAXV d3dq-Zyus] ifWvyeS 19 WNSZ O} L S99ON1 | Z2¢] 31ny aido3ay SqLtt ASO 1ONH93 1} DP 3O1A30 adioay A6o|ouydal ao1aaq ainy 307 asog eIny y07 Ag euny ainy ebuoyd Buran] 8yoy Bul Gus WIAU.S. Patent Nov. 5, 2002 Sheet 5 of 5 US 6,477,432 B1 50 \ CENTRAL COMPUTER SYSTEM 67 62 MON — 63 KB y L176 60 77 mee 72 7” RAM 75) MON 0 cPu 74 — os f ass KB Ks p>—86 L y 186 » SHOP FLOOR} — 87 FABRICATION PLANT 30 FIG. 6US 6,477,432 BL 1 STATISTICAL IN-PROCESS QUALITY CONTROL SAMPLING BASED ON PRODUCT STABILITY THROUGH A. SYSTEMATIC OPERATION SYSTEM AND METHOD BACKGROUND OF THE INVENTION 1. Field of the TIavention ‘This invention relates to quality control systems for a ‘manufacturing process and more particularly to inspection sampling methods and systems therefor. 2. Description of Related Art US. Pat, No, 5150,289 of Badavas for “METHOD AND APPARATUS FOR PROCESS CONTROL” shows a pro- ‘cess control system, A statistical process control system provides asymmetrical nonlinear automatic closed-loop feedback control. The system applies to the control of ‘equipment that responds 10 a controlled variable signal to vary a measurable characteristic of a process. The system uses an accumulated deviation of a measured subgroup means minus a target value, divided by the subgroup stan- ‘dard deviation, to modify the manipulated variable cach time the accumulated value exceeds a decision interval above or below the target. The system permits the use of independent slack variables subtracted from the accumu- lated deviation to model the underlying process. more closely. The system permits the use of independent alarm ‘values and variable gains to permit greater process control, US. Pat, No. 5,497,331 of Iriki et al, for “SEMICON- DUCTOR INTEGRATED CIRCUIT DEVICE FABRICA- ‘TION METHOD AND ITS FABRICATION APPARATUS” discloses a quality control system for a semiconductor manufacturing line. A semiconductor integrated circuit ‘device fabrication technique improves the accuracy of ele~ ‘ment qualities by considering the influence of interaction of ‘element quality parameters in the quality control of semi- ‘conductor fabrication processes and also by improving the product yield estimation accuracy to improve production cfficiency. First an initial value of a membership function is set. Then element quality parameters and a combined quality parameter are expressed by membership funetions in fuzzy ‘control in a semiconductor fabrication apparatus for auto- ‘mating fabrication by connecting a computer with measur- ing instruments and processors by communication devices. ‘The combined quality parameters are fuzzy-inferred from the element quality parameters using these membership functions. Inference rules are adjusted by data ofthe actual processes. Membership functions of the obtained element ‘quality parameters are converted into an element quality control standard, and the semiconductor integrated circuit ‘device fabrication processes are controlled according to the standard US. Pat, No. 5,465,221 of Merat et al. for “AUTO- MATED PROCESS PLANNING FOR QUALITY CON- TROL INSPECTION” shows an automated process plan- ning for quality control inspections. A computer is used for generating a part inspection plan for a coordinate measuring machine, in a feature-based rapid design system, having @ Feature-Based Design Environment, an Episodal Associa- tive Memory, Fabrication Planning, and an Inspection Plan, ‘with features which include form features which define the form or shape of the part, manufacturing features, inspection features, and geometric and design features. An Inspection Plan includes interaction means wherein the inspector inter- acts with the system to guide it toa desired results and the as os ss 4s ss 6s 2 inspector can define setups, measurement points, sequence for the points, and the via points. A learning process is included so that desired sequence input from the inspector is sent 10 discovery means to organize patterns and to deine rules, This creates a sel-improving expert system by recall- ing relevant past experiences, and learning from the desired sequence input from the inspector. Commonly assigned U.S. Pat, No. 5,862,054 of Li for “PROCESS MONITORING SYSTEM FOR REAL TIME STATISTICAL PROCESS CONTROL” shows a method of ‘monitoring for real time process contiol. The method moni- tors process parameters from multiple process machines t0 provide real time statistical process control (SPC). The ‘implementation applied to ion implantation of wafers. The ‘method has applicability where there are a number of process machines having a number of process parameters fand close continuous sampling of data is required. The process parameters are collected on a single computer over a single network, and cach parameter is analyzed and displayed separately for each process and process machine. Statistical variables like Cp and Cpk are calculated and presented on the computer sereen along with graphs of the various parameters for a particular process machine. Data is ‘aged out of the computer to an archival database under the control of a manufacturing information system and con- nected to a company wide network. SUMMARY OF THE INVENTION ‘This invention teaches a statistical in-process quality ccontrol sampling system which adjusts the sampling rate dynamically, ‘There are many inspection steps defined in manufacturing, process to find defects when the process is being tested for the first time. In addition, for a new product, « heavy burden of inspection work is also necessary. However, for a mature product, ia order 10 save costs the inspection Sampling rate may be reduced. Objects of this invention are as follows: 1, Reduce inspection cost and keep high product quality. 2. Adjust the inspection sampling frequency by process stability dynamically. 3. Provide a systematic method for controlling the inspec- tion sampling frequency. In this invention, the stability of the process is considered to be a factor to be considered in selection of the appropriate Sampling rate seling. A systematic management and opera- tion method is also designed to guarantee that the dynamis sampling inspection steps are executed exactly as required {or optimum performance. ‘Advantages of this invention are as follows: 1. The invention provides systematic setting of the sampling Frequency rule based on process stability data to reduce inspection cost and keep high product quality 2. This is a systematic method to manage the inspection operation, In accordance with this invention, a system and method for managing quality control in a manufacturing plant for processing lots of work in process (WIP) for at least one product, comprises a manufacturing process which includes 4 manufacturing executive system (MES) which provides inspection data to a statistical process control (SPC) database, and an SPC analyzer for analyzing the inspection data and providing 2 sampling rate rule output to sampling rate database. A server supplies the sampling rate rule to the MES. The MES tests a condition as to whether a lot of WIP should be sampled. Ifthe condition is met, then provide an