IETE Technical Review
`
`ISSN: 0256-4602 (Print) 0974-5971 (Online) Journal homepage: https://www.tandfonline.com/loi/titr20
`
`Role of Automation and Robotics in
`Semiconductor Industry
`
`Parag Diwan & Dilip Kothari
`
`To cite this article: Parag Diwan & Dilip Kothari (1990) Role of Automation and
`Robotics in Semiconductor Industry, IETE Technical Review, 7:5-6, 368-377, DOI:
`10.1080/02564602.1990.11438675
`To link to this article: https://doi.org/10.1080/02564602.1990.11438675
`
`Published online: 02 Jun 2015.
`
`Submit your article to this journal
`
`Article views: 3
`
`View related articles
`
`Full Terms & Conditions of access and use can be found at
`https://www.tandfonline.com/action/journalInformation?journalCode=titr20
`
`Petitioner STMICROELECTRONICS, INC.,
`Ex. 1019, IPR2022-00681, Pg. 1
`
`

`

`Role of Automation and Robotics in Semiconductor
`Industry
`
`PARAG DIWAN AND DILIP KOTHARI
`Semiconductor Complex Ltd, S A S Nagar 160 051, India
`
`robotics in the semiconductor industry. The
`This paper presents the current trends in automation and
`semiconductor manufacturer, in order to survive in this industry, has to maintain a state-of-the-art facility and
`for that he has to invest in automation. The reasons for automation are myriad; principal among them are yield
`and throughput enhancement, improved process control, faster turnaround time and development of a sophisti(cid:173)
`cated test hed for research activities. Tbe paper dwells extensively on reasons, henefits and objectives of auto(cid:173)
`mation.
`
`Automation systems comprise of a distributed computer architecture; mechanization systems; process con(cid:173)
`trol systems and logistics systems.
`Implementation of automation systems should be carried out in a phased
`manner so as to achie,·e bottoms up integration. Both system buildup and implementation strategies are discussed
`in detail.
`
`In the Robotics section development of robots for clean room application is discussed. Starting from
`repeatable jobs robots have acquired advanced capabilities such as vision and intelligence.
`In this light the anato(cid:173)
`mical parts of clean room robots and special requirements for their construction are discussed. The futuristic
`trends such as guided vehicles are also surveyed.
`
`Towards the end case studies for automation as applied to wafer fab and device assembly ha,·e been presen(cid:173)
`ted along with the blueprint for automation at SCL.
`
`AUTOMATION
`
`SEMICONDUCTOR business is unique in the sense
`that as the design and technology upgrade, the price
`of the product comes down.
`In order to stay in the indus(cid:173)
`try it is imperative to go in for an automation system.
`This is also necessary because of the newly emerging trends
`in the semiconductor industry which force the manufac(cid:173)
`turers to remain competitive and maintain the state-of(cid:173)
`the-art facility (1 ]. These factors are (i) Increasing level
`of integration, (ii) Decreasing minimum feature size, (iii)
`Increasing process complexity, (iv) Thrust for quality,
`(v) Process control and yield optimization, and (v) Ease
`of operation.
`
`In this context we would like to quote from 1984
`SEMICON West Show's inaugral address of George
`Moore, SEMI Director, who said "As the industry moves
`towards higher densities of large scale and high speed
`drcuits, ... automating the fab line will result in tighter
`processing controls and higher yield" [2]. Yield is of
`paramount importance in this industry. Yield and auto(cid:173)
`mation go hand in hand. Thus automation is the only
`hope for the future.
`
`Reasons for automation
`
`The overriding reasons to automate fab lines and
`these reasons are true worldwide are to increase produc(cid:173)
`tivity and lower the cost. Productivity can be increased
`by using larger wafers, making denser circuits on them
`and having lower contamination and defect levels.
`
`Paper No. I 90-A; Copyright c.
`
`I 990 by JETE.
`
`Larger wafers are not only difficult to handle manually
`without damage, but because of their higher densities are
`more prone to the impact of contamination; for both
`reasons machines are better suited to handle them.
`
`The main reasons for automation are listed as under
`[3-5]
`
`( i) Complex processes and high density circuits re(cid:173)
`quire process consistency and repeatability. Only
`computer controlled machines can provide such
`features.
`
`(ii) Improvement in throughput are achieved by auto(cid:173)
`mated systems as they never change shifts and thus
`significantly increase equipment utilization time.
`
`(iii) Automated systems provide faster turn around
`time for fabrication of VLSI devices, as automated
`product queuing systems minimize the idle time
`during which wafers sit and collect defects.
`
`(iv) The investigation of relevant material, process,
`device and circuit limits in sub-micron VLSI
`research requires unprecedented degrees of pro(cid:173)
`cess control and monitoring. A flexible auto(cid:173)
`mated system offers the promise of fulfilling these
`needs. This gives an impetus to research and
`development activities by providing a "cutting
`edge", sophisticated and flexible test bed.
`
`(v) Automated systems promote paperless fab con(cid:173)
`cept, thus eliminating papers used in
`fab for
`maintaining records, logs, batch cards etc. This
`
`368
`
`Petitioner STMICROELECTRONICS, INC.,
`Ex. 1019, IPR2022-00681, Pg. 2
`
`

`

`DIW AN & KOTHARI : AUTOMATION AND ROBOTICS
`
`369
`
`would lead to ultraclean work environment of
`class 1, thus resulting in enhancement of yield.
`
`Considering all these benefits it would be impossible
`to survive in the business without investing in automation
`systems. In this regard there is an interesting quip by
`Doug Lockie, President of Strategic Technologies, Inc,
`who said, "As far as automation in the Semiconductor
`Industry, it looks like the Cobbler's children are finally
`getting shoes!", meaning
`that
`the technology being
`developed by semiconductor manufacturers is being used
`to aid the device manufacturing itself [2].
`
`Defining automation
`
`Automation in semiconductor industry can be defined
`as an integrated system of a number of subsystems which
`include mechanized movement system, information system,
`logistics system, process control system and people/pro(cid:173)
`cedures. All are linked via a host computer
`system
`which forms its backbone and base. Figure 1 shows
`the pegging order of such a sub system [6,7]._
`
`The backbone
`
`The fundamental backbone of an automation system
`should comprise of a set of software applications which
`can consist of programs to control mechanical, electrical
`and process parameters and sequencing of events. Thus
`the software programs consist of management information
`systems, expert systelll6, schedulers and work-in-progress
`tracking programs. The host computer generally
`is
`
`super mini (eg, VAX, Prime) system wherein all the soft(cid:173)
`ware packages would reside and run. All individual
`equipments and work area terminals would be networked
`to this host using SECS 1/11 protocols or through Local
`Area Hosts (LAH).
`
`Mechanization systems
`
`Such systems comprise of cassette-to-cassette transfer
`systems, robotic components and in some cases automatic
`guided vehicles. Details of such system would be covered
`in the section on robotics and while discussing case studies.
`
`Process control systems
`
`One aspect of such systems is automatic process con(cid:173)
`trol through which preprogrammed process parameters
`are fed to the machines and subsequently monitored by
`the computer system. Another aspect is closed loop
`automatic control which detects a process drift and
`generates a warning for the human operator. The third
`aspect is statistical process control which compiles statis(cid:173)
`tical data on various process parameters resulting in
`repeatability and consistancy in the process leading to
`higher yield.
`
`Logistics system
`
`The key to successful logistic system will be a master
`plan for controlling product routing and priority assign(cid:173)
`ment. Such system comprises of a scheduler program
`which allocates tasks to be completed during a shift and
`assigns jobs to equipment and operators. Some auto(cid:173)
`mation experts are of opinion that WIP tracking system
`are a subset of logistic systems. These systems are parti(cid:173)
`cularly helpful in knowing, "What is where?".
`
`People and procedures
`
`This could comprise of an expert system which would
`guide the work force in day to day operating procedures
`and to help in conducting yield improvement and process
`tuning experiments even in the absence of area experts.
`The knowledge base for such a system can be formulated
`from the experience of the experts in various areas of manu(cid:173)
`facturing [9].
`
`Objectives of automation
`
`Once the decision to automate has been made, criteria
`must be established for evaluating equipment and ability
`of suppliers to provide appropriate support. The most
`critical factor in making such a choice is that the system
`selected must be evolutionary and adaptable, thus lending
`itself to phased growth.
`
`While evolving an automation system strategy the
`following questions need to be answered [7].
`
`,Fig 1 Totem-pole representation of the elements of a semiconduc(cid:173)
`tor manufacturing automation system
`
`. How much change in the system is anticipated?
`
`Petitioner STMICROELECTRONICS, INC.,
`Ex. 1019, IPR2022-00681, Pg. 3
`
`

`

`370
`
`IETB TECHNICAL REVIEW. Vol. 7, Nos. 5 & 6, 1990
`
`How much adaptability must be built in?
`
`Operator training
`
`How flexible is the plan to evolve to full automation?
`
`Configurability for line integration
`
`In view of the above questions an analysis of proposed
`automation system should be carried out in the following
`format.
`
`Eliminating product handling by operators
`
`For efficiency and minimization of defects it is impera(cid:173)
`tive to eliminate direct handling of product by operators.
`Research on this has led to significant improvement in
`cassette-to-cassette handling system and a standard called
`'SMIF' has been developed for mechanical interface [81.
`
`Appropriate identification methodology
`
`Many technologies have been developed for the identi(cid:173)
`fication of wafers and cassettes. These include laser
`wafer titler and bar code system for cassettes.
`
`Computer control of process equipments
`
`Some degree of computer control is essential; the user
`should determine just how much. Some key areas to
`look for are, Automatic parameter setup, Multiple levels
`of equipment control, Conformity with SEMI SECS stan(cid:173)
`dards, and Automatic process data collection.
`
`Flexibility of CIM/CAM architecture
`
`The basic architecture of the overall CAM system
`should be carefully considered. An expert opinion is
`that a distributed architecture is most suitable as it works
`in a fail safe mode, has multiple control and is modular
`so that a bottoms up integration can be achieved.
`
`Protection from contamination
`
`The equipments used for automation should be such
`that they do not generate particulate contamination resul(cid:173)
`ting in killer defects on the product.
`
`Adequate equipment reliability
`
`Suppliers should guarantee a certain mmtmum level
`of reliability on equipment. Generally it is demanded
`that overall system up time should be greater than 95 %,
`which means component reliability should be better than
`99%.
`
`Acceptable throughput levels
`
`The automation strategy should be implemented in
`such a way that the cverall throughput should necessarily
`increase.
`
`Usability
`
`Usability is a broad term denoting the ability to make
`a successfull transition to automation. Following key
`factors should be looked into for a smooth transition.
`
`Automatic setup and recalibration
`
`Supplier assistance.
`
`Implementation strategies
`
`Since a large number of components are involved in
`any automation system it is essential to evolve a strategy
`and a sequence for putting the system together. In con(cid:173)
`text of semiconductor automation, a five phase buildup
`strategy is recommended [7].
`
`Phase 1: Establishment of the CAM system base
`
`CAM system base should be established first since it
`forces the user to define basic system architecture for the
`total automated line. The computer system is obviously
`a natural place to start because it can be put into place
`even before the automated process machines. Once this
`back bone is set; an easy bottoms up growth is possible by
`using Local Area Host (LAH) distributed throughout
`the floor.
`
`Phase 2: Mechanization of process machine systems
`
`With computer system ready to serve as the binding
`force individual machines can be automated one by one.
`This is implemented through cassette feed systems, stan(cid:173)
`dard interfaces, automatic process sequence control, auto(cid:173)
`matic recipe set up and monitoring.
`
`Phase 3: Adding CAM application at LAH
`
`At this juncture central computer becomes involved,
`linking local process stations (eg, Implanters, Steppers)
`for the purpose of process data monitoring, equipment
`status monitoring, failure prediction, diagnostics etc.
`
`Phase 4: Sectorize into critical area groups
`
`In this phase fully automated discrete tools can be
`organized into integrated sectors such as lithography,
`diffusion, etching etc. Within each sector the previous
`three phases are fully implemented.
`
`Phase 5: Robotics based logistic system
`
`At this point user can take a giant leap forward and
`add clean room robots and mechanized cassette carrier
`systems which carry products between the sectors.
`
`Overview of some leading CAM systems
`
`There are some commercially available CAM sub(cid:173)
`systems which cater to automation in various areas of
`Semiconductor Industry. Notable among them are des(cid:173)
`cribed briefly [6].
`
`(i) BTU Engineering Corporation, a traditional
`
`Petitioner STMICROELECTRONICS, INC.,
`Ex. 1019, IPR2022-00681, Pg. 4
`
`

`

`DIW AN & KOTHARI : AUTOMATION AND ROBOTICS
`
`371
`
`supplier of diffusion furnaces, has become involved
`in CAM market through F ACS, a furnace analysis
`and control system. This was later enhanced to
`WICS,
`a wafer
`inventory control
`system.
`Through this management information reports
`could be generated providing information on
`wafer lots in process.
`
`(ii) BBN Software Corporation has developed a
`software package titled RS-1 which is a data
`management and analysis package that provides
`data entry and retrieval, two and three dimen(cid:173)
`sional graphics, curve fitting, statistical analysis
`and analytical modelling.
`
`(iii) Consilium Associates Inc have developed a CAM
`system called COMETS (Comprehensive Online
`Manufacturing and Engineering and Tracing
`System).
`It is composed of eleven modules
`dedicated to various specific functions such as
`WIP tracking, capacity planning and scheduling,
`engineering and managerial data analysis.
`
`(iv) Fairchild have developed an Archival Computer
`Aided Yield Tracking and Evaluation System
`(ACYTE). This is a real time process control
`system providing feedback and monitoring for
`all the key elements in the wafer fabrication,
`assembly and testing cycles.
`
`(v) I P Sharp and Associates have developed pro(cid:173)
`bably the most advanced LAH system called
`PROMIS (PROcess Management
`Information
`System). This has several facilities to create and
`display user defined descriptive information or
`diagrams about any object in the wafer fa b.
`
`ROBOTICS
`
`The new crop of intelligent clean room robots owes
`its existence to the developments in the semiconductor
`industry. Faster, more powerful microprocessors and
`cheaper memory have made the robots more versatile and
`economically much more feasible to IC manufacturers.
`In this section we propose to look at what robots do in
`clean rooms today, and how they are designed to accomp(cid:173)
`lish these tasks [10-12].
`
`Historical perspective
`
`Compared to application of robots in other traditional
`industries such as steel making and automobile manu(cid:173)
`facture the development of robotic application in semi(cid:173)
`conductor manufacturing has been rather slow. Several
`reasons could be assigned for this slow pace; notable among
`them are as follows [10].
`Until recently, robots were no cleaner than human
`operators, thus they were not acceptable for clean
`room applications.
`
`The established robot manufacturers did not pursue
`the clean robot with same zeal as they spread the
`use of their machines in more traditional appli(cid:173)
`cations.
`
`Wafer fabrication process is not as operator inten(cid:173)
`sive as some other areas in semiconductor manu(cid:173)
`facturing. These areas, particularly device assembly
`were subcontracted by the western manufacturers
`to cheap labor oriented Asian countries.
`
`Moreover, the faltering economy had put the semi(cid:173)
`conductor industry in a capital investment crunch
`and revolutionary equipment like robots were the
`major casuality.
`
`Developmental stages of clean room robots
`
`The evolution of clean room robots into a workable
`tool in semiconductor manufacturing process is a good
`news for more than just the robot vendors. Robotics
`also represents a potentially brave new technology for the
`producers of semiconductor production equipment to
`In the clean rooms it would mean that human
`exploit.
`operators would have less repetitive, higher skilled jobs.
`Also it would present a new challenge to the engineers and
`technicians who will have to adapt to the ways of
`"ROBOT" [10].
`
`Figure 2 shows the developmental stages of the robot
`for semiconductor industry. As the robots become more
`advanced their capabilities and sphere of application in(cid:173)
`creases. Fundamental differences exist between familiar
`industrial robots and their clean room counterparts.
`Significant design changes in the robot, such as sealed
`joints are required to eliminate particulate contamination.
`The robot's outer cover must be constructed with a
`material that resists flaking and shedding of particulates.
`
`Robots in their simplest form are machines which per(cid:173)
`form the basic function of repeatability. This is the
`ability to return again and again, to previously taught
`point or points, as such application of these robots was
`limited to pick-and-place tasks. Earliest use of these
`robots was to put in and off-load boats of wafers from
`diffusion and CVD furnaces.
`
`A step further in evolution was imparting accuracy
`in addition to repeatability. This is the ability to locate
`a point on its own by being given a mathematical descrip(cid:173)
`tion of that point in its software programming. Such
`robots can handle tasks of transferring wafers from boats
`to cassettes or putting cassettes into various machines.
`
`More sopl:rlsticated robots have intelligence added to
`repeatability and accuracy. Such robots offer vision,
`tactile sense and decision making abilities. Vision, imple(cid:173)
`mented through pattern recognition systems, enables the
`
`Petitioner STMICROELECTRONICS, INC.,
`Ex. 1019, IPR2022-00681, Pg. 5
`
`

`

`IETE TECHNICAL REVIEW, Vol. 7, Nos. 5 & 6, 1990
`
`II
`
`Ill
`
`IV
`
`INTERACTIVE
`INTELLIGENT
`ASSEMBLY
`
`INTELLIGENT
`ASSEMBLY
`
`ASSEMBLY
`
`PICK AND
`PLACE
`
`TYPE
`
`OF
`
`ROBOT
`
`COMMUNICAT I~ S
`
`INTEC.RAT ED
`INTELLIC.ENCE
`
`INTEC.RATED
`INTEL J(£NCE
`
`ACCURACY
`
`ACCURACY
`
`ACCURACY
`
`CAPABILITY
`
`REPEATABILITY
`
`REPEATABILITY
`
`REPEATABILITY
`
`REPEATABILITY
`
`APPLICATIO"'S PUT TIN(, AND OFF PUTT lNG WAFERS. PUTTING WAFERS, PUTTING WAFERS,
`lOADING WAFER
`CASSETTES IN TO
`CASSE TrES IN10 CASSETTES INTO
`FROM DIFFUSION
`VARIOUS MACHINE MACHINE ONLY IF MACHIN[ ONLY IF
`·FURNACES.
`AND EQIJPMENTS SPACE IS THERE RIC.HT PROCESS
`CONDITION EXIST
`
`Fig 2 Development stages of clean room robots
`
`robot to "see" whether an object is present or not, and
`perform a pre-determined function. Tactile sense gives
`a robot "force feedback". For instance if a robot tries
`to insert an object which is bent tactile sense will tell it
`that the insertion cannot be made and the robot will not
`force the placement. Decision making enables the robot
`to determine its next action based on information recieved
`from its own senses, another machines or an operator.
`Such capabilities make robots suitable for operation
`such as die mounting and automatic wire bonding.
`
`Robots with still advanced capability add communi(cid:173)
`cation ability to other features named. Such robots can
`interact with the other automatic equipments in their
`environment.
`
`However, for semiconductor industry application, the
`vital requirement is protection from particle generation.
`Sophistication of robots is of secondary importance. We
`now look into the anatomy of clean room robot in this
`light [12].
`
`Anatomy of a clean room robot
`
`An intelligent robot, like any other computer consists
`of hardware, software, and communication capabilities.
`However, robots differ significantly from plain computer
`in that their dominant characteristic is a mechanical peri(cid:173)
`pheral the robot arm with its associated end-effector.
`First, computer oriented aspect of the robot design would
`
`be taken up. This would be followed by mechanical
`aspects of the robot and stringent requirement that clean
`room application places on it [11).
`
`Hardware requirements
`
`Robots require tremendous amount of computing
`and number crunching power. A robot must perform
`kinematic transformation when asked to move from one
`point to another in a cartesian coordinate system. These
`transformations are required to determine what rotary
`movements of it's motor shafts would lead to desired
`movement. Computations are further compounded by
`the fact that all motors move by unequal amount, at
`varying acceleration and deacceleration and yet produce
`a synchronized motion. Also, the robot is recieving real
`time inputs from material feeders, processing equipment,
`safety equipment and high level control computer which
`it analyses and makes decisions accordingly. All these
`activities impose terrific demand on computing power.
`Fortunately with advances in semiconductor technology,
`in form of the
`such computing power is now available
`inexpensive yet powerful microprocessors
`that make
`robotics viable for clean room automation. Key micro(cid:173)
`chips used for robot brain would now be described:
`
`An Intel 8086/8087 a 16-bit microprocessor and
`powerful math coprocessor pair is most commcnly
`used. This pair handles number crunching re-
`
`Petitioner STMICROELECTRONICS, INC.,
`Ex. 1019, IPR2022-00681, Pg. 6
`
`

`

`DIW AN & KOTHARI : AuroMAnoN AND RoBOncs
`
`J73
`
`quired to do the kinematics transformation fast
`enough to meet required duty cycle time.
`A pair of z-80, 8-bit microprocessors is often used
`to control the motor driving each axis. One z-80
`receives signals from the CPU regarding desired
`motor movement and signals from feedback loop
`indicating actual motor movement. It compares
`these, and issues commands for corrective motor
`pulses, if required. The second z-80 is a slave to
`the first and issues the actual motor pulses.
`
`An Inlet 8088 is the CPU normally used for auxi(cid:173)
`liary computer which controls end-effectors or end
`of arm tooling.
`
`All these microp ocessors are arranged in a hierarchical
`control system. The memory requirements range from
`128K over 750K. Thus it is apparent that robotic sys(cid:173)
`tems could not exist without the advances in micro-elec(cid:173)
`tronics, which they are now helping to implement.
`
`Software requirements
`
`Robots are, by definition, "soft-automation" or re(cid:173)
`programmable, and the programming language of a robot
`largely determines its usefulness. For typical clean room
`applications, a robot needs the capability to be program(cid:173)
`med in a high-level language.
`
`For this purpose a programming language called Robot(cid:173)
`BASIC (or R-BASIC) bas been developed._ This is a
`version of popular ·Micro-soft BASIC, with commands
`added for control of the arm, end-effector, vision system,
`other sensors and ancillary equipment. With Robot
`BASIC, an operator can write one program, offiine, which
`integrates control of all the basic elements of a robotic
`workcell. BASIC bas the additional advantage of being
`an interpretive language. This allows the operator to
`single step the robot through a program and observe
`robot operation.
`
`Programs are down loaded into robot memory either
`through a plug-in programming computer or a "teach(cid:173)
`pendant".
`
`Communication requirement
`
`Automation in a clean room requires extensive com(cid:173)
`munications between the various pieces of equipment, if
`the degree of automation is to reach the point where it is
`cost effective.
`
`Minimum requirements for any robot working with
`semiconductor equipment should be able to communicate
`over a SECS II interface. Compatibility with RS-232
`(with SECS I modification) is also very useful.
`
`Mechanical considerations
`
`Significant design changes in the robot, such as sealed
`
`joints, are required to prevent particulate generation.
`Additionally, the robots outer cover must be constructed
`with material that resists flaking and shedding.
`
`For arms having rotary joints, the reduction in parti(cid:173)
`culate generation could be achieved by the use of ferro(cid:173)
`fluid seals at the joints, with the remainder of the arm
`designed so that it is nearly or entirely enclosed.
`
`Preferably there should be no parts with exposed lubri(cid:173)
`cants, and if lubricants are exposed, they should have a
`very low vapour pressure.
`
`All external surfaces should be of non-shedding variety,
`such as bare metal. Plastic could be used but in vacuum
`operation requirement, phenomenon of outgassing could
`create problems. If paints are to be used they should be
`of a durable, non-flaking variety.
`
`Motors used should be brushless to minimize parti(cid:173)
`culate generation. Figure 3 summarises the building
`blocks of a clean room robot.
`
`A. ~LEXISLE AND EXPANDABLE ARCHITECTURE AFPROACH
`
`VISION
`o INTEl 8086
`l2K ROM !EXPANDABlE TOI281<)
`0
`. o 1281< RAM- VISION DATA
`0 641< RAM- DISPlAY
`o SERIAl/ PARAllEl 1/0 PORTS
`
`PlUG· IN
`o 121< 1128 KEPROM
`o 161< EPROM.ll6KRAM
`o FUTURE
`ENHANCEMENTS
`
`H OST
`o TER
`MINAL
`o MICR
`OCOMPUTER
`o MAIN
`FRAME
`COMP
`UTER
`
`l_____
`
`CONTROlLER
`INTEl 808618087
`0
`o 128K RAM. 128K ROM
`o 8 K EEPROM. 41< CMOS RAM
`o SERIAL/ PARALLEL I I 0
`
`c PORTS
`
`AUXIliARY COMPUTERS
`o INTEL 8088
`o 81< EPROM 2K RAM
`o CONTROL LINES
`o SAFETY 110 PORTS
`
`'
`. --_J
`
`---
`
`I TE
`
`MOTOR CONTROllER I
`
`~ROBOT ARM I
`I ENO EFFECTORS I
`Fig 3 Anatomy of a clean room robot
`
`Petitioner STMICROELECTRONICS, INC.,
`Ex. 1019, IPR2022-00681, Pg. 7
`
`

`

`374
`
`IETE TECHNICAL REVIEW, Vol. 7, Nos. 5 & 6, 1990
`
`buted processing communications and control network,
`a software system which controls system components and
`flexible guide paths for automated vehicles to follow.
`
`Thus a scenario, where a number of guided vehicles
`with robotic arms perched on the top, whiz around the
`aisle, transferrng cassettes of wafers from one equipment
`to another is not way out in the future. Facilities with
`such systems are coming into vogue and would proliferate
`in coming years.
`
`CASE STUDIES ON AUTOMATION AND ROBOTICS
`
`Wafer fab case study
`
`The company. NMB Semiconductor Company Ltd,
`Japan was formed in May 1984. A 43,000 square feet
`wafer fab along with 21,000 square feet energy center were
`built by April 1985 and in August 1985 first working
`samples rolled out. In April 1986 NMBS offered first
`commercial grade CMOS 256 K DRAMs which have
`high speed access level [14, 15].
`
`Total automation. The facility at NMBS consists of
`an energy center, production plant and employee cottages/
`dormitories. Energy center provides services like nitro(cid:173)
`gen, oxygen, D I water etc. Production plant houses
`wafer fab, assembly and test units.
`
`The wafer fab is totally automated. Cassette move
`in an ultraclean main transfer line suspended immediately
`below the ceiling. Complementing this transfer line are
`branch transfer lines robot hands run through dual struc(cid:173)
`ture ultraclean rooms as shown in Fig 4. These rooms
`have class 1 'operator free' processing section and class 30
`maintenance section. Operators work out side the ultra(cid:173)
`clean room in a class 300 corridor within which they
`monitor and adjust the processing equipment via remote(cid:173)
`control systems and closed circuit TVs.
`
`To minimize the expensive clean room floor space the
`wafer fab has been located in a two storied structure with
`all gases, chemicals and other services coming from the
`basement. Figure 5 shows this implementation.
`
`Towards foture-.Inter-equipment transfer
`
`There are basically three approaches to inter equipment
`transfer [13].
`
`(i) The two pieces of equipment can be directly linked
`together, both mechanically and electrically.
`
`(ii) Cassettes can be transported over a fixed modular
`track.
`
`(iii) In the third approach cassettes can be transported
`by a guided "robot-on-wheels" type of vehicle.
`
`The direct link approach is well suited for some small
`groups of processing steps such as scrub, coat, bake and
`align. In fact a lot of such systems have been developed
`and are doing a good job in the industry.
`
`The second and third approaches to inter-equipment
`transport are both based on cassette transport technique.
`Presently, cassettes are carried manually between the
`equipment therefore greatly susceptible to contamination
`problems.
`
`The only fixed track system that is currently com(cid:173)
`mercially marketed has been developed by Nacom Indus(cid:173)
`tries, Inc, Tustin, California. Their Namtrak system
`is currently at use at AT&T's Kansas City facility. This
`system consists of modular sealed clean environment
`section through which cassette boxes are moved. By
`attaching straight sections and turn tables together, a
`bi-directional or continuous loop system can be installed
`in a facility at work station level or elevated above doorway
`height. System complexity varies from simple relay
`logic controlled 'call/send", single rail car units to fully
`automated systems with robotic auto-load/unload stations
`and multiple railcars. One major disadvantage of the
`fixed track approach is the problem of loading and un
`loading of cassettes to and from the track system. This
`can, however, be solved by using a robotic arm at each
`gateway.
`
`The third approach is being pioneered and promoted
`by Veeco Instruments Automation division of Dallas,
`Texas and Flexible Manufacturing Systems (FMS) of
`Los Gatos, California. The FMS system consists of four
`major components; a mobile transport unit (MTU), a
`docking module, a central control computer, and an intelli(cid:173)
`gent work-in-progress (WIP) station. The MTU is a
`vehicle that transports wafer cassettes or boxes from one
`work station to another by a robotic arm. The docking
`module is mounted on the production tools or WIP sta(cid:173)
`tions to provide docking point for MTU. The central
`control computer manages all MTU guidance and loca(cid:173)
`tion monitoring facility. The WIP station provides
`local storage and inventory control for 16 wafer cassettes
`or cassette boxes. The Veeco system consists of auto(cid:173)
`mated guided vehicles with robotic capabilities, a distri-
`
`B•anch uansler 11r.cs
`
`I
`
`\
`
`s ! s m~W00 s -
`v]l,l.l 1 Q
`
`Automated VLSI Fabricat•on Area
`Ma1n ovetheao
`tr.1nspo11 system
`
`Si"lf.ll)l' buller
`~l.lhon!.
`
`4
`
`I
`
`f
`~
`2
`n.'
`
`I Ys:1
`
`Class 30
`ma•ntcnancc
`area
`
`ultraclean
`processmg area
`
`\ /
`
`C1ass 300
`opcrJ:ot a·ca
`
`c
`
`~
`c
`~
`
`c
`g
`'---
`
`JL
`
`g'
`~
`s
`~
`
`~
`
`\L:':alls
`
`Fig 4 Wafer fab layout at NMBS facility. Ultra clean process
`areas are joined by branch transfer lines to the main overhead
`transport system
`
`Petitioner STMICROELECTRONICS, INC.,
`Ex. 1019, IPR2022-00681, Pg. 8
`
`

`

`DIW AN & KOTHARI : AUTOMATION AND ROBOTICS
`
`375
`
`Total control. The complete automation system including
`various software packages were designed and developed
`in house. The system is controlled by a host computer,
`and has capability of a lot and individual wafer tracking.
`CCTV's are located in the control room to monitor indi(cid:173)
`vidual process areas in the fab. All operations are carried
`out by operators controlling from the control room.
`
`Automation pays back. NMBS plant produces 20,000
`to 24,000 wafers per month or 4 million devices a month.
`The defect density levels of 1 to 2 defect/em/wafer for
`particles of 0.17 ·fLm shows the level of contamination
`control. This has enabled high yields (theoritical 56-74%
`yield