Envisioned was a completely automated wafer-fabrication line that could potentially manufacture built-in circuits at a pace significantly faster than previously imaginable. In a bygone era, a mere 54 years ago, pioneering efforts were undertaken to develop complex integrated circuits, which would have been considered bold even today, when cutting-edge manufacturing facilities can churn out sophisticated ICs within weeks rather than days. Manufactured on a batch-by-batch basis, integrated circuits, commonly referred to as random-access memory chips, were typically assembled at a leisurely pace over the course of several weeks through a laborious process involving multiple manual workstations.
On the occasion when Harding held the position of supervisor over IBM’s Manufacturing Analysis group,
. The venture that would bring his vision to reality, though its specifics remained unclear, was dubbed Project SWIFT. To achieve a remarkably swift turnaround time, a fundamental transformation in design was essential, necessitating the implementation of automation through a paradigm shift in strain development. With Harding’s team overcoming significant challenges, they successfully reached breakthroughs that have the potential to be replicated globally across the semiconductor industry. While many of SWIFT’s pioneering advancements have become standard in today’s highly automated chip manufacturing facilities, its exceptional speed remains unparalleled in terms of rapid turnaround times.
While SWIFT fabricates layers at a pace averaging just 5 hours, its competitors require 19 hours per layer, with industry averages reaching as high as 36 hours. Despite advancements in built-in circuits featuring numerous extraneous layers on massive wafers resembling large pizzas, the processing complexity remains a significant hurdle. Swiftly, Harding’s automated manufacturing line proved to be remarkably efficient.
A Semiconductor Manufacturing Manifesto
I first met Harding in 1962, and prayed it would be my last encounter.
Was preparing to release its first fully solid-state PC, the. The meeting was a notably tumultuous affair. “What practical use do you propose for something so minuscule?” he thundered, his incredulity evident as I showed him the intricacies of handling those microscopic, unboxed semiconductor cubes on a large scale for quality control and categorization purposes?
Jesse Aronstein, the creator and manager of the gear group at Venture SWIFT, traded his technical responsibilities for a creative outlet when he played French horn with the Southern Dutchess Pops Orchestra one night a week. One additional key supervisor was Walter J. Wally Kleinfelder, standing at parade rest, led the methodology group within Venture SWIFT as its commanding figure. William E. “Invoice” Harding, a no-nonsense World War II combat veteran and ingenious inventor, is captured here in 1973. He led the development of IBM’s SWIFT venture, successfully pioneering the creation of built-in circuits in record time.What’s the significance of this pair?
Was a pioneering visionary and ingenious inventor. As he approached his fourth year of growth and development within IBM’s manufacturing ranks, a pivotal change was underway: the corporation’s new direction took shape in 1961. As Harding rose through the ranks, he became a mid-level supervisor in the new division, responsible for driving growth and production of the components necessary for crafting the pioneering System/360’s cutting-edge solid-state devices and complex circuit modules.
He possessed a rugged exterior, uncharacteristic of a typical IBM supervisor. Growing up in Brooklyn, New York, it’s little wonder that someone with a background like his would have anticipated the outcome; after all, he’d been shaped by the harsh realities of war, having suffered three wounds while fighting in World War II under General George S. Patton’s Third Military. Following the battle, Harding pursued higher education, earning both bachelor’s and master’s degrees in mathematics and physics before joining the Institute of Electrical and Electronics Engineers (IEEE) as a member.
I began my career with IBM in 1961, having transitioned from a role focused on enhancing rocket engines at Convac Corporation. As a typical engineer of that era, I was unfamiliar with the intricacies of semiconductor fabrication. Five years earlier, I had enrolled in a vacuum-tube electronics program where my instructor dismissed these innovations as “laboratory curiosities” – entities that could potentially amount to something, but whose value was uncertain.
At the IBM’s sprawling East Fishkill semiconductor facility, Venture Swift occupies a modest abode, aptly illustrated by a bright yellow marking on building 310. IBM
The rough-edged persona of Harding consistently emerged whenever our paths intersected. The possibility of him venturing into IBM’s premises would have been a significant milestone in his life.
Administration coaching lacked concrete evidence to substantiate its effectiveness. Despite the challenges, he successfully accomplished his objective. As production of solid-state logic modules for IBM’s groundbreaking System/360 computers reached full throttle by 1964, the Elements Division’s newly established East Fishkill facility was humming along, transforming a former farm into a hub of technological innovation.
After a three-year sabbatical pursuing graduate research, I rejoined IBM in July 1970. Prior to taking an academic break, I held the position of first-level supervisor for four years, after which I did not require another administrative role. I pursued a purely technical profession, and upon joining East Fishkill’s Manufacturing Analysis (MR) group, I hoped to secure one.
The unexpected reunion with Harding had me pondering our past encounters. On August 15, 1970, he ascended to the role of highest authority at MR. Prior to that, he invested 12 months in developing an IBM-approved methodology for long-term production and utilization.
(VLSI) circuits. Commanded to lead the MR team, he endeavored to validate the feasibility of his innovative manufacturing concepts.
A meeting of Management Representatives personnel was convened to announce an administration change. After launching, Harding outlined his vision for future VLSI functions and manufacturing capabilities. His most crucial factors had always been:
- VLSI circuits initially leveraged advancements in field-effect transistor technology, a significant departure from the dominance of bipolar-junction transistors at the time.
- High-quality outputs are essential.
- Manufacturing could be totally automated;
- Optimal results are likely to arise when processing individual wafers separately.
- Fleeting instances of turnaround will yield substantial benefits.
- By scaling up, we could replicate and expand our most successful manufacturing processes to increase overall quantity.
As the academic lecture concluded, Harding underwent a transformation, morphing from professor to commander in a manner reminiscent of General George S. Patton. The MR’s primary goal was to elucidate Harding’s ideas and ongoing projects, and any undertakings not directly supporting this objective were free to be reassigned within IBM or abandoned. A highly advanced automated system could potentially process around 100 wafers daily, yielding exceptional results while achieving a remarkable one-day turnaround time, independently operating without human intervention.
Did I hear that proper? What we once considered a monumental challenge – a one-day turnaround from blank wafers to fully functional circuits – is now a testament to the incredible pace of innovation. It commonly required more than a month to complete this task at that time. Did he actually imply it?
Harding grasped the theoretical feasibility of his goal, and with unwavering resolve, he set out to make it a reality. IBM would reap significant benefits from rapidly producing prototype experimental IC designs – potentially within a single day – rather than the usual timeframe of months. He demanded that the circuit designer deliver testable circuits just 24 hours after submitting the digital design specifications to production.
One-day turnaround from blank wafers to fully functional circuits – an endeavour that would today be dubbed a moon shot in terms of its audacity and scope.
Upon Harding’s prompt initiative, he swiftly established a gear team and a training program within his organization, appointing me to lead the former. I was spared the added responsibility of overseeing others. As a former second-level supervisor, I was tasked with spearheading the development of processing and wafer-handling equipment for an as-yet undefined manufacturing line, a challenge that was only just beginning to take shape in my mind. My tenure in my dream analysis job had been remarkably brief, lasting barely more than a month.
Transferred to manage responsibility for handling the methodology group effectively. To design the semiconductor product, they would select the components to manufacture and map out the manufacturing process, meticulously detailing each step – from processing a pristine silicon wafer to creating intricate integrated circuits on its surface with high yield performance.
Kleinfelder selected an IBM RAM II, a random-access memory chip, for our demonstration purposes. The East Fishkill facility had been producing this product locally, allowing us to readily access all necessary components and assess our performance against that of the existing non-automated production line.
IBM’s SWIFT pilot wafer fab had an innovative transportation system: a monorail “taxi” that efficiently moved workers and supplies.
The process of built-in-circuit manufacturing involves fabricating transistors and other components onto a silicon wafer, followed by wire bonding them together through the selective deposition and etching of thin layers of aluminum to form the desired circuit pattern. The thin film, typically made of a conductive material like copper, that acts as a conductor in a microelectromechanical systems (MEMS) device is commonly referred to as the wiring or metallization layer.
IC manufacturing makes use of
To fabricate an integrated circuit (IC), one aimed to design and manufacture numerous layers, each featuring a unique sample. The interconnects cradle multiple metallic wiring layers, exceeding a dozen in complexity for a modern integrated circuit. The metallic layer on the wafer is then coated with light-sensitive material, following which an image of the sample is projected onto its surface. The regions where conductors should be formed are shaded from sunlight. Upon development of the image, the resist material separates from the exposed regions, allowing those areas to be selectively treated with acid and shaped accordingly. The remaining portion of the floor remains shielded from corrosive substances due to its acid-resistant coating. After etching is completed, the residual protective resist is removed, resulting in a pure wiring layer within the specified sample.
The IC course employs lithography to fabricate transistors and other components onto a silicon wafer. Tiny amounts of specific impurities are infused into exposed areas of pure silicon through openings etched in insulation layers, thereby modifying the material’s electrical properties. The fabrication of RAM-II ICs necessitated a laborious process involving four distinct photolithography steps, each employing unique mask patterns: three dedicated to transistor and element formation, and one focused on metallization. To effectively produce the chips, the four patterns must be meticulously synchronized with one another.
While being a crucial component in IC manufacturing courses, It typically takes several weeks for a RAM-II wafer to complete its processing within our existing production line. In just under 48 hours, the wafer underwent an uncooked transformation, navigating multiple thermal, lithographic, chemical, and deposition stations with precision and speed. The majority of a wafer’s existence was devoted to preparing for the forthcoming sequence of steps. If wafers advanced quickly through each process, some chemical cleaning steps might potentially be eliminated.
Their task was to identify potential areas for streamlining and opportunities for expediting processes within Kleinfelder’s organization. The ensuing period was less than 15 hours long. It then fell to
I would appreciate it if my supervisor, responsible for chemical-equipment improvement, could review the proposed plan and provide feedback. Researchers painstakingly crafted individual wafers with a diameter of approximately 1.25 inches, utilizing a makeshift “pots and pans” laboratory arrangement to meticulously evaluate and refine their product. The abbreviated process efficiently produced working circuits within approximately 15 hours, as expected.
The emergence of a self-sustaining automated system became manifest. Initially conceived as a series of interconnected devices, each responsible for executing a single stage of the process, akin to a manufacturing assembly line where individual components come together to form a cohesive whole. Despite requiring adjustments for planned maintenance and repair of equipment failures to be factored in. This achievement was enabled by the strategic implementation of short-term storage “buffers,” which temporarily stored wafers at select points along the production chain as needed.
The course of chain ideas was significantly disrupted by concerns related to. Publicity surrounding the photoresist on wafers was typically accomplished through an analogous process akin to traditional photographic contact printing methods. The lithographic masks, through which sunlight gently shone when exposing the silicon wafer, suffered equally from photodegradation, akin to a photographic negative’s vulnerability to damage. Any imperfection or contamination on the mask will result in a consistent defect pattern on the integrated circuit, precisely at the same site, regardless of the wafer being processed.
The East Fishkill Lithography Group had successfully pioneered a revolutionary non-contact technology offering a substantial 10-to-1 price reduction.
. Here is the revised text in a different style:
A miniature photographic display unit emulated a slide projector, casting a reduced image onto a microchip, showcasing a single layer’s contents. As it “stepped” along the wafer’s surface, revealing one microchip location after another in sequence. In relation to contact masking, the anticipated reduction in sensitivity to particulate contamination stems from the diminution of the shadow’s dimensions, enabling a potential decrease of 10:1 for stray particles. Diverse advantages featured enhanced visual acuity and extended mask lifetimes.
– YouTube
Due to its gradual nature, multiple step-ups were required to meet the desired throughput objective. To achieve optimal sample alignment on each wafer, it is necessary that the wafer be re-routed to the same stepper for exposure of every layer throughout the fabrication process. To ensure that picture distortions caused by slight variations between machines are cancelled out, To fabricate the RAM-II circuits, a silicon wafer had to undertake four distinct trips to its designated stepper module. The linear sequence was divided into five distinct sectors. A monorail ‘taxi’ system transported wafers from one processing sector to their designated steppers, then returned to collect the wafers for subsequent sectors.
Each of the five sectors was designed as a self-contained module, housing all necessary automated wafer-processing equipment and machinery required to execute its corresponding step in the manufacturing process. The sector enclosures and taxis could be engineered to provide a controlled environment, mimicking a clean room quality native atmosphere, ideal for wafer storage. Within a sector enclosure, wafers typically transition seamlessly from a wet chemistry module to compact furnaces, followed by a stop at the photoresist utility module, before finally arriving at the taxi pickup station. Within the wet-chemistry module, the wafer undergoes a series of processes, including cleaning, improving and removing photoresist, as well as etching, among other procedures.
Effective management of the entire operation aimed to be accomplished across three distinct tiers. Centralized management of production-line operations, including record-keeping, taxi logistics, and monitoring, can effectively be handled through a computer-based system. Skilled sector controllers, assigned to each area, would manage wafer supply chains from start to finish, sharing data with visiting wafers and processing information seamlessly with the central hub. Within each sector enclosure, the person-processing and wafer-handling modules would possess dedicated control systems, designed to ensure impartial setup and maintenance.
Once fully configured, our automated demonstration line for the RAM-II chips will comprise five distinct sectors: a taxi area, lithographic-pattern imaging module, and three others, all controlled by a central computer. After assuming leadership for six months, the company, MR, initiated the development of a customized system requiring meticulous planning and execution.
In a surprising turn of events, the brash center supervisor discovered an unlikely source of inspiration: literature.
Harding made frequent journeys to
In Armonk, New York, he traveled to present updates, solicit resources, counter criticisms, and convince top officials that the investment would ultimately yield exceptional returns. It was a troublesome mission. The tension emanating from his weekly team meetings was often a direct reflection of the anxiety he struggled with. He delivered erudite lectures on subjects he assumed we were already familiar with, shared parabolic stories that required interpretation, and wove intricate comparisons.
At the time, it wasn’t clear to me that he was leveraging employees’ gatherings to iterate on ideas for the exhibits at Armonk. As he observed our responses, he tailored his presentation to resonate with us more effectively. He presented his findings to the top echelons with remarkable proficiency. Throughout its approximately three-year venture, MR enjoyed unfettered access to all necessary funding and resources, leveraging these assets to comprehensively develop, design, construct, and operationalize the entire system.
During a worker’s assembly, President Harding had someone read aloud Heywood Broun’s engaging short story “
The elusive allure of an untouchable ideal, perpetuated by a tantalizing mantra that dares us to strive for the seemingly impossible? At his level, we were seeking an exceptionally outstanding title for the venture. “SWIFT” was finally chosen. While Harding repeatedly asserted that it wasn’t an acronym, people still speculated it stood for “Semiconductor Wafer Built-in Manufacturing Facility Expertise”.
SWIFT’s remarkably speedy processing times have yet to be surpassed.
IBM’s Elements Division tailored Swift’s processing and wafer-handling equipment entirely. The initial design objectives aimed to mechanize the processing of wafers consistently, maintaining their integrity and clarity throughout the process. Wafer-handling experiments have successfully optimized the most refined and delicate procedures to ensure the highest level of cleanliness and gentleness. Dealing with gear was intended to assist in holding the wafer securely rather than providing a firm grip on it. A novel wafer handler was developed, leveraging a controlled stream of air to suspend and transport wafers without physical contact, seamlessly integrating this innovation into various wafer-handling operations.
The singular exception to SWIFT’s consistently ‘clear and mild’ approach to equipment management lay in its handling of this particular case. The administration at the Elements Division’s Burlington, Vermont-based website exerted pressure on Harding to employ “air-track” wafer-transport equipment, which they had designed and developed. This mechanism utilized airflow to lift and transport wafers, much like the motion of a puck on an air-hockey table. Harding sought to maintain a strong partnership with Burlington, and as such, he decided to incorporate air-track technology into the existing SWIFT system. Despite addressing wafer contamination and reliability issues, these concerns remained unresolved.
A patchwork approach to sector management arose from SWIFT’s top-down decrees, resulting in a mismatched pairing of techniques that undermines the principles of maintainable design. As the company’s latest innovation became available, our team was tasked with integrating it into the customized controllers, which were already in production – five distinct models, each tailored to a specific industry.
Developed specifically for industrial applications such as factory equipment and process control. Although IBM did not integrate PCs into their internal operations, it remained an open question: why not? However, when SWIFT successfully employed a System/7 in its venture, this endorsement could have contributed to promoting the System/7’s adoption. The company ultimately designed four bespoke controllers and a single System 7 controller to accommodate the five distinct sectors. Each sorts labored effectively.
Gear reliability proved to be a significant vulnerability for SWIFT. To ensure maximum reliability and minimize maintenance requirements, standardised mechanisms and controls were implemented throughout the system, prioritising functionality over aesthetics or innovative design. Upon observation of the system in action, one would notice that numerous movements were accomplished through distinct, consecutive steps rather than a solitary traversal. Beneath this distinctive characteristic lay a reliance on the widespread application of robust and trustworthy
Originally designed centuries ago for clockmaking, these precision mechanisms have since been adapted to cater to the demands of modern industry, where accuracy and reliability are paramount, particularly in applications requiring precise locking and smooth motion, whether linear or rotary. Each deliberate rotation of the Geneva drive’s central shaft resulted in a precise incremental movement. The lengthy traverses necessitated multiple rotations of the shaft, resulting in peculiar, seemingly aimless motions.
Within an enclosed chamber situated within a sector, a silicon wafer underwent a series of fully automated processing procedures in a controlled environment. Here are two of the initial concept designs, as depicted below. Wafers entering the higher chamber had a sample exposed on the resist, subsequently undergoing a series of processing steps: improvements, hardening, etching, and others, as detailed.
Spin-coating the wafers with centrifugal force to evenly distribute liquid photoresist across their surfaces, typically applying it to the centre of each wafer. Currently, unsuitable spin velocities are consistently identified as the primary cause of resist-related wafer processing defects and rejections. To eliminate spin velocity as a factor in the experiment, SWIFT’s spinners were powered by synchronised AC motors operating at a precise 3,600 revolutions per minute, thanks to their connection to a stable 60-hertz AC power source – a common setup similar to that used in phonograph turntables. No velocity controllers are typically needed. The targeted photoresist film thickness can be attained through modifications to the process parameters: temperature, viscosity, or spin duration. In the long term, the elimination of four redundant velocity controllers significantly enhanced system dependability.
As Swift transitioned from concept to tangible hardware execution, Harding reorganized MR’s team and secured buy-in from key stakeholders. He recognized that his team members had the necessary resources to tackle the project and would focus on the venture. As I arrived, I was impressed by his exceptional organizational prowess and his ability to identify and attract high-caliber talent from across the organization.
Harding pioneered the development of SWIFT’s grasp management system, which tracked the progression of each wafer as it traversed the manufacturing sectors. This Execution Management System (ECS), primarily built upon
. Each wafer was assigned a unique serial number, allowing for precise tracking throughout its journey along the production line. The ECS continuously monitored and controlled every wafer’s processing parameters in real-time, promptly detecting and responding to any deviations from specifications. In retrospect, the innovation of punch playing cards and tape cartridges may seem antiquated compared to modern standards, yet their introduction represented a significant leap forward in manufacturing control and monitoring on the wafer line.
The company also relocated a comprehensive Instrumentation Division, led by Sam Campbell, from its IBM Endicott facility to East Fishkill. Following Campbell’s division, innovative tactics were crafted to manage the SWIFT course in real-time and on-site.
A Pioneering Career: Leaving a Lasting Impact on the World of Semiconductor Production.
Prototypes of furnaces and chemical processing units have been built and thoroughly scrutinized.
The team within East Fishkill’s Manufacturing Engineering department, led by a dedicated division, carefully crafted the sectors and intricately assembled the processing equipment modules within them. The Harding Act was introduced to establish procedures for setting up, debugging, and operating the road system. As the gear and amenities converged in SWIFT’s meticulously designed 4,000-square-foot facility, Wu took responsibility for overseeing the installation, start-up, and troubleshooting of the newly developed vacuum metal-deposition equipment, which he had played a significant role in designing alongside his team.
One single operation throughout the entire course of production remained non-automated. The precise alignment of the wafer to expose the sample onto the photoresist still relied heavily on the expertise of a well-trained operator. At its peak, SWIFT operated a 10:1 optical stepper alongside a 1:1 contact-mask machine; however, since then, most chips were manufactured using the 1:1 process due to its significantly higher throughput.
By early 1973, IBM’s headquarters had already concluded that the successful automation of wafer processing was a realistic possibility. To achieve this critical objective, a newly designed wafer-processing line was established with the primary goal of manufacturing circuits for IBM’s cutting-edge “FS” personal computer.
). The proposed new line, designated as the “Future Manufacturing System,” was rebranded; concurrently, the SWIFT initiative was relabelled the “FMS Feasibility Line.”
Bevan Wu demonstrated exceptional leadership skills in overseeing the project’s successful completion, showcasing his expertise in resource allocation, staff development, and process optimization through meticulous planning, quality control, and timely adjustments to equipment, methods, and protocols. He initiated the process of obtaining certification to manufacture circuits for IBM’s products. Between mid-1974 and early 1975, the system executed five consecutive operational runs. During downtime, his team meticulously examined results and implemented improvements. The longest continuous run of streaks without a missed day lasted an impressive 12 days. Wafer throughput averaged 58 wafers per day, representing 83% of its designed capacity. The typical turnaround time from receiving bare wafers to producing testable circuits is approximately 20 hours. The unadjusted duration of time was 14 hours? The yield ultimately matched the highest level of quality attained by East Fishkill’s established RAM-II production process.
Around 135 professionals, comprising technicians, engineers, and managers from various IBM regions globally, received training in operating the system. They fabricated 600 high-quality wafers featuring 17,000 RAM-II FET memory chips.
Like General George S. Patton, who was famously bypassed by higher command, Harding found himself overlooked in favor of others, specifically tasked with guiding “the big show” – namely, the development of the revolutionary FMS automated production line. With a departure from his administrative roots, he ascended to the esteemed role of IBM Fellow, a pinnacle of distinction within the company, transcending traditional management ranks.
The Swift FMS Feasibility Line concluded its steady operation in the early months of 1975. It had achieved its goals. Individuals were asked to help establish the FMS production line to manufacture FS computer systems. By late 1975, the FS initiative had been scrapped, rendering FMS redundant. The company’s production line was redirected to meet the sudden surge in demand from FMS, with a substantial allocation of gears manufactured at East Fishkill.
The AS/400, a pioneering IBM showcase that is more widely recognized and celebrated than its lesser-known antecedent, the Venture SWIFT.
Although Swift’s lifespan was brief, its innovations remain evident in modern semiconductor fabrication facilities. Like SWIFT, these state-of-the-art fabs boast cutting-edge automation and precision-controlled processes, featuring a central transport system that relies on “Bernoulli” handlers to elevate wafers with pinpoint accuracy, sans physical contact; instant resist application follows oxide or metallic film formation; lithographic sample exposure is enabled by steppers; and real-time process management ensures seamless operations. For five decades, all of these pioneering ventures had been trailblazing milestones for Venture SWIFT.
My time spent working under Harding’s guidance on the SWIFT project for three years had a profoundly transformative impact on my professional development. What initially filled me with trepidation ultimately gave rise to profound admiration. I’ve grown to appreciate Invoice Harding as an undeniable genius, in his own unique way. Fueled by the innovative spirit that defined his leadership style, a dedicated team exceeded even the most optimistic expectations. Far exceeding our own lofty expectations.
We regard the pioneers behind a successful business as the “founders” or “pioneers” of the modern incarnation of their groundbreaking innovations. Pioneers in their respective fields, such as Thomas Edison, Alexander Graham Bell, Henry Ford, and the Wright brothers, are commonly regarded. In that sense, William E. Is the pioneer behind the trendy, automated, and billion-dollar fabric industry indeed Harding?
From Your Web site Articles
Associated Articles Across the Net
