[ad_1]
The semiconductor business has a repute for extremely technical and ultraefficient manufacturing, seemingly providing a blueprint for cutting-edge operations. This notion could fall quick in a single space, nevertheless. Regardless of offering the intelligence for the digital financial system, semiconductor firms usually lag behind different industries in automation and digital transformation, particularly back-end actions, which contain reducing wafers into particular person chips, testing processes, and packaging.
The dearth of automation in back-end operations can have main penalties, since these actions usually account for 20 to 25 % of producing prices. The implications may turn into extra extreme as firms transfer to the next generation of advanced packaging, or in the event that they relocate manufacturing nearer to house, usually in higher-cost international locations, as many firms are actually contemplating.
By growing back-end automation, built-in gadget producers (IDM) and outsourced meeting and take a look at (OSAT) gamers can considerably improve throughput functionality whereas lowering or eliminating the necessity for expensive capital expenditure expansions in the long term. Automation would additionally give firms extra visibility into and management over provide chains, and price efficiencies, resulting in commensurate uplifts on the underside line.
The present state of back-end operations
At current, simply 30 % of semiconductor firms deploy synthetic intelligence or machine studying at scale (Exhibit 1). The explanations, business leaders say, embrace a scarcity of capabilities, information safety issues, connectivity and technical points, and generally unclear enterprise circumstances.
Even semiconductor firms which have made progress in AI and machine studying (ML) nonetheless have many alternatives to seize extra worth. Our analysis exhibits that AI/ML contributes between $5 billion and $8 billion to annual earnings, earlier than curiosity and taxes, throughout the business. Whereas that is a formidable sum, it displays solely about 10 % of AI/ML’s full potential. Furthermore, the actions more than likely to profit from AI-related worth creation are manufacturing and R&D—two areas the place they might be utilized the least.
Most not too long ago, business challenges on the store flooring had been highlighted throughout the world pandemic. Amid lead instances that stretched six months or longer, chip shortages precipitated slowdowns in industries from autos to agriculture. Even now, because the impression of the pandemic lessens, chip shortages stay in a number of areas, particularly on a regular basis functions resembling vehicles, house home equipment, medical gadgets, and electronics. In the meantime, manufacturing progress will likely be gradual till at the very least 2026, regardless of big funding allocations, most of which go to front-end capabilities. In an business predicted to hit $1 trillion in annual gross sales by 2030, there may be a lot room for enchancment.
Automation boosts throughput, efficiency, and productiveness
Our current expertise working with main IDMs and OSATs exhibits {that a} concerted back-end manufacturing transformation can produce important efficiency advantages over a 12- to 18-month interval. Essentially the most profitable transformations take a holistic strategy, overlaying general gear effectiveness (OEE) potential (together with digital implementation and automation), in-plant planning, provide chain coordination, complete price of efficiency, and output high quality (yield, scrap, and failures). Successfully carried out, motion in these areas can result in a 20 to 30 % rise in shop floor throughput, a 20 % decline in unit manufacturing prices, and a 30 % drop in buyer complaints—all at close to net-zero capital expenditure.
The keys to capturing back-end efficiency advantages are automation and digital innovation. In sensible phrases, this could play out throughout a number of elements of operations. Nevertheless, an essential first step can be establishing a digital management room—a “bridge on the ship” that will facilitate evaluation of bottleneck course of steps resembling grinding, wire bonding/lid connect, testing, or burn-in. The impression of making a holistic, digitally knowledgeable overview can increase OEE by as a lot as 20 %. In a single use case, take a look at parameter optimization leveraging digital and analytics led to a 13 % discount in take a look at time (Exhibit 2).
A standard thread working by means of these experiences is that information is the lifeblood of profitable transformation, supporting optimization finish to finish and guaranteeing that upkeep—for instance, of handlers and testers—is well timed and efficient (Exhibit 3). In a single pilot program, an organization achieved a 40 % discount in downtime by introducing predictive upkeep processes. One other data-driven unlock is RTD (real-time dispatching), which might be linked to broader provide chain coordination. As well as, information can inform administration of complete price efficiency—for instance, in relation to procurement and labor.
Three game-changing focus areas
The most important hurdles to operationalizing back-end modifications are sometimes in three distinct enterprise areas: working techniques, administration techniques, and mindsets and behaviors. Profitable transformations will virtually inevitably absorb these parameters, with options designed to deal with particular ache factors within the manufacturing course of.
Working techniques
A standard problem is that firms do not need the infrastructure to detect or treatment inefficiencies in shop-floor operations. Thus the transformation would ideally embed superior capabilities into the working system. Enhancements would come with visible administration strategies, together with coloration coding tons or lot placements in particular labeled areas. They might additionally embrace work-in-progress movement routines and normal working procedures (SOP) to observe changeovers and operator efficiency. An usually ignored issue is machine structure, which must be optimized to cut back ready instances and enhance materials movement.
Some main firms are turning to a variety of Trade 4.0 levers to spice up productiveness, which have to be seamlessly built-in into standing working techniques. Among the many only are these:
- Web of Issues sensors will increase OEE transparency and provide detailed perception into microstoppages in addition to OEE losses that aren’t readily obvious from high-level information (for instance, unmonitored/unaccounted OEE losses or defective inputs, resembling misallocation of loss buckets).
- Actual-time dispatch will increase utilization charges. Right here, software program displays materials in manufacturing and allocates it to machines to reduce wait instances and maximize throughput. In the meantime, information can inform planning to extra successfully handle distribution, dispatch materials to underutilized corridors, and keep away from waste.
- Superior yield analytics can hint faulty dies to front-end wafer positions. This turns into more and more essential for superior packages, the place each electrical and thermal failure modes are important. It’s going to additionally facilitate the gradual introduction of superior transistor constructions—together with gate-all-around (GAA) field-effect transistors (FETs) and nanosheet FETs—that are at the moment resistant to traditional defect detection resulting from their small measurement and three-dimensional form.
- Superior take a look at analytics would take away pointless take a look at blocks, resembling these with low failure charges. This will likely be more and more essential as chip complexity rises, pushed by ever-growing numbers of transistors per chip and elevated bundle complexity. These, in flip, would drive necessities for longer take a look at instances on excessive capital expenditure take a look at gear.
- Automated guided autos (AGVs) and autonomous cellular robots (AMR) can enhance materials dealing with and streamline operations in greenfield back-end fabs. Many greyfield fabs are usually not set as much as maximize their utilization. A key motive is area constraints and the necessity to accommodate the variety of AGVs/AMRs required.
- Video analytics can streamline guide meeting. In distinction to front-end manufacturing, there are a number of visible inspection steps on the again finish that require important operator time. Examples embrace the inspection and maintenance of burn-in boards, defect inspection after wire bonding, and inspection of molds. Right here, computerized visible inspection, powered by machine-learning algorithms, may help streamline processes and scale back labor.
- Analytics-driven course of parameter optimization, particularly testing and burn-in testing, is commonly tormented by suboptimal handler operation, in addition to defective or inactive take a look at websites, resulting in constraints on throughput. Superior analytics can align handler speeds throughout tester fleets, whereas machine-learning algorithms on take a look at boards can predict website failures.
- Digital twins (that’s, the digital mannequin of a product/manufacturing system) are already deployed by some back-end producers in capability, upkeep, and supply-chain optimization. A extra highly effective utility would holistically mannequin the interaction between provide chain, WIP (work in course of), operators, and machine and upkeep to pinpoint yield points, predict WIP bubbles and ODDs (on-time supply dates), and speed up manufacturing ramps. A standard problem is a scarcity of knowledge availability paired with lacking real-world suggestions loops.
Administration techniques
At senior ranges, a typical problem is a weak KPI construction and guide or semi-automated information gathering from unconnected techniques. This results in a scarcity of transparency and infrequently delayed motion. As well as, high-level targets are usually not translated into shop-floor and shift targets, reflecting a disconnect between senior administration and each day operations. Generally, shop-floor operators are usually not clear on expectations, resulting in restricted engagement and low retention charges.
Disconnect might be mitigated by means of clear KPIs and using dashboards for oversight and monitoring. These must be listed to each day shop-floor operations (as an illustration, dies produced or examined per shift and space or general yield per shift). Common conferences on the senior stage must be held to observe processes and resolve bottlenecks. The place targets are missed, senior administration ought to help concern decision, take part in Gemba walks (a lean strategy the place managers bodily go to the store flooring), and work to take away potential roadblocks. Conversely, the place targets are met or exceeded, achievements must be acknowledged and rewarded, then adjusted to seize enhancements.
Mindsets and behaviors
Efficient organizational change relies on tradition and mindset. Easy measures resembling common suggestions classes, presentation of successes to senior managers, and tailor-made studying modules (for instance, on automation, manufacturing rules, and high quality) can stoke enthusiasm and guarantee resolution making is aligned.
Usually, a change in management fashion is required. Leaders will apply the chief “Commonplace Work,” defining what, when, and the way leaders ought to take motion. This may increasingly imply, for instance, being on the store flooring for Gemba walks and course of confirmations, supporting root trigger problem-solving, and providing common suggestions and training. Shift managers, in the meantime, ought to work to spice up autonomy at shop-floor stage and lead efficiency dialogues. Workforce-building initiatives may help forge a way of collaboration and studying. On the identical time, a tailor-made capability-building program round manufacturing excellence and industrial automation will assist embed and maintain the transformation.
Lastly, firms shouldn’t be embarrassed to have a good time their successes. It is smart to articulate and talk the transformation change story, deal with cultural resistance head-on, and showcase how personal- and operational-performance enhancements go hand in hand.
Transformation of back-end operations can have a big impression on plant throughput, productiveness, and efficiency, however the impression is contingent on the effectiveness of change initiatives. Essentially the most profitable are constructed on three pillars—working techniques, administration techniques, and mindsets and behaviors. Our expertise exhibits {that a} mixture of analytical and automation instruments, enhanced oversight, and cultural change can produce ends in only one or two months, whereas capability positive factors accrue over one to 2 years. With these advantages in thoughts, resolution makers have a chance to refocus on execution, resolve bottlenecks, and transfer towards a extra strong and decided back-end technique.
[ad_2]
