New Challenges Facing Semiconductors

The semiconductor industry faces the challenges and opportunities of increased product demand in the immediate future. The growth of artificial intelligence (AI) and the Internet of Things (IoT) and the ongoing demands from the smartphone sector and other high-tech industries will place stress on the semiconductor supply chain. The challenge will be further complicated by ongoing international trade disputes, which may drive up the cost of semiconductor materials and interfere with global collaboration within the industry.

New solutions are needed to face these demands. Recycling semiconductor materials at device end-of-life continues to prove difficult, as does the need to produce semiconductor devices using more sustainable processes to reduce the emissions of toxic pollutants during the manufacturing process.

 

Challenges to automation in the semiconductor industry

The need to streamline the semiconductor-industry supply chain is evident. At present, it can take up to six months to complete production of an integrated circuit, not counting packaging and delivery of chips to the buyer. Semiconductor companies need smoother, more efficient manufacturing processes.

Automation offers potential solutions. However, automating semiconductor-manufacturing processes isn’t without its own challenges.

Why automate semiconductor manufacturing?

To grasp the effects that challenges to automation have on the industry, one should first understand the basics of why the industry should automate to begin with.

Automation improves ROI

Automating workloads and improving process reliability results in greater and more consistent wafer yields and less material waste. According to an IBM study, 100 percent of electronics executives—including semiconductor manufacturers—plan to implement or are in the process of implementing AI into their manufacturing process, with 83 percent reporting moderate to significant ROI due to improved yield predictions.

Automation improves process reliability

Given the length of time needed to process semiconductor devices, downtime at any point in the manufacturing process represents a financial and material cost. In wafer transportation, human handling would carry with it the risk of human error. Given the thousands of steps in the manufacturing process, errors in transporting wafers to the correct machine in the correct manufacturing sequence are all too likely.

Furthermore, human handling would risk contaminating wafers with dust particulate, even in a clean room, which controls airborne particulate concentration. Semiconductor manufacturers use front-opening unified pods (FOUPs), which automated material handling systems (AMHSs) control, to prevent contamination and ensure that each wafer is transported and positioned with precision.

Errors can also occur when testing performance criteria. To reduce this risk, performance evaluation is increasingly automated, using smarter systems to test components for efficiency both individually and when assembled into larger semiconductor devices.

This is not to say that automated systems are infallible. Automated equipment failure can bring a manufacturing process to a standstill. The use of industrial IoT sensors and predictive maintenance helps semiconductor manufacturers detect developing equipment problems in their early stages, reducing production downtime.

Intelligent automation prevents reliability issues and yield losses

AI goes hand in hand with automation. Smart machines help increase facility capabilities while identifying and helping eliminate production bottlenecks.

To prevent product reliability issues and yield losses, intelligent automated systems employ fault detection and classification (FDC) systems to identify production abnormalities—outliers in machine parameters and sensing data. Given the narrow margin for error when manufacturing integrated circuits, this feature alone makes automation an important industry tool.

What is the main challenge to automating processes in the semiconductor industry?

Semiconductor automation systems must be able to control wafer fabrication throughout its many steps, which must continuously change in the face of evolving demands placed on semiconductors themselves. This puts pressure on automation systems to continuously tune the fabrication process and its many steps.

Semiconductor manufacturers mustn’t stop investing in human and AI oversight to make sure their automation frameworks stay on top of tuning-wafer fabrication and other processes. In other words, automation is a must, but it comes with a hefty price tag—attached not only to retrofitting facilities for automated processes but also to keeping those processes working.

How outsourcing and collaboration affects semiconductor automation

The need for automation has increased the willingness of many companies to outsource aspects of wafer processing. Doing so allows them to shift the costs of automation to partners while focusing on internal resources and strategic differentials. Outsourcing noncore engineering has the benefit of streamlining the supply chain without the financial cost of retrofitting facilities with automated features.

Outsourcing wafer processing, however, is not without its potential problems. The foundry used to manufacture wafers may be in a different country (many of the major foundries, for instance, are in the Asia-Pacific region). This adds a global element to a company’s supply chain that, while profitable during stable economic times, can become an issue when trade wars develop and nations use tariffs and other economic weapons to punish each other.

How are new developments in automation impacting semiconductor manufacturing for AI?

At present, semiconductor manufacturing produces 7 nanometer (nm) chips, and the industry is exploring the possibilities of 5 nm and even 3 nm chips. Manufacturing components at such small scales exposes production to electrostatic and quantum effects that can impact performance and reduce wafer yields.

Similarly, imperfections in the silicon substrate that could once be ignored now have the potential to render entire runs of wafers unusable. AI can help address these issues in ways manual production cannot.

For instance, AI may be able to detect anomalies in production test results or sensor data and judge how likely certain causes, such as material impurities or damage from electrostatic discharge (ESD), are responsible. From this information, control engineers can adjust fab processes accordingly to mitigate production downturns from substrate imperfections.

Read more about new challenges facing semiconductors in the IRDS™ Roadmap

Access the 2020 IRDS™ Roadmap

 

Recycling semiconductor materials

Many of the materials used in semiconductor manufacturing are valuable and well worth recycling at the end of a device’s use life. The minuscule amounts of materials found in individual integrated circuits make reclaiming used materials difficult, however, and almost 70 percent of electronic products (PDF, 13 MB) wind up as unrecycled trash.

If the future of semiconductors is to be sustainable, significant efforts will need to be made in the fields of recycling and materials reclamation.

How does recycling introduce new challenges to the semiconductor industry?

According to the World Materials Forum (PDF, 13 MB), recycling a million smartphones reclaims the following materials:

  • copper: 16,000 kg (35,274 lb)
  • silver: 350 kg (772 lb)
  • gold: 34 kg (75 lb)
  • palladium: 15 kg (33 lb)

The obvious problem is one of scale. The amount of materials the industry reclaims through recycling is not economically viable. Individual semiconductor devices contain such minuscule amounts of valuable materials that reclaiming them is a costly endeavor.

Recycling semiconductor materials not only costs money—the process of reclaiming semiconductor materials from e-waste also generates its own waste, much of it toxic. Complicating matters further, recycling e-waste is often outsourced to developing countries, where local recycling firms often use child labor to sort recyclables, manually recycle products, and burn unusable parts.

According to the World Health Organization (PDF, 2 MB), children working in such conditions are exposed to lead, lithium, cadmium, mercury, and chromium. While it’s difficult to determine how much child labor is used in electronics recycling at a global level, the International Labour Organization estimates 35,000 to 45,000 children (PDF, 2 MB) work as waste pickers and dismantlers in Delhi, India, alone.

Furthermore, wastewater produced during deposition and etching can contain arsenic and other contaminants while incinerated chemicals are released into the atmosphere. It’s at the production stage, rather than device end-of-life, that the industry has the best chance of reclaiming wasted materials, including tantalum, an important component of next-generation semiconductors.

How important is the quality of tantalum used in semiconductor chips?

Tantalum is a rare, hard, and corrosion-resistant metal commonly used to create semiconductor capacitors. Considered a technology-critical element due to its demand in the semiconductor industry and relative scarcity, tantalum is found in mineral groups such as tantalite, columbite, and coltan, and always in the presence of niobium.

To be of use in the semiconductor industry, tantalum must be extracted from minerals and refined. This is usually accomplished by exposing tantalum-containing minerals to superheated hydrofluoric or sulfuric acids, dissolving the tantalum, niobium, and other elements. The resulting slurry is then filtered and processed to extract the tantalum. Wastewater from the process includes acids and trace amounts of rare earth metals, thorium, uranium, manganese, and titanium, among other substances.

Highly refined tantalum is needed due to the increasingly thin layers used to make integrated circuits. Semiconductor layers are now less than 50 nanometers thick. At such levels, any impurities will negatively impact tantalum’s ability to improve particle free–process consistency and metal-sputtering uniformity (the depositing of thin metal films on the wafer).

Why is silicon mostly used as a semiconductor?

Despite advances in semiconductor materials, silicon remains the most common base for wafers. This is due to its ability to act as conductor and insulator, depending on heat and energy exposure, and to the simple fact that it’s one of the most abundant elements on earth. Silicon and silicon dioxide are easily refined from common sand.

The most abundant semiconductor material may also be the one most easily recycled. Short-loop silicon wafers are used during process development. These are used to evaluate the reliability of particular components and then recycled for reuse.

Read more about new challenges facing semiconductors in the IRDS™ Roadmap

Access the 2020 IRDS™ Roadmap

 

Robotics, artificial intelligence, and semiconductor manufacturing

Robotics has long been part of the semiconductor industry. When dealing with the microscopic tasks needed to create semiconductor applications, the use of highly precise robotics has proved the most efficient way to prepare devices for market.

How can the semiconductor industry use robotics to meet new challenges?

The market for semiconductors demands integrated circuits and semiconductor applications quickly, while insisting on high standards of reliability and performance. Using small robots to increase production speed without sacrificing quality only makes sense.

High-speed delta-style robots with six-axis articulation offer the speed and precision required to orient semiconductor parts in two or more planes while also assembling circuit board packaging. Such robots perform these tasks more efficiently than high-speed chip shooters.

What hardware components will be critical in driving innovation in new IoT devices?

Automated manufacturing will make it possible for the industry to meet the hardware demands of the Internet of Things, 5G infrastructure, and autonomous vehicles.

Microelectromechanical systems, sensors, LEDs, and flexible display technology are needed for 5G devices, smart cities, smart homes, and autonomous vehicles. Robotic manufacturing processes will be able to meet demand while providing high degrees of performance and reliability—and reliability will be essential. Flaws in industrial applications and autonomous vehicles could have fatal consequences.

What should semiconductor manufacturers consider when designing systems for the Chinese robotics market?

The relationship between robotics and semiconductor manufacturing is a symbiotic one. Robots and AI allow for more precise chip manufacturing, while integrated circuits and other semiconductor applications are essential for the robotics market.

China is a major player in the robotics market. Backed by the state, China’s robotics industry can shift gears quickly in response to market conditions. When the economy is strong, the industry expects semiconductor manufacturers to provide needed applications quickly, which can put a strain on the semiconductor supply chain.

Despite the increasing importance of Huawei’s native-made chips, the Chinese robotic industry still relies heavily on US-made semiconductor devices. China’s own chip foundries lag behind when it comes to manufacturing technology—while Samsung and Intel are using 7 nm technology, SMIC (the leading Chinese foundry) is still working with 145 nm.

What the Chinese semiconductor sector lacks in technology, however, they make up for with strong state influence, which could have a significant impact on global semiconductor manufacturing and the AI market, given current economic tensions between China and the USA.

Read more about new challenges facing semiconductors in the IRDS™ Roadmap

Access the 2020 IRDS™ Roadmap

 

The implications of the US/China trade war for semiconductor firms

Despite the global collaboration that drives the IRDS™ process, semiconductor manufacturers remain vulnerable to disruptions to the supply chain caused by international trade disputes. At present, the ongoing trade war between the United States and China has semiconductor firms in both countries worried, as tariffs and political negotiations threaten the manufacturing and delivery processes.

What role do semiconductors play in international trade?

Semiconductors are the bedrock of the international electronics trade, as they are needed for every electronic device in use today, from televisions and computers to electric cars and utility sensors. As an example of their importance, semiconductors are the USA’s fourth-largest export after airplanes, refined oil, and crude oil—and aircraft themselves require multiple semiconductor components. The computers and devices used to track international trade are themselves dependent on semiconductors.

American semiconductor firms control almost half the industry’s global market share. Most American firms have supply chains that rely on companies in other countries and regions for the testing, manufacturing, assembling, and packaging of their core products. Many of these countries and regions, including Malaysia, China, and Taiwan are located in the Asia-Pacific region.

This setup lends itself to a global market. A company might develop and test semiconductor designs in its home country, purchase raw materials from another, and use firms in a third country to manufacture and package its products. The system has worked well for semiconductor firms internationally but relies on a degree of trade cooperation between nations—cooperation that can be disrupted by trade wars.

How are semiconductors affected by the trade war with China?

Unfortunately for the industry, semiconductor manufacturing is an area where American and Chinese interests clash. US tariffs on Chinese imports have made semiconductor materials and semiconductor devices more expensive. The industry has concerns that, should the trade war continue, China may opt to restrict or ban exports of rare earth metals to the USA, which would drastically impact the US semiconductor industry’s ability to manufacture components.

China, currently a major importer of semiconductors, plans to develop its own chip design and manufacturing industry. The Chinese government has announced its intention to supply 40 percent of its own semiconductors by the end of 2020, with the longer-term goal of reaching 70 percent homegrown chip production by 2025.

This represents a challenge to the USA’s semiconductor market share. US firms Qualcomm, Broadcom, and Micron see more than 50 percent of their profits come from China. US tariffs on semiconductor-related products threaten to slow China’s economic development and encourage retaliatory action by the Asian nation. US tariffs also affect US companies: a company that has outsourced manufacturing to China will see its imports rise in expense, affecting revenue.

How do states exercise power through global economic networks?

Broadly speaking, states flex their global economic muscles by controlling access to needed nodes of production. In China’s case, this means controlling the export of rare earth metals. At present, China controls 80 percent of the rare earth metals imported by the United States, and 85 percent of the global processing of rare earth metals.

Rare earth metals are essential for multiple industries, including the production of rechargeable batteries, wind turbines, electric cars, computers, televisions, fiber optics, and supercomputers. They are also needed for military applications, including missile-guidance systems, satellites, and jet engines. Should the US-China trade war continue, restricting US access to rare earth metals would grant China serious bargaining power.

Read more about new challenges facing semiconductors in the IRDS™ Roadmap

Access the 2020 IRDS™ Roadmap

 

Supply chain issues and semiconductor manufacturing

Historically, the semiconductor industry enjoyed a relatively stable global market, with the bulk of demands on the supply chain coming from smartphones and industrial equipment. This made monitoring supply chain inventory, manufacturing, sales, and even research and development fairly predictable. If semiconductor manufacturers met the demands of these sectors, they would most likely see moderate growth and increased profits.

The advent of AI, the IoT, and autonomous vehicles have changed the semiconductor-industry outlook. Today’s semiconductor firms need to be more flexible, with a greater focus on research and development; increased functionality; and shorter, more efficient production times. Such changes have led to significant challenges to the traditional semiconductor supply chain.

What new challenges are facing the semiconductor industry because of supply chain changes?

The greatest challenge to the semiconductor supply chain remains Moore’s law, or rather, finding ways to circumvent the limitations of Moore’s law to provide smaller, more powerful chips at affordable costs. System in package and similar innovations make this possible but also increase the complexity of manufacturing processes.

Few companies will be able to meet this manufacturing challenge alone. Collaboration will become more important. All levels of production, including manufacturers themselves, foundries, assembly and testing firms, material providers, manufacturing tool suppliers, and component companies will need to form an interactive, organic supply chain to meet new demands for greater capacity, performance, and manufacturing costs.

More efficient supply chains are essential. AI and IoT companies demand an ever-increasing range of functionality and performance with shorter delivery times—a problem when supply chain lead times for most semiconductor devices are as long as twenty-eight weeks. Firms that can streamline their supply chains while shortening lead times will have a distinct advantage.

What is the role of chip manufacturing in the semiconductor supply chain?

Efficient, timely chip manufacturing is vital to the semiconductor supply chain, both at the front end and back end.

Front end refers to the manufacturing of semiconductor chips, starting with the purification of the wafer and building the multiple layers needed to create integrated circuits. The most complex step in the supply chain, it is at this point that the greatest risk of bottlenecks, errors, and mechanical delays exists.

The back end refers to the assembly and packaging of multiple components into the final semiconductor product. A highly precise process, the back end must function efficiently to ensure prompt lead times and minimize yield loss.

Changes to the supply chains pose challenges, of course. As AI and the IoT become part of the manufacturing process, companies need to invest in retrofitting or completely overhauling chip foundries. Doing so will result in delays in production, but not doing so risks falling behind competitors who embrace the new technology.

What functions in the supply chain are critical to semiconductors?

It is perhaps indicative of the complex, interrelated nature of the semiconductor supply chain that no single function stands out as more important than the others. Manufacturing, sales, research and development, and marketing must come together as a whole for success. That said, most firms can stand to improve performance in four specific functional areas:

  • Marketing requires careful attention to customer input, a pricing strategy in tune with market forces, and a deep understanding of device architecture and how products address and resolve customer pain points. With AI and IoT devices often requiring devices customized to their needs, semiconductor marketing needs to focus more on individual consumer needs.
  • Research and Development needs to look forward, not only to new products and solutions but also to product sustainment, legacy, and end-of-life cycles. With demands for ever-smaller, ever-more-powerful chips, research and development is an essential part of the semiconductor supply chain and requires high levels of investment from companies.
  • Manufacturing is vital to supply chain strategy and needs to expand beyond production and quality control to consider capacity planning, logistics, and manufacturing outsourcing. The more efficient a semiconductor supplier’s manufacturing process, the faster and more efficiently the company can meet consumer demand.
  • Fulfillment must find new, more efficient ways to manage orders, distribute products, and deliver within lead times while developing the ability to operate at deeper capacity. Fulfilment offers the chance to develop long-lasting relationships with satisfied customers.

The semiconductor industry faces many internal challenges, from materials handling to recycling to process improvements. It also faces growing external challenges such as geo-politics tied to resource availability and trade. However, the overall semiconductor market continues to grow rapidly. Growth in demand shows no sign of slowing down. Companies that are willing and able to invest in the future and successfully navigate these challenges are sure to thrive. 

Interested in learning more about challenges facing the semiconductor industry? Consider reading the International Roadmap for Devices and Systems (IRDS™). The IRDS™ is a set of predictions that examine the future of the electronics, semiconductor, and computer industries over a fifteen-year horizon. It encompasses a number of critical domains and technologies, from application needs down through devices and manufacturing. Join the IRDS™ Technical Community to download the roadmap and stay informed of our latest activities.

How to Download IRDS™

Access the 2020 IRDS™ Roadmap