Part of the  

Solid State Technology

  Network

About  |  Contact

Transition to ISO 9001:2015: Starting the Journey

October 21st, 2015

The official introduction of ISO 9001: 2015 in September provides an opportunity for a completely fresh look at Quality Management Systems (QMS). While at first it may appear that there are many changes within the detail of the standard, it is important to remember that the key management principle of CUSTOMER FOCUS remains unchanged.

The most visible change from the 2008 standard is the adoption of the High Level Structure. This will form the basis of all ISO management systems going forward to give them all the same look and feel along with identical core text and common definitions. This will greatly aid organizations in integrating multiple management system standards effectively and efficiently. For example, their management systems for quality (ISO 9001), environment (ISO 14001) and safety (ISO 45001), which will replace OSHAS 18001 next year) will all have the same structure and definitions, thus avoiding conflicts, duplication, and potential misunderstanding across standards.

Within the standard and the guiding principles, there is now much more emphasis on LEADERSHIP (section 5) and ENGAGEMENT rather than management and involvement. This is consistent with other organizational development processes that move from a basic implementation of the management principles to one that is embraced and ‘lived’ by the whole organization.

This engaged activity needs to both reflect and support the CONTEXT OF THE ORGANIZATION (section 4). This is a new section that defines the external and internal issues that can impact what the organization does not only in its current activities but also in its strategic direction. This allows for the management system to not only cover legal, regulatory, or contractual requirements, but also more forward-looking market assurance and governance goals. The scope of the management system needs careful definition to ensure it is meeting the needs and expectations of all interested parties.

This more forward-looking and proactive approach is also reflected in PLANNING (section 6) and IMPROVEMENT (section 10) where there is a new focus on RISK and OPPORTUNITY management all within the context of the organization. This provides greater emphasis on improving processes to not just prevent non-conformities, but also on improving products and services to meet known and predicted requirements. For the electronics industries in particular, this last requirement is critical. Together we face demanding customer needs to meet advanced technology node requirements in terms of purity and process control; the revised ISO 9001 quality management system provides guidelines to help materials providers work toward these requirements in a structured and consistent manner.

We know our customers’ requirements continually evolve and our management systems need to reflect this. This new ISO standard provides impetus to once again revise and update our quality and other management systems to ensure customer needs are met both now and in the future. Our teams are actively planning for the ISO 9001: 2015 changes and we are happy to discuss our high-level planning with our partners.

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++


This blog post was contributed by Greg Shuttleworth, Technical Quality Manager, Linde Electronics. For more information, contact Francesca Brava at francesca.brava@linde.com.

Advanced Node Semiconductor Materials Reliance on SQC

October 8th, 2015

The semiconductor manufacturing industry is driven by the continuous quest for economical production of smaller, more powerful, and more energy-efficient chips in a fiercely competitive environment. Fabrication facilities operate up to 24 hours a day, 7 days a week, 365 days a year producing hundreds of identical die (unpackaged chips) on single silicon wafers that each undergo over 600 processing steps, taking up to 60 days of processing time from start to finish.

Companies at the forefront of the industry are utilizing advanced node semiconductor manufacturing techniques to produce the current generation of chips used in mass market consumer electronic devices. A chip can contain billions of transistors and the structure of these units has evolved from planar to 3D architecture. The latest technology node in production uses devices where the smallest printed feature is just 14nm in width (by contrast, a single piece of paper is 100,000nm thick) and deposited films can be counted in numbers of atomic layers.

Implementing an additional dimension to transistor architecture increases both manufacturing complexity and cost. Enormous financial commitments are required to be competitive in this industry, with annual R&D and capital investments in the order of $10B USD. It is no wonder that quality expectations of materials used in advanced node semiconductor manufacturing are unrivalled in their stringency.

One tool Linde uses to ensure that their suppliers meet these meticulous quality requirements is Statistical Quality Control (SQC): the use of statistical methods to monitor and control a process. In SQC, process data is used to calculate upper and lower control limits, which are distinct from specification limits. There are a number of control chart trend rules (Western Electric Rules) used in SQC to identify unusual occurrences. The most basic of these is rule #1 – one point beyond either control limit – and this must be adopted in application of SQC.

Linde requests SQC control charts from their suppliers on a quarterly basis to ensure that incoming materials received are high-quality, consistent, and predictable in performance and that potential instabilities in product characteristics are identified early.

The figure below shows a simplified example of a SQC chart for Material Y produced by a supplier using moisture content as a key control parameter being monitored and assuming one batch produced per day. All values are hypothetical.

This level of focus on stability is crucial to semiconductor customers, who can already link deposited film thickness variations to minor cylinder-to-cylinder product differences. As semiconductor manufacturing technology continues to advance to smaller and more complex geometries, there will be even greater emphasis on impurity specifications and tighter control limits.

Linde’s semiconductor customers inhabit a global, fast-paced, high-precision manufacturing environment. Supporting customers in this environment requires Linde to engage in active collaboration with their suppliers in areas such as implementation of robust SQC processes. Through this approach, Linde contributes to establishing the high-quality supply chain required to enable success for their customers.

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

This blog post was contributed by Ryan McGrath, Head of Global Quality Management, Linde Electronics. For more information, contact Francesca Brava at francesca.brava@linde.com.

Fabs Seeking Higher Quality Electronic Materials to Meet Technology Demands

August 27th, 2015

Manufacturers are striving to overcome limits to stay on track with Moore’s Law. Typically as the technology node gets smaller, the number of processes goes up and yield potentially goes down with each added process step.

Critical process steps in high-volume semiconductor device manufacturing at aggressive feature sizes require stringent control of variability. For a silicon wafer with 100 or more advanced logic chips, each with up to 4 billion transistors and billions of connections, it is critical to remember:

  • Essentially all the transistors and connections have to work as intended on each chip and
  • The process has to be repeatable from wafer to wafer while chip production proceeds at rates of up to 80,000 wafer starts or more per month through a fab.

A modern high-volume semiconductor fab hugely amplifies value and cannot afford any process excursions. There must be stringent focus on controlling variation in all inputs to the chip fabrication process.

Variation among transistors on a chip lead to poorer overall chip performance and must be minimized. Even trace contaminants – including those that are not specified on a standard Certificate of Analysis – can cause measurable shifts in semiconductor processes and affect chip performance in advanced devices. Given that process materials are a critical input in wafer processing, it is easy to see how the quality of electronic materials (EM) products becomes increasingly important for chip manufacturers at leading technology nodes.

Another important consideration is the challenge of the unknown: engineers don’t know how a specific impurity might impact performance. This can lead to needing additional processes and controls, which can mean higher operational costs and more risk from higher investments. Any misstep along the way – an impurity in a gas, for example—might interact in the process in unknown ways. Such a misstep can cost thousands or even millions of dollars per month.

Ensuring consistent product requires a holistic approach to quality. Instead of limiting responsibility to a quality department, it must be a priority that runs through the entire organization. As is seen in this wheel, a comprehensive quality strategy cuts across all functions that touch a product.

To meet the demands for rigorous quality control, organizations may need to hire materials scientists, chemists, and process engineers and change the culture of their organization so that every department has a strategy and plan that contributes to the overall quality vision.

Process stability across the supply chain is made possible through SPC (Statistical Process Control), SQC (Statistical Quality Control), MSA (Measurement System Analysis), and BCP (Business Continuity Planning) systems. Fingerprinting and metrology furnish the means for rigorous measurement, reducing variability, and tightening controls. Gas purity, consistency, and reliability are then delivered as an integral part of the final product.

IC technology step changes are driving electronic materials purity and analytical requirements. The bottom line? Materials suppliers must reduce variability and tighten control limits to help fabs meet market demands for more complex devices.

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

This blog post was contributed by Dr. Anish Tolia, Head of Global Marketing, Linde Electronics. For more information, contact Francesca Brava at francesca.brava@linde.com.

Larger Fabs + Smaller Devices = More Gases

August 6th, 2015

Semiconductor manufacturing fabs are faced with intense business and technical challenges to meet the demands for and costs of ever smaller and more complex devices. Semiconductor manufacturers are pushing the limits of physics and driving a constant need for new materials. The highly competitive mobile devices market is forcing fabs to ramp to higher volumes faster than ever before to meet market demands. 2014 saw a 10% upsurge worldwide in integrated circuits. Additionally, development costs for new technology can exceed $2B.

There are several factors driving the increased consumption of gases, the first being the rapid deployment of very large fabs to realize economies of scale, which are essential for profitable operation. A typical logic foundry is now running at 80,000 WSPM (wafer starts per month) and a typical memory fab is running at 120,000 WSPM. Fabs are also concentrating in clusters such as the Hsinchu Science Park in Taiwan.

In order to meet the demand and technological challenges, a larger volume and variety of gases is needed. Hundreds of gases and chemicals are used in several hundred process steps — etching/cleaning, deposition, doping, purging, and lithography/patterning — in the manufacture of semiconductors.

As process technology nodes are getting smaller, the minimal feature size at 20 nm becomes smaller than the wavelength of light and necessitates workarounds like multi-patterning to overcome physical limitations. This increases the consumption of gases per wafer.

Because of the need for low power and high performance, which 2D devices cannot handle, the industry is moving to 3D devices, which increases circuit density. This move to 3D FinFET and 3D NAND and the corresponding move to increased transistor processing — epitaxy, etch, and ALD (atomic layer deposition) — drive the need for new and increased materials to construct more complex devices.

Here are a few examples of gases that semiconductor manufacturers are using in increasing quantities.

Increased use of nitrogen

The gas most consumed in the production of electronics is nitrogen (N2). Nitrogen is used for purging vacuum pumps, in abatement systems, and as a process gas.  As process nodes are driven down and the typical fab size has increased, nitrogen consumption has grown substantially. In large advanced fabs, there can be as much as 50,000 cubic meters per hour of nitrogen consumed, which compounds the need for cost-effective, low-energy, on-site nitrogen generators.

Increased use of hydrogen

Another electronics manufacturing gas that is seeing an increase due to larger fabs and increased capacity is hydrogen. Hydrogen is used during epitaxial deposition of silicon (Si) and silicon germanium (SiGe), as well as for surface preparation. Significant volumes of hydrogen usage are also anticipated to be used in extreme ultra violet (EUV) in the future as 450mm wafers enter production streams.

Increased use of rare gases

There is also an upswing in the need for rare gases such as neon, krypton, xenon, argon, and helium. This increased usage of gases that are not as readily available as nitrogen has driven sporadic temporary worldwide shortages, particularly helium and neon. Rare gases are also used for wafer cooling (helium), as source gases in lasers (neon), and as sputtering gases (argon and krypton).

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

This blog post was contributed by Dr. Anish Tolia, Head of Global Marketing, Linde Electronics. For more information, contact Francesca Brava at francesca.brava@linde.com.

The Increasing Demand for UHP Nitrogen and the SPECTRA Generator

June 25th, 2015

The electronics-related industries are fast moving, with a relentless drive to lower costs and higher performance. One factor that has remained constant throughout this change is the need for high-purity nitrogen for use in almost all of the process steps involved in converting a blank silicon wafer into a set of sellable working semiconductor devices. As the scale of the wafer fabs where these devices are produced has grown from the small and medium sized 6-inch and 8-inch fabs seen in the 1980s and 1990s to the latest wave of 300mm mega fabs, so too have the volumes of nitrogen being used in a particular location.

Typical nitrogen demand at an 8-inch fab in the early 1990s was 3,000 Nm3/hr.  Contrast this with the volumes being consumed now in the latest 300mm fab lines being added to the major campuses of customers, which can now reach as much as 50,000 Nm3/hr or more.

This remarkable progression has been accompanied by a similar growth in the size of the N2 generator plants we have deployed at such fabs, with the capacity of Linde’s SPECTRA-N plant increasing from the 1,000 Nm3/hr size of the first of the series (installed in 1990 in Japan) to the current development of a 55,000 Nm3/hr unit for use in Taiwan.  Throughout this time we have built and installed almost 40 SPECTRA-N plants to serve the semiconductor and TFT-LCD industries. The total combined production capacity of these generators is close to 450,000 Nm3/hr of nitrogen.

The success of the SPECTRA-N is based on three critical factors, which fit closely with the characteristics of the electronics industry: low power consumption, cost-effective capital, and speed of deployment. Having the capability to lead in all three of these dimensions has enabled the SPECTRA-N to become an established benchmark standard for on-site nitrogen generators throughout the industry.

The largest sized plant for the past five years has been the SPECTRA-30000, which can deliver up to 36,000 Nm3/hr of N2 at ultra high purity levels (each impurity below 1ppb), while having the lowest relative power consumption of any such plant in the industry.  This plant has contributed a great deal to the recent successes achieved by our business teams in the U.S. and Asia. SPECTRA-30000 generators have been an integral part of recent major project wins in LCD-TFT fabs in China and new semiconductor wafer fabs in Taiwan and the U.S.

With the strong support of Linde Engineering, we are continuing to drive to ever larger plants to meet the requirements of the latest mega fabs in both semiconductor and TFT-LCD, including ensuring we are prepared for the anticipated jump in nitrogen demand that will come with the introduction of 450mm wafer processing in the next few years.  Linde Engineering has completed the design for the SPECTRA-50000 generator, capable of delivering up to 55,000 Nm3/hr of UHP N2. The first of these plants has been installed in Taiwan and will begin operations in 2015.

The SPECTRA-N generator has been a cornerstone of our success in the electronics market over the past 20 years and one place where this is especially noticeable is Hsinchu in Taiwan.  The first nitrogen plant installed by Linde (then as BOCLH) to serve these early fabs had a capacity of 4,500 Nm3/hr.

As more wafer fabs were built in the Park in the late 1980s and early 1990s, the demand for high purity N2 increased, and BOCLH successfully won contracts to supply the majority of these new fabs and began to install additional nitrogen generators. Deciding to focus on a centralized supply model with a pipeline network to customers was key in building the future success of our electronics gases business in Taiwan. Through the mid 1990s a further three new plants were added, each a SPECTRA-N 5000 generator. A second operating site was established when the first became full, and the pipeline network expanded into new areas of the Park.

This pipeline was designed as a loop and as it developed, it became a series of several interlocking loops which provided more than one pathway for gas to travel from the plants to an individual customer. This ensured a level of redundancy that could prevent a failure in one section from interrupting nitrogen supply. It also allowed the pressures to be balanced across the system to enable optimum power loading to be achieved on the growing network of generators.

By 1998, the total N2 demand in the Hsinchu Park had risen close to 30,000 Nm3/hr and was continuing to grow with more and larger fabs being added by key customers, and particularly the foundries. In order to match the scale of this continuing new fab investment, it was clear that a larger scale of plant was required. In the period from 1998 to 2001, a total of four of our then largest SPECTRA generator – the SPECTRA-N 10000 – were installed, trebling the Hsinchu N2 production capacity.

Following this period of rapid growth, the demand in Hsinchu slowed for a few years; at the same time, the advanced semiconductor production industry was getting started in mainland China with a number of fab projects being invested by Taiwan-related companies. BOCLH was able to make use of one of the earliest of the Hsinchu SPECTRA 5,000 plants by relocating this to Shanghai to capture one of these new fab opportunities there and establish a new operating site and small pipeline grid.

The next phase of Taiwan development took off in 2004, by which time the largest SPECTRA-N plant made was now 15,000 Nm3/hr.  Two of these were installed in Hsinchu taking the total N2 production capacity to above 100,000 Nm3/hr. By now, the primary manufacturers in the Hsinchu Park had begun investing in 300mm wafer fabrication plants, the throughput capacities of which were also growing, further adding to the rate of increase in N2 demand. To keep pace with this, and to take advantage of the lower unit production cost, the first SPECTRA-N 30,000 was installed in Hsinchu in 2007.

In the years since then, the N2 demand has continued to rise, driven largely by the huge amounts of investment made by major customers in adding further 300mm wafer fab capacity in the Park. A second SPECTRA-N 30,000 plant was added in 2011 and in 2014 the largest ever SPECTRA-N generator at 55,000 Nm3/hr was added.

When this begins operations in 2015, it will bring our total production capacity in the Hsinchu Park to 220,000 Nm3/hr of high purity nitrogen, with a total of 15 plants having been installed there in the 27 years of our development of this critical piece of infrastructure. This Hsinchu cluster represents about 10% of advanced global semiconductor manufacturing capacity and Linde is proud of the role that our SPECTRA-N plants have played in supporting its development.

This blog post was contributed by David Pilgrim. For more information, contact Francesca Brava: francesca.brava@linde.com.

On-Site Hydrogen Solutions: Smart, Reliable, and Flexible

June 1st, 2015

Hydrogen in Electronics

Hydrogen (H2) is a commonly used gas in various steps of semiconductor manufacturing. It is used as a co-reactant in deposition and etch processes, for surface passivation, and for cleaning. One of the major uses of hydrogen is in epitaxial deposition of silicon to form the n and p wells in the transistor.  With the advent of 3D transistors, there has been an increase in the number and thickness of epitaxial Si, Si-C, and Si-Ge layers. Another potential source of H2 consumption is EUV, which requires a large continuous flow of H2 in order to maintain a reducing atmosphere in the source.

There has also been a trend to build larger fabs in clusters at a single site to enable greater economies of scale. Together, these factors have greatly increased the demand for hydrogen at semiconductor fab sites.

Typically hydrogen has been delivered in two forms: compressed gas in tube trailers or liquid hydrogen. The latter has been the preferred means of delivery in the U.S. and Europe since higher volumes can be delivered each time. However, transportation regulations in Asia prevent liquid hydrogen transport by road and there tube trailers are more commonly used.

Once the consumption of hydrogen exceeds 50m3/hr, it may be more economical to put a generator on the customer site. Linde offers a wide range of on-site H2 generators.

Hydrogen On-Site Solutions by Linde ECOVAR®

On-site supply solutions, including those to the electronics industry, fall under the Linde ECOVAR umbrella. All three main industrial gases—nitrogen, oxygen, and hydrogen—are represented within the ECOVAR portfolio. The portfolio gas generators/plants produce products over a wide range of flow, pressure, and purity to meet the most stringent customer demands.

The ECOVAR hydrogen supply solutions are known under the trade name HYDROSS™. They consist of three unique technologies:

  • Steam methane reforming (SMR)
  • Electrolysis
  • Methanol cracking

All three technologies require additional gas purification steps to remove impurities. These impurities are either created in the hydrogen generation process as by-products or are carried into the process by the feed gases themselves. Product hydrogen purities can range from 99.9% to 99.999% and can be achieved through the use of conventional pressure swing adsorption purifiers.

Choice of technology depends on a number of factors:

  • Customer preference
  • Flow
  • Purity
  • Cost and availability of electricity and feed stocks

In general, electrolysis will be used for low flows <200 Nm3/h and where power costs are low. Application of the methanol cracker technology is limited to geographic regions where stability of methanol feed stock pricing is achieved through subsidy. SMR technology is the most common means of hydrogen production due to the wide and increasing availability of low-cost natural gas.

Hydroprime™ HC300: Small-Scale SMR for Cost-Efficient Hydrogen Production

With the demand for secure on-site supply of high-quality hydrogen growing and with the availability of low-cost natural gas increasing, the development of a small-scale, standardized, skid-mounted SMR unit was what Linde had in mind to meet the market demand when it introduced the Hydroprime product line.

Within Linde, a development initiative was started in 2009 that incorporated cutting-edge design ideas with latest commercially available technology. Since inception, ECOVAR, HydroChem, and HYCO operations divisions have combined resources and best practices, which resulted in the design available to the market today. Currently, there are several of these plants, located throughout Europe and Asia, which reliably and safely serve a wide range of markets.

One of Linde’s latest HC300 designs, serving the metals market in Austria as of December 2013

The Hydroprime HC300 can produce 330 Nm3/h of hydrogen at 14 barg and with a purity of 99.999%.

Since these plants are skid mounted, multiple units can easily be connected in parallel to meet a wide product flow range.

In South Korea 4 x HC300 plants are connected side by side. They are supplying high-purity hydrogen to a common hydrogen header.

The HC300 plant footprint is 14 L x 3 W x 4 H meters. It is delivered to a customer’s site after a full factory authorization (cold test) to insure ease of field installation and to reduce commissioning time and costs.

A typical Hydroprime installation consists of the hydrogen plant, ancillary utility systems, and a backup system that consists of either tube trailers or liquid hydrogen with vaporization. At sites where merchant product is too far for reliable supply for the backup, a redundant HC300 plant can be installed.  Linde plants are monitored 24 / 7 via a remote operations center for quick response to plant issues and to insure uninterrupted product supply.

The development of the Hydroprime has resulted in a low-cost, reliable, flexible hydrogen supply solution that is a good alternative for the more established supply forms like electrolysers, tube trailers, or liquid hydrogen.

This blog post was contributed by Tony Moes, Head of Linde ECOVAR – Standard Plants, Global Business Unit Tonnage. For more information, contact Francesca Brava: francesca.brava@linde.com.

New Approaches to Small Problems

March 20th, 2015

The market expectations of modern electronics technology are changing the landscape in terms of performance and, in particular, power consumption, and new innovations are putting unprecedented demands on semiconductor devices. Internet of Things devices, for example, largely depend on a range of different sensors, and will require new architectures to handle the unprecedented levels of data and operations running through their slight form factors.

The continued shrinkage of semiconductor dimensions and the matching decreases in microchip size have corresponded to the principles of Moore’s Law with an uncanny reliability since the idea’s coining in 1965. However, the curtain is now closing on the era of predictable / conventional size reduction due to physical and material limitations.

Thus, in order to continue to deliver increased performance at lower costs and with a smaller footprint, different approaches are being explored. Companies can already combine multiple functions on a single chip–memory and logic devices, for example–or an Internet of Things device running multiple types of sensor through a single chip.

We have always known that we’d reach a point where conventional shrinking of semiconductor dimensions would begin to lose its effect, but now we are starting to tackle it head on. A leading U.S. semiconductor manufacturer got the ball rolling with their FinFET (or tri–gate) design in 2012 with its 3D transistors allowing designs that minimize current leakage; other companies look set to bring their own 3D chips to market.

At the same time, there’s a great deal of experimentation with a range of other approaches to semiconductor redesign. Memory device manufacturers, for instance, are looking to stack memory cells vertically on top of each other in order to make the most of a microchip’s limited space. Others, meanwhile, are examining the materials in the hope of using new, more efficient silicon–like materials in their chips.

Regardless of the approach taken, however, this step change in microchip creation means new material demands from chip makers and new manufacturing techniques to go with them.

The semiconductor industry has traditionally had to add new materials and process techniques to enhance the performance of the basic silicon building blocks with tungsten plugs, copper wiring / CMP, high–k metal gates, for example. Now, however, it is beginning to become impossible to extend conventional materials to meet the performance requirements. Germanium is already added to Si to introduce strain, but its high electron mobility means Germanium is also likely to become the material of the Fin itself and will be complemented by a corresponding Fin made of III–V material, in effect integrating three semiconductor materials into a single device.

Further innovation is required in the areas of lithography and etch. This is due to the delay in production suitability of the EUV lithography system proposed to print the very fine structures required for future technology nodes. Complex multi–patterning schemes using conventional lithography are already underway to compensate for this technology delay, requiring the use of carbon hard masks and the introduction of gases such as acetylene, propylene and carbonyl sulphide to the semiconductor fab. Printing the features is only half of the challenge; the structures also need to be etched. The introduction of new materials always presents some etch challenges as all materials etch at slightly different rates and the move to 3D structures, where very deep and narrow features need to be defined through a stack of different materials, will be a particularly difficult challenge to meet.

The microchip industry has continuously evolved to deliver amazing technological advances, but we are now seeing the start of a revolution in microchip design and manufacturing. The revolution will be slow but steady. Such is the pattern of the microchip industry, but it will need a succession of new materials at the ready, and, at Linde, we’re prepared to make sure the innovators have everything they need.

This blog post was contributed by Greg Shuttleworth, Global Product Manager at Linde Electronics. For more information, contact Francesca Brava: francesca.brava@linde.com.

The Greening of Semiconductors

March 3rd, 2015

There is an increasing demand for and focus on sustainable manufacturing that will contribute to a greening of semiconductors. This greening must be robust and responsive to change and cannot constrain the individual processes or operation of a fab.

Fabs are being driven to choose materials by the needs to:

  • Create ever-higher-performance devices to stay competitive
  • Reduce the material costs in a device
  • Realize process efficiencies and achieve zero negative process interaction throughput

 What can be achieved by the greening of semiconductor materials?

  • Lower facilities use of electricity, water, and abatement chemicals
  • Lower emissions of harmful substances into the atmosphere, waste water, and solid waste streams
  • Lower overall volume of solid and liquid wastes
  • Lower consumption of materials, especially those that are costly, finite, or associated with harmful extraction

 All of the above lead to reduced costs at scale.

How are these benefits achieved?

These benefits are achieved through investments in:

  • Material supply and packaging
  • Tool, pump, abatement, and waste stream design
  • Recovery technology

 Often the economic benefits require upfront CAPEX investment by suppliers and end users to achieve later OPEX cost savings. Boundary conditions can determine who bears the cost and who realizes the benefits.

How can materials suppliers contribute to the greening of semiconductors?

Materials suppliers can do the following to reduce the environmental footprint of fabs:

  • Limit emissions and waste over a product’s lifecycle, to include material production, delivery, and return/reclamation/disposal.
  • Where possible within material selection constraints, make direct, process-compatible substitutions that have less environmental impact. An example of this is using F2 (fluorine) rather than NF3 (nitrogen trifluoride) as a cleaning gas in plasma CVD chambers.
  • Package and process for efficient use. Often, headspaces are exhausted and heels are unused to prevent light and heavy contaminants. Better purification, quality control, packaging, and material property knowledge can reduce the amount of material lost to safeguarding quality.
  • Recover material from waste streams of such gases as He (helium), Ar (argon), Xe (xenon), and H2SO4 (sulphuric acid). Reference Linde article “Sustainability through Materials Recovery” at http://semimd.com/materials-matters/2014/07/21/sustainability-through-materials-recover/.
  • Consider the trade-offs of production location of global vs. regional vs. local vs. on-site that includes a yields matrix of benefits and costs for emissions, material cost, and supply chain stability.

 What can equipment suppliers and end users do to contribute to the greening of semiconductors?

They can:

  • Design equipment and processes for efficient use and easier recovery
  • Design pumps, abatement, and facilities to enable better abatement, lower dilution of waste, and the ability to isolate waste streams from specific recipe steps
  • Budget for material recovery in new fab construction
  • Develop strategies for recycling of valuable and toxic materials from finished products

 

This blog post was contributed by Paul Stockman, Commercialization Manager, Linde Electronics.

Paul Stockman

For more information, contact Francesca Brava: francesca.brava@linde.com.

Emerging Requirements for Electronic Materials Product Quality and Metrology

October 13th, 2014

In order to keep pace with Moore’s Law, semiconductor market leaders have had to adopt increasingly challenging technology roadmaps, which are leading to new demands on electronic materials (EM) product quality for leading-edge chip manufacturing.

 Critical process steps in high-volume semiconductor device manufacturing at aggressive feature sizes require stringent control of variability. For a silicon wafer with 100 or more advanced logic chips, each with up to 4 billion transistors and billions of connections:

  •  Essentially all the transistors and connections have to work as intended on each chip and
  • The process has to be repeatable from wafer to wafer while chip production proceeds at rates of up to 50,000 wafer starts per month through a fab!

 Variation among otherwise identical transistors on a chip will lead to poorer overall chip performance and must be minimized. This issue is exacerbated by the fact that critical feature sizes on the latest semiconductor devices are now several thousand times smaller than the thickness of human hair.

 Variation in the form of a process excursion—when something unexpected happens in the manufacturing process—can also be very expensive. A starting wafer has a typical cost of $120. After the wafer has been subjected to several hundred process steps over a period of six to eight weeks, the investment in the finished wafer can be of the order of $1,000 to $4,000. However, based on the selling price of the chips, this same finished wafer can represent a value of as much as 10 times the processing cost in terms of the revenues it can bring in for the chip maker!

 Major industry players have pointed out that even trace contaminants—including those that are not specified on a standard Certificate of Analysis—can cause measurable shifts in semiconductor processes and affect chip performance as device geometries continue to shrink. There can sometimes be a lag of several days to weeks in the detection of problems during wafer processing. Small problems not detected early in the supply chain of the IC chip fab will increase exponentially in impact as value is added during wafer processing and can reportedly cause lost revenues of hundreds of millions of dollars in a worst-case scenario.

Given that EM products are a critical input in wafer processing, it is easy to see how the quality of EM products becomes increasingly important for chip manufacturers at leading technology nodes. Apart from focusing on major assay components, which are the impurities detailed in a Certificate of Analysis (CoA), some customers are also asking that minor assay components or other trace impurities must be controlled for critical materials used in advanced device manufacturing. EM suppliers usually only look for specified impurities and do not carry out a broad spectrum analysis due to the additional costs involved.

The need for tighter and more extensive control on gas purity now demands broad spectrum characterization. When carried out on product to be supplied to a customer, such “fingerprinting” can help us detect and measure impurities not formally specified on the CoA, but which could be present in the EM product and impact IC chip manufacturing processes where the EM product will be used. Broad spectrum characterization is also sometimes required to be carried out reactively in more of a forensic setting, e.g. when something goes wrong with a product during its use in a fab.

Thus, technology changes in semiconductor processing and demands for higher-purity and better-characterized electronic materials have driven the need for advanced analytical metrology. These and related changes are shown in the evolving roadmap of EM product quality below.

 In response to these stringent emerging requirements, Linde Electronics is expanding its analytical capabilities to enable broad spectrum characterization of EM products.

 In addition to the need for more advanced analytical technology, the EM supplier community is focusing on implementing a robust overall process control management system that leverages tools such as Statistical Quality Control (SQC), Statistical Process Control (SPC), measurement system analysis (MSA), and automated laboratory information management systems (LIMS) in response to the evolving customer requirements to control variability. 

As seen above, a full understanding of the electronic materials supply chain is required for successful introduction of new EM products and for their continued use in critical manufacturing steps in a fab. Achieving this also requires collaboration and trust between the supplier and the fab customer.

This blog post was contributed by Dr. Atul Athalye, Head of Technology, Linde Electronics.

 

 

TLIMS/SQC Process Control System

October 5th, 2014

Driving a high performance organization (HPO) to achieve LeadIng status in the industry requires establishing more cost-effective Process Control Systems with linkage to higher-level business results. To that end, Linde Electronics is in the process of implementing a system capable of real-time monitoring and reporting, across the supply chain, for the production of specialty gases.

Automating our processes allows us to allocate a larger percent of our resources toward continuous improvement and strategic efforts:  business continuity planning, improved cost management, improved productivity and traceability.

An increasing number of customer quality requirements is now mandatory to sustain current business and/or support new business opportunities.  Automating our process control systems allows us to achieve efficient response times, more visible quality control across product supply chains, improved tracking and traceability speed for supply chain routing – leading to improved customer confidence in Linde products and increased customer scores.

Linde Electronics has developed the TLIMS/SQC System. This system includes an information management database plus SQC/SPC software and delivers connectivity with SAP, electronically pulling order information from SAP to TLIMS and pushing CoA data from TLIMS to SAP.

Typically manually operated and separately managed, TLIMS/SQC System integrates and electronically links the following processes:

  • Data management and OQC-to-IQC traceability
  • Incoming raw/source material (IQC) data management
  • Control limit management across supply chain (IQC, IPQC, OQC)
  • SQC/SPC monitoring, flagging, and e-notification across supply chain (IQC, IPQC, OQC)
  • Defects/CONQ management and reporting

There are many benefits to doing this:


Our customers have given us very positive feedback on this system so far.  Here are some examples of what they have been saying about the TLIMS/SQC System:

  • “We recommend regular monitoring of SPC trends as well as implementing SPC rules to ensure targets are met and maintained. TLIMS is an excellent tool to provide this capability, including real-time SPC monitoring, electronic data connectivity plus instrument management and validation.”
  • “An integrated LIMS system is critical to providing real-time process monitoring, necessary to protect our interests, and therefore our confidence in Linde. We see this as a key component to maintaining and growing new business with Linde.”
  • “In the semiconductor industry, it is critical to have real-time SPC data to identify concerns and potential problems in the process so that we can provide our customers with the best quality product as possible. We expect our suppliers to be aligned and have the same capabilities to fully support us as a key customer.”

This blog post was contributed by Patricia Clarke, Technical Programs Manager, ESG SHEQ, Linde Electronics and Specialty Gases.

Next Page »