iNEMI ROADMAP IDENTIFIES TRENDS IMPACTING ELECTRONICS THERMAL MANAGEMENT | Electronics Cooling

iNEMI ROADMAP IDENTIFIES TRENDS
IMPACTING ELECTRONICS THERMAL MANAGEMENT

PREFACE

The International Electronics Manufacturing Initiative (iNEMI) is an industry-led consortium of approximately 100 leading electronics manufacturers, suppliers, associations, government agencies and universities. One of iNEMI’s key initiatives is its biennial roadmap, which looks at the future technology requirements of the global electronics industry. It provides a 10-year outlook for electronics manufacturing, anticipating technology needs and identifying gaps.

The iNEMI roadmap is unique in scope. It covers the full electronics manufacturing supply chain and is an important tool for focusing research & development (R&D) priorities. This article discusses highlights from the Thermal Management Chapter of the 2015 Roadmap [1].

1. INTRODUCTION

The Thermal Management Chapter of the 2015 iNEMI Roadmap describes future thermal management technology needs across a broad range of product sectors, including: high-end systems, consumer/office systems, portable and wireless products, medical devices and systems, and light emitting diodes (LEDs). It assesses the current state of the art for all areas and then identifies the technology gaps that the industry will need to address to create tomorrow’s products in these sectors.

The Thermal Management Roadmap also addresses the need to develop improved cooling technologies in terms of heat transfer processes, materials and innovative designs. If successfully implemented, enhanced thermal management will contribute to continued performance improvement trends and increased competitiveness of packaged electronic products. The roadmap identifies needs for further advances and developments in the following thermal technologies:

• thermal materials and thermal spreaders
• refrigeration cooling
• heat pipes
• liquid cooling
• thermal interfaces
• air cooling
• direct immersion cooling
• thermoelectric cooling, and
• thermal design modeling tools.

Here are highlights of some of these technologies, along with key challenges and R&D recommendations.

1.1 Overview/Situation Analysis

Demand for more effective and cost-efficient means of removing heat from electronic systems continues to grow across all segments of the industry—from compact, portable electronics to large, high-end systems. It is a particular concern for power electronics associated with transportation and the electrical grid. In the absence of major new breakthroughs in thermal management technology, this demand is being met by broader, more aggressive use of, and incremental improvements in existing techniques, along with increased emphasis on reducing the generation of heat through improved product design.

Aluminum and copper heat sinks, with forced or natural air circulation, continue to dominate in terms of unit volume in the high-volume personal computers and consumer electronics applications. The growing trend toward higher performance electronics in smaller packages is forcing designers and manufacturers to consider phase change and liquid cooling techniques, albeit at higher cost. These have not yet made it to high volume because of limited data about cost, reliability, and longer-term efficacy..

Increasingly, there is recognition that reducing system power (power in, heat out) will not only reduce the demands on thermal management technologies, hence their cost, but will also result in direct and indirect energy savings at system and facility/plant levels. This reduced energy consumption will eventually result in operating cost reduction—and is leading to improved design practices that focus more closely on reducing the sources of heat in electronic devices. One example of this is non-uniform heat generation on devices, which causes localized “hot-spots” and requires thermal management based on a worst-case scenario. In high-volume and critical applications, where cost is no object, custom designs of cold plates can effectively manage hot-spots locally, without need for the entire thermal solution being scaled for the worst-case hot-spot scenario.

Expanded adoption of multi-core central processing units (CPUs) continue to help lower the thermal impact of high-performance devices in computing applications and are helping to mitigate, though not eliminate, the need for even more costly and aggressive cooling solutions. Also, increasing acceptance of electric and hybrid vehicles, renewable energy solutions and evolution of the smart grid is placing growing emphasis on thermal management techniques suited to the unique requirements of power electronics. Most high-performance applications have specific thermal solution expectations and must be dealt with based on specific design requirements. High-volume customer solutions, where cost is a key concern, are still driving the technology toward simpler and lower-cost “elegant” solutions.

Cost and time-to-market continue to play a critical role in maintaining competitiveness for all product sectors. To keep pace with the shrinking design-cycle time and to reduce development costs, the industry will rely on advances in computer-aided thermal design tools.. Developments in thermal modeling tools to integrate electrical, thermal, fluid flow, and mechanical analysis and simulation in one user-friendly package continue to lag industry needs. The chapter identifies these tools as a major development need going forward. While many start-ups are addressing this gap, there have not been breakthroughs.

Thermal management has been a key enabling technology in the development of advanced micro-electronic packages and systems, and has facilitated many of the so-called Moore’s Law advances in computers and electronic products. Thermal management entails a balanced combination of materials and techniques to optimize performance-cost designs. Increased complexity, density, and higher clock frequencies continue to push the thermal fluxes at chip level. These thermal demands propagate through the consumption chain (sub-system, system and facility).

The tradeoff between peak and average power is an important concern: high peak power will often mean a more expensive thermal management solution, while high average power will result in higher energy cost over the life of the system. There is reconsideration of the die junction temperature and there is a tendency to lower the acceptable temperature from previous norms. This will improve reliability and device life, while putting even greater demands on the thermal management system.

The state of the art of appropriate cooling solutions for electronics depends on the system power and the financial budget available for the cooling solution. There are thermal solutions to handle some of the most demanding needs, but the technology to do so does not always meet cost constraints.

Whenever possible, air-cooling is used, as it is the lowest cost. For cost-sensitive computing (desktop PC, a declining sector) this means air cooling of a heat sink, usually with fans to force air onto the heat sink to improve performance. Notebook computers often use a heat spreader with a heat pipe and a fan. However, weight and volume limitations constrain laptops’ cooling performance, and that requires the use of lower power CPUs. Server computers use high-performance heat sinks and multiple fans to extend the limits of air cooling, and some medical equipment can afford liquid cooling. Acceptance of liquid cooling continues to increase as users get a better understanding of the costs and benefits associated with this solution.

The thermal management technology roadmap for typical power systems with different cooling schemes is shown in Figure 1. For 1U horizontal boxes with natural convection, systems now in the market are typically around 50W at ambient of 55°C at sea level.

Figure 1: Thermal Management Technology Requirements and Choices [Source: Huawei Technologies].

Thermal management costs have historically been a small percentage of total system cost, ranging from less than 1% for some PCs to 3-5% for some large servers, and approaching 10% on the largest supercomputers. Failure to consider thermal issues while the design is still flexible limits the ability to make design tradeoffs that might significantly improve thermal management cost, reliability and performance. System and chip-level product designers are now engaging the thermal design team early in the design cycles to take advantage of the potential to improve the performance contribution and cost of the thermal solution.

Additionally, from a design perspective, a unified approach, looking at the entire system design rather than “stacking” margins and specs at sub-system levels, may be another way of achieving a more efficient thermal management solution.

1. THERMAL MANAGEMENT TECHNOLOGY CHALLENGES

2.1 Power Electronics

Applications in power electronics have grown dramatically in the last few years because of greater need for electric power management and control (smart grid), renewable energy generation and control, and electric transportation—as well as a desire to improve operating efficiency of heavy systems, such as trains, industrial motors, and electric vehicles.

Power electronic converters are found wherever there is a need to modify the voltage, current or frequency. These range in power from few milliwatts (mW) in mobile phones to hundreds of megawatts (MW) in high-voltage direct current (HVDC) transmission systems. Usually, electronics are thought of in the framework of information technology, where speed is the primary interest. However, in the context of power electronics (Figure 2), there is a critical need for improved efficiency and reduced power losses.

Figure 2: Range of power electronics applications [1].

There is a fundamental difference between planar microelectronics devices and power electronics devices. While microelectronics is essentially a surface-based technology (i.e., the active area is on surface of the chip), in power electronics the current passes through the chip. This means that electrical contacts are on both sides and, because of the high voltages, insulation is necessary. For higher power, the chips are often stacked or connected in parallel in a multichip module. As 3-dimensional (3D) chip solutions get incorporated into volume production, the need for bulk thermal management will drive the development of cooling techniques such as liquid channels and graphene[1] layers.

Designers call on materials in power electronics systems to provide electrical and thermal conduction, insulation, protection, and mechanical stability, all with the objective of achieving the desired reliability. These properties usually cannot be considered in isolation. The coefficients of thermal expansion (CTEs) of the semiconductors and insulators are fixed. Metals can be matched by adding fillers such as aluminum-silicon-carbide (AlSiC), a mixture of aluminum, silicon and carbon/graphite, where silicon-carbide (SiC) has the low CTE to interface with silicon. Clearly, it is important to use materials with high thermal conductivity and closely matched CTEs; applications using graphite must consider the fact that graphite is highly anisotropic with high thermal conductivity in the x-y plane only.

This metal-matrix composite (MMC) material is used in high-reliability traction applications because of the CTE match between the DCB/AMB (direct bond to copper-active metal braze) substrate to the AlSiC base plate, which protects the large area solder joint between them. Because of its lower weight, AlSiC may find use in aircraft applications.

2.2 LED Thermal Management Challenges

One particular challenge of light emitting diode (LED) thermal management is that, as with many semiconductor devices, the junction temperature of LED chips must be typically maintained well below 125°C — well below operating temperatures of traditional incandescent light sources.

The LED bulb is significantly more efficient in converting electricity to visible light, and the spectrum of that light is more effectively tailored to the photopic response of the human eye. As a result, the 60W equivalent LED bulb consumes only 13.5W of input power to deliver 2.5W of visible light power equivalent to 800 lm. This dramatic, 80 percent reduction in heat dissipation relative to the incandescent, 11W versus 55W, would seem to make thermal management of LED bulbs trivial. However, with a maximum allowable junction temperature of only 125°C, there is a much smaller temperature difference available to drive heat transfer from the source to the ambient. As a result, thermal management of the LED bulb is more challenging—requiring one-fourth to one-fifth the thermal resistance (°C/W). Making matters worse, natural convection and radiation heat transfer processes are significantly less effective at temperatures closer to ambient.

The vast majority of the waste heat in an LED system is generated in a very small volume within the millimeter-scale LED chip(s). While LED efficiency continues to improve, device manufacturers are packaging devices in increasingly smaller footprints, only exacerbating the heat density challenge.

2.3 Direct Immersion Cooling

It is possible that, in some cases, even with improved thermal interface materials, the internal temperature rise from the case-to-chip-junction may be too large because of the projected increase in power. In such instances it may be necessary to resort to direct immersion cooling with a dielectric liquid contacting the chip. Such cooling schemes could take the form of single-phase liquid-impingement jet cooling, pool boiling, or two-phase liquid spray cooling. Spray cooling of electronics within an enclosure has been implemented in military systems and in supercomputer modules. Whatever form the application of direct liquid immersion cooling may take, the major requirement will be that it is done at a reasonable cost, is reliable and occupies the minimum possible packaging volume.

2.4 Refrigeration Cooling

Both large servers and workstations have employed vapor compression cycle refrigeration to lower temperatures of the processor. Current technologies exhibit improvements of approximately two percent for every 10°C reduction in chip temperature. With this technology, the evaporator is mounted directly on the processor module. The remaining hardware (i.e., compressor, condenser, valves, etc.) is typically packaged in a separate enclosure attached to the bottom of the system (workstation) or mounted inside the rack (servers). This technology has achieved chip temperatures in the range of -20 to 40°C.. As with water cooling, the major requirement is to develop a refrigeration cooling technology that is low-cost, reliable, and occupies a minimum volume within the system.

2.5 Thermoelectric Cooling

Thermoelectric coolers (TECs) offer the potential to enhance the cooling of electronic module packages to reduce chip junction temperatures or accommodate higher power. They also offer the advantages of being compact, quiet, and moving-parts-free—and they can provide an active control of temperature. TECs are limited in the magnitude of the heat flux that can be accommodated. TECs also exhibit a lower coefficient of performance (COP) than conventional refrigeration systems. The COP of a TEC will vary depending upon the usage conditions, but will typically be less than unity. This means that the electrical power consumed by the TEC will be as great as, and often more than, the power dissipated by the component being cooled. These limitations are due to the currently available materials and methods of fabrication. As a result, thermoelectric devices have been restricted to applications characterized by relatively low heat flux.

Efforts are underway to improve the performance of TECs by the development of new thermoelectric materials and thin film coolers. If successful, these efforts promise increased heat pumping capability and higher COPs, which could open the door to a much broader application of thermoelectric devices to augment electronic cooling.

2.6 Thermal Materials

Heat removal, thermal stresses, warpage, weight, and cost are critical packaging issues. Traditional thermal management / packaging materials all have serious deficiencies. In general, traditional materials with high thermal conductivities have high CTEs, and materials with low CTEs have high densities and thermal conductivities that are similar or modestly better than that of aluminum. Chemical vapor deposition (CVD) diamond is a key exception.

In response to this problem, an increasing number of advanced materials have been investigated, and some are available for use. These materials offer: thermal conductivities up to more than four times that of copper, CTEs that can be tailored from –2 to +60 ppm/K, electrical resistivities ranging from very low to very high, extremely high strengths and stiffness, low densities, low cost and net-shape fabrication processes.

The payoffs are: improved performance or simplified thermal design, reduced power consumption, and reduced thermal stresses and warpage. Use of CTE matched materials allows direct solder attachment with hard solders (hard solders have better fatigue resistance than soft solders and fewer metallurgical problems), increased reliability, improved performance, weight savings up to 90%, size reductions up to 65%, reduced electromagnetic emissions, increased manufacturing yield and potential cost reductions. In addition, advanced materials make it possible to have low CTE, thermally conductive PCBs that can greatly increase the range of conductive cooling (in the aerospace industry, heat is removed entirely by use of PCB cold plates) and convective cooling.

A number of advanced materials are now being used in commercial and aerospace applications including: servers, plasma displays, notebook computers, printed circuit boards (PCBs), PCB cold plates, radio-frequency (RF) modules, power modules and optoelectronic packages. The rate of growth in the use of these materials has been dramatic. Components include carriers, heat spreaders, heat sinks, thermoelectric cooler substrates, LED packages and laser diode packages.

A significant need exists for new fluids that can be used for indirect liquid cooling (replacing water). That fluid would need to provide safer, more reliable operation in hostile environments, and also lower-cost direct (immersion) liquid cooling for the broad range of potential applications in commercial and military systems.

The critical issue of cost is complex and must be considered in the context of total cost over system life. Efficacy and appropriateness are prime factors, but beyond that, other issues needing to be evaluated include: reliability, life, mean time between failures (MTBF) and maintenance-repair-operating costs. Many factors also play a role in cost, such as complexity, size, flatness, surface finish requirements, and production run size.

Cost effectiveness depends on a particular application. For example, higher price and higher performance systems such as high-end servers and aerospace systems can tolerate higher thermal component costs than mobile phones. A key issue is the cost of competing approaches. For example, if the alternative is liquid cooling, an advanced material that is more expensive than a traditional one may be cost effective if it allows the use of convective air-cooling.

2.7 Thermal Design Modeling and Tools

Sophisticated thermal design tools are now an essential element in the day-to-day design of electronic components, packages and systems. These tools take a variety of forms. Thermal conduction codes are used to model heat flow and temperatures within a package. Computational fluid dynamics (CFD) codes are used to model fluid flow around and through package assemblies—along with the associated pressure drop and heat transfer from exposed package surfaces to the fluid stream. In addition, some CFD and thermal conduction codes have conjugate capability, making it possible to model thermal conduction within the package structure simultaneously with modeling fluid flow and heat transfer in the cooling fluid.

Over the past decade, much has been done to improve the graphical user interface for problem definition and data input, especially with CFD codes tailored for use to model electronic equipment. Nonetheless, the industry needs further improvements to reduce the time consumed in defining the package geometry and structure and to enter related data preparatory to running a model. Seamless integration of computer-aided design (CAD) solid modeling tools, electronic design automation (EDA) tools, and CFD tools is needed to provide thermal designers the ability to take CAD solid-modeling-generated data and EDA-generated data and move them effortlessly into finite element thermal conduction modeling tools or CFD modeling tools.

Other needs requiring further effort:

• An improved ability to optimize thermal analysis codes for parallel processing to reduce solution time and provide the capability to model more complex thermal problems;
• A better way to enable CFD codes to better model turbulence and convective heat transfer in the transition flow regimes; and
• More extensive benchmarking to validate the accuracy of CFD codes.

3.0 Summary of Technology Gaps & Show-Stoppers

The thermal technology improvements needed for each product sector to fill gaps and avoid show-stoppers are summarized in Table 1.

Table 1: Thermal improvements needed by product sector [1].

Product Sector	Requirements
Common Needs	·         Improved thermal interfaces. ·         Improved thermal spreaders. ·         High-performance air cooling solutions. ·         Advanced modeling tools. ·         3D designs cooling: ability to insert heat spreaders and phase change layers.
Portable/Wireless	·         No significant improvements needed as long as battery power remains constrained.
High-End Systems	·         Thermal integration with electromagnetic compatibility (EMC) shielding. ·         Low-cost, compact and reliable water cooling. ·         Low-cost, compact, reliable and efficient refrigeration. ·         High heat flux, efficient thermoelectric cooler. ·         Mechanically robust packages that minimize the thermal resistance path to air. ·         Low-cost, compact, and reliable dielectric liquid cooling. ·         Abatement of heat load impact on installation. ·         Outdoor structure (sheds) for remote stations: use materials that can passively reduce the need for active heating, ventilating, and air conditioning (HVAC)/fan load. ·         Quieter fans with efficient airflow design.
Automotive	·         Low-cost, reliable heat pipe technology for automotive environment. ·         Passive electrical components/system level packaging materials capable of operating at 150oC. ·         Low-cost liquid or refrigerant cooling systems utilizing automotive cooling components. ·         Low-cost, self-contained, phase change materials to handle transient thermal events. ·         Analog and digital ICs capable of operating with TJ = 175oC. ·         Power transistor capable of operating with TJ = 200oC.
Medical	·         Low-cost, compact and reliable water cooling. ·         High heat flux, efficient thermoelectric cooler. ·         Low-cost, compact, and reliable dielectric liquid cooling.
LEDs	·         Development of LED packaging with low thermal resistance. ·         Low-cost, compact, and reliable dielectric liquid cooling. ·         SSI LED products have color shifts and lower lifetime performance — develop thermal management technologies to dissipate heat associated with high brightness light sources. ·         SSI-specific software and modeling tools to optimize assembly of LED and organic light-emitting diode (OLED) SSI devices are limited — develop SSI-specific software for designing and fabricating LED light engines and light sources within environmental and thermal constraints.
Power Electronics	·         Geometries of electrolytic capacitors and magnetic components should be optimized to transfer and exchange heat better. ·         To best take advantage of the SiC junction field effect transistors (JFETs) (capability for higher voltages and higher operating temperatures in contrast to silicon devices) packages must be improved: materials, inductive wiring, creepage distances, etc. ·         Overall lower thermal resistance within power electronics system, especially in solders, glues, etc. — silver sintering as an attachment medium is in early stages of commercial use, and may offer other benefits. ·         Heat spreading: reliability of joining of heat pipes needs improvement; MMCs (such as aluminium/carbon) should be explored. ·         For optimization of heat exchange to ambient developments in two-phase cooling, pumpless liquid loops.

RECOMMENDATIONS

The following constitutes the major cooling technology areas identified for development and innovation by the Thermal Management Roadmap:

• Low-cost, higher-thermal conductivity packaging materials, such as adhesives, thermal pastes and thermal spreaders, for use in products ranging from high-performance computers to automotive applications.
• Advanced cooling technology, such as high-performance heat pipe / vapor chamber cooling technology, thermoelectric cooling technology, direct liquid cooling technology, as well as high-performance air-cooling and air-moving technologies.
• Closed loop, liquid-cooling solutions, which are compact, cost-effective and reliable.
• High-performance cooling systems that will minimize the impact on the environment within the customer’s room and beyond.
• Advanced modeling tools that integrate the electrical, thermal, and mechanical aspects of package / product function, while providing enhanced usability and minimizing interface incompatibilities.
• Advanced 3D packaging techniques that can effectively remove heat from die not directly in contact with the PCB / substrate.
• Advanced experimental tools for flow, temperature and thermo-mechanical measurements for obtaining local and in-situ measurements in micro-cooling systems.

It is further recommended that industry participants should pool resources to fund cooling technology development, promote the involvement of university / research labs and establish a closer working relationship with vendors.

Finally, the industry needs to consider and evaluate changes in design processes to optimize system performance by (i) eliminating margin redundancies so costs may be minimized, and (ii) modified partitioning of component / system building blocks.

Reference

[1] iNEMI, 2015 Roadmap Thermal Management Chapter, http://www.inemi.org/store_product.asp?prodid=55

About the Author

Azmat Malik is President of Acuventures and Chair of iNEMI’s Thermal Management Technology Working Group (TWG) for the 2015 Roadmap. He will also chair the TWG chapter for the 2017 Roadmap. Work is scheduled to kick off February 10 on the 2017 edition. Participation is open to the industry, and anyone interested in getting involved with the next Thermal Management Roadmap should contact Azmat ([e-mail]) or iNEMI ([e-mail]).

CONTACT DETAILS  

Name: Azmat Malik

Affiliation: Acuventures

Email:  [e-mail]

[1] Graphene has been shown to have unusually high thermal conductivity.  Multiple layers of graphene show strong heat conducting properties that can be harnessed in removing dissipated heat from multilayer electronic devices.  This has led to dramatic improvement in thermal characteristics, leading to lower temperatures even at higher processing speeds and higher power dissipation.