Author Archives: Pallavi Madakasira

In the Power Electronics Industry, Infineon Leads the Pack on Innovation and Execution


Key players in the power electronics industry, including Infineon, ABB, Cree, and Rohm, are all looking to capture market share in individual end markets through technology innovation and differentiated market strategies. Some of this critical innovation includes pushing the limits of silicon’s performance to investing in new material flavors, such as gallium nitride (GaN) and silicon carbide (SiC). As for technologies, strategies also vary in terms of internal R&D versus partnerships, value returned to R&D and M&A activities, and geographic and application breadth. The question is, who comes out on top and which companies are struggling to make it in the power electronics industry? We benchmarked 12 leading power electronics companies, contrasting them on their innovation performance and their business execution. For the former, analysis of R&D spending and the last five years of patent portfolios, innovation-impacting acquisitions, divestments and partnerships, and overall new product portfolio was analyzed, weighted and aggregated. For business execution, factors included profitability, liquidity, solvency and acquisitions, whereas partnerships focused on market access.

Infineon is a clear winner on both metrics, having consistently maintained its market share (on a revenue basis) despite other companies clamoring to compete with it across all markets. In terms of technology innovation the company has invested and successfully released products not only in silicon – 1200 V and 1700 V IGBT modules, 1700 V integrated power modules, 600 V CoolMOS, and 650 V trenchtop IGBTs for automotive applications – but also in new flavors of materials, including SiC and GaN. NXP, Microsemi, and Rohm follow behind Infineon. NXP, for example, scored lower on its patents and product releases (it has released only two new power electronics products in the last few years), bringing its overall innovation score down. Without any major acquisition or a breakthrough product, none of these companies is likely to offer Infineon tough competition in the near future.

At the opposite end of the spectrum are the likes of Renesas, ON Semiconductor and Panasonic. Renesas, for example, scored poorly on both business and technology-related acquisitions and partnerships as well as new product releases. While Renesas has a core strength in microcontrollers for the automotive industry, it lacks strength in any other product segment. ON Semiconductor has some ongoing work in GaN with Transphorm to develop power supplies, although it has no concrete strategy to execute on this. Panasonic falls to the bottom on this grid given its weak scores across the board. The company acquired Sanyo in 2010, but that did not give Panasonic the strength it needed in power electronics. The company has a very narrow portfolio of products, which it has failed to expand on or offer any real technology differentiation. Fairchild, too, has barely managed to stay alive; without drastic measures, it could be wiped out entirely.

The laggards should look to diversify away from power electronics in to the Internet of Things (IoT) market. These companies will need to invest in sensors-related technologies – whether that has to do with developing sensors or powering them – to survive. ON Semiconductor has made a big move in this space by acquiring Aptina – a CMOS image sensor company. Fairchild released its first MEMS product in July 2015, which appears to be the silver lining in the company’s story. However, if it does not act fast, it will be wiped out of the industry. In contrast, Infineon will need to watch out for competition from high-profile GaN startups to stay ahead of the innovation game in GaN. The company may ultimately need to acquire a startup like GaN Systems or VisIC that gives it the competitive edge in GaN.

Innovation and business strategies are available for any existing or aspiring market participant to shore up their respective activities or acquire pieces that go on the offense around the vulnerabilities of others. Knowing those strengths and vulnerabilities in the markets that will grow is the key.

Cree Demonstrates 50 kW String Inverter as It Looks to Further Its Interests in Downstream Modules

What They Said

Cree recently demonstrated a 99.1% efficient 50 kW solar string inverter. The company used its in-house silicon carbide (SiC)-based metal oxide semiconductor field effect transistors (MOSFETs) and diodes in the solar inverter. It also claims the resulting inverter is one-fifth the average weight and size of an incumbent, similarly rated silicon inverter, and costs about 15% less than a comparable silicon inverter.

What We Think

Cree’s 50 kW string inverter demonstration further reinstates the company’s attempt to move downstream into power modules, away from being just a pure play discrete device manufacturer. The efficiency of Cree’s 50 kW inverter is 1% more than the highest demonstrated efficiency of a comparable silicon inverter although the result is not surprising given Cree is using all SiC (that is inherently much more efficient than silicon) components within an inverter. Since SiC MOSFETs and diodes have less switching losses, Cree can get away with having far fewer discrete components (both active and passive) and smaller heat sinks, resulting in a substantially smaller and lighter inverter overall. The company’s lower cost claims appear to be largely based on an improved inverter efficiency impacting the $/W costs instead of explicit costs of Cree’s inverter (which is quite likely very high). SiC MOSFETs and diodes are substantially (more than 10x) more expensive than silicon MOSFETs and diodes – offset to a more limited extent by the savings in the bill of materials related to fewer passive components and smaller heat sink. Unfortunately, despite the efficiency improvement, a price sensitive solar inverter market is unlikely to adopt Cree’s SiC inverter until the explicit costs of the inverter come down and until the reliability of these inverters are proven at scale.

GaN-on-Si Will Dominate the GaN Power Electronics Market for the Next Decade, Reaching $1 Billion by 2024


As Si struggles to keep up with increasing demands on performance in several power electronics applications, emerging materials like gallium nitride (GaN) and silicon carbide (SiC) look to displace silicon. While silicon and SiC (SiC-on-SiC ) come in only one flavor, GaN comes in many different flavors, including GaN-on-Si, GaN-on-SiC, and GaN-on-GaN. Each variety of GaN has advantages and disadvantages while also being better suited to different power electronics applications. For example, while GaN-on-Si offers price benefits over the other GaN types, GaN-on-SiC can offer benefits of efficient high-temperature operation.

The total market for GaN power electronics overall grows at 32% CAGR, reaching $1.1 billion by 2024, or more than 5% of the total market share. Within this, the GaN-on-silicon (Si) market will grow at over 30% CAGR to reach nearly $1.0 billion by 2024, representing 90% of the total GaN market. This is largely due to the sheer number of companies developing GaN-on-Si solutions for power electronics markets. GaN-on-Si in transportation and renewables and grid markets will reach about $380 million and $350 million respectively in 2024, proving to be the runaway leaders for adoption of GaN-on-Si. GaN-on-SiC and GaN-on-GaN will have a limited play in power electronics until SiC and GaN substrates become substantially cheaper. That said, GaN-on-Si carbide (SiC) will be best adopted in the transportation segment, reaching nearly $100 million in 2024. SiC substrates’ ability to function efficiently at high temperatures will be a key driver for GaN-on-SiC adoption. In contrast, GaN-on-GaN will experience lukewarm adoption across all applications.

The success of GaN-on-Si will attract investments not only from silicon incumbent device manufacturers but also from foundries like Taiwan Semiconductor Manufacturing Company (TSMC) that are already invested in GaN-on-Si for light-emitting diodes (LEDs). GaN-on-Si leverages silicon infrastructure, and with foundries becoming increasingly cash-rich, expect to see select foundries move upstream through direct partnerships, investments, or acquisition of device manufacturers in GaN-on-Si longer term.

Source: Lux Research report “Breaking Down the Gallium Nitride Power Electronics Market” — client registration required.

Mixed Signal Driver IC Developer xSi’s Technology is Disruptive; a Strong IP will be Key to its Success

We recently spoke to Rajesh Swaminathan, the founder and CEO of xSi Semiconductors, a developer of mixed signal integrated circuits (ICs) for light-emitting diode (LED) applications. He explained that the main difference between xSi’s mixed-signal and competing mixed-signal solutions was the programmability of the firmware (programs and/or data structures that internally control various electronic devices) within the IC. He said that most mixed-signal ICs have a combination of analog and digital IC functionality; for example, when the driver functionality needs to be expanded to include more LED strings in a traditional mixed-signal IC, the manufacturer would have to use silicon components to build out a new IC entirely. In contrast, xSi’s IC can be “tuned” by only using firmware to expand functionality, lowering the total cost of ownership. The company also claims it can offer this product at analog IC prices, which are typically at 50% the cost of a
digital IC.

Driver Innovation is important to lower costs, as the potential cost savings is squeezed from the LED dies. In addition to cost, the LED industry looks beyond the die at thermal management, optics, and drivers to not only reduce system cost but also optimize system functionality. The total LED driver market for LED luminaires in commercial, industrial, and street lighting application is set to reach $4.3 billion in 2023 led primarily by the aggressive rate of adoption of the recessed modular luminaires for the commercial lighting market (see the report “Powering Light: Sizing the Commercial, Industrial, and Street LED Driver Market” — client registration required). Digital ICs are prevalent in the LED industry but can be very expensive, and their accuracy is not as good as the analog ICs that are also the cheapest on the market. xSi’s claims about being able to offer mixed-signal ICs at the price of analog ICs is definitely very promising; the company plans to release its first product in 2015.

From a customer perspective, the rationale for using these application specific ICs (ASICs) is clear. They simplify the design and circuitry challenge, embed necessary firmware, reduce the footprint for printed circuit boards, while reducing the items on the bill of materials (BOM) and the total BOM. Given the time-to-market and cost pressures especially in the LED industry, these components usually make a lot of sense. However one big challenge with ASICs is that they become too specific for an
application which forces the customer to become dependent on one single source or vendor for these ICs. LED industry is especially very conservative and forcing a single sourcing behavior can backfire; xSi hence needs to navigate cautiously. Just like in solar, vendors/suppliers with strong balance sheet, portfolio of products and strong IP are desired in the LED industry. xSi currently owns only two patents, and as such is not very well positioned; it will be critical for the company to secure its technology differentiation through a strong IP portfolio.

Given that the development cycle for driver ICs is about three years, xSi probably has a three-year lead over its competitors; not long enough to keep competition from incumbents such as Texas Instruments and Maxim Integrated at bay. For xSi which is hoping to leverage its IP and design expertise there’s a tricky path to navigate. That being said, if the company can successfully demonstrate its value, it could become an attractive acquisition target medium-term. With slim pickings in the mixed-signal driver IC market, clients looking for an entry point into the LED industry should engage for investment opportunity.

The Market for LED Drivers to Hit $4.3 Billion in 2023, From Only $490 Million Today


Light-emitting diodes (LEDs) offer superior operating lifetime, efficiency, and color quality, and will replace incumbent technologies like fluorescent and high-pressure sodium in the global lighting market. The tremendous growth in the LED luminaire market will clearly flow down to the suppliers of key components, but to different degrees in different market segments. Indeed, LED performance and durability, two of the key drivers for adoption, will not evolve without efficient driver power electronics within the LED luminaire, a need that will also vary greatly by market segment. The choice in LED drivers needs to account for the electrical characteristics of the LED system as well as the optical and the thermal characteristics that are needed to ensure the total LED system operates efficiently. While resistor- and transistor-based LED drivers are typically suited to low-power LED applications (below 20 W), linear and switched-mode LED drivers are used for medium and higher power LED applications respectively.

The market for LED drivers in commercial, industrial and street lighting applications is best broken into recessed modular, high-bay and street LED luminaires within each application, respectively, given the growth expected in these applications for LEDs overall. In aggregate, the total LED driver market will grow from about $490 million in 2014 to reach $4.3 billion in 2023, led primarily by the aggressive rate of adoption of the recessed modular luminaires for the commercial lighting market. Drivers for commercial lighting will account for $3.7 billion in 2023. While the growth in this LED luminaire segment overall is a major component of the growth in driver revenue, LED drivers in a commercial recessed modular luminaire will also grow as a percentage of the bill-of-materials, accounting for nearly 30% of the cost stack by 2023, up from 21% in 2013. In contrast, slow adoption of LED luminaires in high-bay industrial and street segments will hold driver adoption to $300 million and $320 million, respectively, in 2023.

As the market grows, advancements in driver technology will determine the winners and losers. Driver designs will become more sophisticated even as research in magnetics picks up momentum, enabling significantly more power density with dramatically smaller footprints. This trend will benefit recessed luminaire applications, in particular where there are extreme space constraints. Fabrication innovations are also in sight, with leading-edge, high-voltage clean room processes and 8-inch (or larger) wafers set to play key roles in lowering the cost of drivers and lighting. Emergence of new wide-bandgap materials like gallium nitride (GaN) that have a high switching frequency could also dramatically improve efficiency of LED drivers across all segments.

For the prospective players in the space, LED driver design will still serve the incumbents well – with few and limited opportunities for start-ups. Large companies, including Texas Instruments, Infineon, and Maxim, will continue to dominate much of the LED driver landscape. That said, Asian manufacturers’ participation in this market will gain steam, as in other LED markets, threatening margins in a price-to-performance battle. This trend has already been set in motion; Chinese analog IC manufacturer Silergy Corp. has already received LED lighting driver orders from Philips, NVC Lighting Technology, and Everlight. Champion Microelectronic Corp. launched an LED driver product in 2013 targeting the industrial and commercial lighting sectors. The path for new entrants? Smaller companies that can demonstrate a technology edge such as xSi Semiconductors will become fodder for acquisition, or lose out entirely if they play hard to get.

The growth in LED drivers overall is hard to ignore, but companies planning to benefit from it will need a strategy that captures technology, application and geographic forces in the market.

Source: Lux Research report “Powering Light: Sizing the Commercial, Industrial, and Street LED Driver Market” — client registration required.

Ammono-Unipress Hybrid GaN Process Looks Very Promising; Commercial Viability Needs to be Proven

What They Said

Ammono and the Institute of High Pressure Physics of the Polish Academy of Sciences (Unipress) recently announced that they have developed a hybrid ammonothermal-hydride vapor phase epitaxy (HVPE) process to produce low-dislocation-density gallium nitride (GaN) substrates. The traditional HVPE process is a two-chamber process using sapphire as the seed crystal while the traditional ammonothermal process is a single-chamber process using GaN substrates as the seed crystal; the details about this hybrid process were not disclosed. The partners claim that they have achieved smooth GaN layers up to 2.5 mm thick (crystallized with a stable growth rate of about 240 μm/hour) without any cracks, and at a dislocation density of 5 x 104/cm2 (which is about three orders of magnitude better than a typical HVPE GaN substrate dislocation density). They also used GaN wafers from this hybrid process as the seed crystals for Ammono’s ammonothermal process; the company claims that using a hybrid ammonothermal-HVPE seed crystal in a subsequent ammonothermal process still resulted in very low dislocation density of about 2 x 104/cm2.

What We Think

Ammono-Unipress demonstration of a hybrid GaN substrate process looks very promising especially in terms of rate of growth and dislocation density although its commercial viability remains unproven. The crystallization growth rate of 240 μm/hour is about 20x faster than that of the traditional ammonothermal process and can result in a dramatically more cost effective process – about 40% cost reduction overall. The dislocation density achieved in the hybrid GaN wafer is only slightly worse than the 2 x 104/cm2 that can be achieved using the ammonothermal process further validating the promise of such a hybrid approach. Ammono (client registration required) is also currently participating in one US Department of Energy (DOE) Advanced Research Projects Agency-Energy (ARPA-E) project under the Strategies for Wide-Bandgap, Inexpensive Transistors for Controlling High-Efficiency Systems (SWITCHES) program with Kyma to develop a similar hybrid manufacturing approach.

GaN substrates with such low dislocation density are especially attractive in not only lasers but also high-voltage power electronics applications that typically mandate a vertical device structure (see the report “Price or Performance: Bulk GaN Vies with Silicon for Value in LEDs, Power Electronics and Laser Diodes” — client registration required). Until now, the prohibitively high costs of GaN substrates have prevented their use in power electronic devices in power electronics – should this process become commercially viable, that scenario could change quickly; clients should continue to scope out for investment opportunities in GaN substrate players such as Ammono and Fairfield Crystal (client registration required.)

Carbodeon’s Diamond Nanoparticle-Infused Polymer Holds Promise

What They Said

Carbodeon recently announced a thermally conductive polymer with improved performance by using tiny particles of diamond as the thermal filler. It started with a polyamide 66 (PA66) reference material containing 45% by weight boron nitride as the thermal filler. It then created a new material using 44.9% boron nitride and 0.1% diamond nanopowder, and found the thermal conductivity of the PA66 increased by 25%, averaged across all planes. The company also claims that it has modified surface chemistry so that the particles are driven to disperse and to become consistently integrated throughout the polymer. The company is specifically looking to target electronics and light-emitting diode (LED) applications.

What We Think

Carbodeon’s announcement is notable in that it claims uniform distribution of diamond nanoparticles that is used as a filler material in the polymer. Whether this polymer composite is ultimately cost effective and easy to manufacture remains to be seen. Most other companies developing conductive polymers including Ovation Polymers, Bergquist, and Cool Polymers use multiwalled nanotubes (MWNTs), graphene nanoplatelets (GNPs), metal or ceramic nanoparticles as fillers; with each filler material enabling different levels of thermal conductivity. Start ups like Ovation Polymers have struggled to improve the thermal conductivity beyond 20 W/mK and uniformly across all planes; mainly due to agglomeration of the MWNTs that it uses a filler material. In contrast Bergquist has successfully commercialized polymers with metal and ceramic nano particles as fillers and become an established name in the thermal management industry. Carbodeon’s use of diamond nano particles presumably enables superior thermal conductivity, although it still has a long way to go to effectively compete with the likes of Bergquist and Cool Polymers. The total conductive polymer addressable market is set to grow from $280 million in 2013 to $1.9 billion in 2020, a CAGR of over 31% (see the report
Cooling Heats Up: Sizing the Opportunity for Conductive Polymers in Thermal Management” — client registration required). To tap into this growth, clients should consider strategic acquisition of or investment in companies like Cool Polymers, Carbodeon, or Bergquist.

The Polymer-Based Thermal Materials Market Reaches $1.9 Billion in 2020, Led by LED Lighting


Significant improvements in luminous efficacies in light-emitting diodes (LEDs) and current densities in power electronics necessitate efficient and cost-effective thermal management solutions. Where incumbent secondary heat sinks like aluminum and thermal interface materials like thermal grease, pads, and fillers have been traditionally used, conductive polymers look to substitute these traditional materials in different applications. The question is, how much will these emerging materials solutions displace the incumbents over the coming decade?

The overall thermal management materials market will reach $4.8 billion in 2020, with LED lighting being the primary driver. The LED lighting thermal management market for secondary heat sinks and Thermal Interface Materials (TIMs) combined will grow from $1.5 billion in 2013 to $3.8 billion in 2020. The smart phone and tablet thermal materials market will triple from about $100 million in 2013 to more than $300 million in 2020, while the thermal materials market for solar micro inverters as well as the HEV/EV power module market will grow from about $28 million in 2013 to over $170 million in 2020, a CAGR of 29%. Importantly for polymer materials developers, conductive polymer developers are set for a major move over this same time period. The total conductive polymer addressable market will grow from $280 million in 2013 to $1.9 billion in 2020, a CAGR of over 31%. Polymer secondary heat sinks in general lighting will account for a majority of this market share, reaching $1.1 billion in 2020, due mostly to replacement of bulky aluminum secondary heat sinks in the residential market near term and commercial market medium to longer term. This growth is mainly driven by the sheer volumes in the general LED lighting industry. The auto LED secondary heat sink market will account for only about $92 million in 2020. The volumes in the auto LED market are substantially lower compared to the general LED lighting industry, although will allow for potentially higher priced products due to specificity. The total polymer TIM market will grow from $280 million in 2013 to $785 million in 2020, a CAGR of 16% largely due to adoption in smart phones and tablets as well as in general LED lighting as TIM.

The consumer electronics segment offers additional value for companies looking for opportunities for materials-enabled innovation. Rather than be restricted by size (area) and shape, as is the case of TIMs for LEDs, microinverters and HEV/EV power modules, polymers for electronics such as tablets and smart phones will migrate to encompass the entire body of the smart phone or tablet – making heat dissipation more effective and in the process lowering the bar for polymer thermal conductivity. This will push the boundaries of innovation in conductive polymers where aesthetics such as color and texture will become important metrics for adoption.

As with many materials-to-device integrations, partnerships will be critical to success. Already, alliances such as Philips-DSM not only reinstated Philips as a leader in innovation in LEDs but DSM’s polycarbonate solution became commercially validated as a dependable thermal management solution for MR16 bulbs made by Philips. Across LED lighting, automotive lighting, consumer electronics and larger format power electronics, device manufacturers and materials developers can mutually enable the other’s success.

Source: Lux Research report “Cooling Heats Up: Sizing the Opportunity for Conductive Polymers in Thermal Management” — client registration required.

GaN and SiC-Based Power Electronics Set to Deliver Big Value in Distributed Solar Installations

Inverters’ importance in the solar market has only been emphasized by the oversupply and price pressure that has driven down the cost of components around it. Suppliers are doing their part to reduce costs as well, with incremental improvements in efficiency and component count reduction, but the holy grail for solar inverters is the implementation of wide bandgap semiconductors – specifically, silicon carbide (SiC) and gallium nitride (GaN). They offer the promise of higher efficiencies, as well as superior thermal management – critical for temperature-sensitive applications such as solar inverters. GaN and SiC offer indirect cost savings in addition to direct performance benefits – superior thermal conductivity of SiC over Si reduces the size of the heat sink in inverters. Higher switching frequencies of SiC and GaN reduce the failure probability and count of passive components, while high power density enables footprint reduction and installation cost savings. The question is, what is the opportunity for introducing diodes and transistors using these higher cost, but higher performance materials?

Microinverters offer the best absolute $/W premium for SiC or GaN  diodes with Si transistors (SiC + Si, and GaN + Si, respectively) and represent the ideal niche entry for these devices in the residential segment. However, string inverters are the most attractive segment for price premiums relative to silicon with the introduction of SiC and GaN transistors in addition to diodes. Acceptable string inverter price premiums of all-GaN and all-SiC systems versus all silicon top $0.10/Wp in the residential market segment enabling price premium of greater than 20% relative to silicon-based inverters. Importantly, string inverters enable ready access to the growing commercial and residential segments, delivering both volume and price in the two segments set to dominate new solar installations in the developing world for the coming years.

Notably, SiC diodes are already hitting the market through microinverters. As GaN diodes and SiC and GaN transistors become more commercially available, they should take the same path – and will have a similarly beneficial impact, while enabling discrete device developers to penetrate the large-scale inverter market at a healthy 10% price premium. As devices fully featuring GaN and SiC hit the market, they’ll hold the biggest competitive advantage in small systems – microinverters and small string inverters, for residential and commercial solar installations – with a powerful proposition: lowering the levelized cost of energy (LCOE) and increasing margins on electricity sold through PPAs.

The race is on to position for technology-driven differentiation in these growing markets. Little surprise to see inverter mainstay Advanced Energy acquire REFUsol on this basis given the latter’s valuable SiC-based inverter IP and products. Though the payback will take some time, the $77 million Advanced Energy paid for that IP will look like a bargain down the road. Others would be wise to take note of this and ABB’s similarly SiC-related acquisition of PowerOne and act accordingly.

Source: Lux Research report “Reaching for the High Fruit: Finding Room for SiC and GaN in the Solar Inverter Market” — client registration required.

Thermal Management is a Critical Need and Innovation Opportunity to Drive LEDs Forward

Cost is the name of the game with LEDs, but most of the time, the focus is entirely on the package. Significant opportunities for cost reduction lie in materials and technology innovation in the balance of system, including thermal management, drivers, and optics. In this respect, today’s technology solutions fall short of the dramatic cost reductions needed to mirror the LED package and alternate solutions are ineffective and uneconomical – presenting opportunities for technology innovation. Based on an LED bulb equivalent to a 60 W incandescent, with a SMD configuration, aluminum based thermal management, non dimmable drivers and standard lenses for secondary optics, thermal management accounts for about 27% of the bulb cost in 2011, or $6.00. While this figure will fall to $3.95 in 2020, that figure will amount to a larger share of the bulb cost, at 36%. The size of the heat sink and the choice of material largely determine the cost – aluminum is the incumbent heat sink material and the cheapest option on the market today. Switching to more thermally conductive materials such as copper can improve performance of the heat sink, but they are currently about two to four times more expensive, and can thus increase bulb costs by 50% or more.

Material replacements with better conductors like copper are unlikely to result in cost savings in the next 10 years, and while active thermal management is a promising approach to cost savings in LEDs, its impact is unlikely to be felt outside of niche, newly enabled, applications. Further opportunities to improve thermal management will be critical for ongoing future LED cost reductions. The share of the cost stack will only rise and serve to cap device capabilities unless the opportunity is addressed.

Source: Lux Research report “Cheaper, Brighter, Cooler: The Need for Cost Reduction Past the Package” — client registration required.