Category Archives: Device Architecture

Cautious optimism


June 15, 2018

By Walt Custer

Updated global GDP forecast

The World Bank just updated its multiyear forecast for GDP growth both globally and by country (Chart 1).

It noted: “Despite recent softening, global economic growth will remain robust at 3.1 percent in 2018 before slowing gradually over the next two years, as advanced-economy growth decelerates and the recovery in major commodity-exporting emerging market and developing economies levels off.

“This outlook is subject to considerable downside risks. The possibility of disorderly financial market volatility has increased, and the vulnerability of some emerging market and developing economies to such disruption has risen. Trade protectionist sentiment has also mounted, while policy uncertainty and geopolitical risks remain elevated.”

Chart 1

Semiconductor growth outlook strong (Chart 2)

The WSTS updated its world semiconductor shipment forecast. This new forecast (endorsed by SIA) projects worldwide semiconductor sales will be a record $463 billion in 2018, a 12.4 percent increase from 2017. WSTS projects year-to-year increases across all regional markets for 2018.

Chart 2

This revised semiconductor forecast coupled with very robust global semiconductor capital equipment sales (Chart 3) paint a positive outlook for 2018.

Chart 3

Very strong end market growth in first quarter (Chart 4)

Based upon the combined 1Q’18 financial reports of 213 large, global OEMs, electronic equipment sales (consolidated into U.S. dollars) increased globally an estimated (and very robust) 10.6 percent in 1Q’18 vs. 1Q’17. While this world growth result is very heartening it was significantly inflated by exchange rate effects as stronger non-dollar currencies were converted into weaker dollars.

Chart 4

Looking at world electronic equipment sales consolidated into both dollars and euros, 1Q’18 growth rates are MUCH different (Chart 5). 1Q’18 vs.1Q’17 electronic equipment sales grew 10.6 percent in dollars but declined 4.3 percent in euros!

Chart 5

Certainly the first quarter was strong globally but the currency chosen for analysis can have a BIG effect.

U.S. supply chain expansion continues

Looking at the U.S. market (in dollars – therefore not distorted by exchange rates) domestic electronic equipment orders rose 6.7 percent in February-April 2018 versus the same three-month period in 2017. The U.S. electronic industry is doing reasonably well at present.

www.census.gov/manufacturing/m3/

Expect the recent exchange rate based amplification of dollar denominated global growth to taper off quickly.

Keep a careful watch on the geopolitical situation.

Walt Custer of Custer Consulting Group is an analyst focused on the global electronics industry.

Originally published on the SEMI blog.

pSemi™ Corporation (formerly known as Peregrine Semiconductor), a Murata company focused on semiconductor integration, announces the availability of the PE29101 gallium nitride (GaN) field-effect transistor (FET) driver for solid-state light detection and ranging (LiDAR) systems. The PE29101 boasts the industry’s fastest rise times and a low minimum pulse width. This high-speed driver enables design engineers to extract the full performance and switching speed advantages from GaN transistors. In solid-state LiDAR systems, faster switching translates into improved resolution and accuracy in the LiDAR image.

“As GaN is proving its relevance in applications like solid-state LiDAR, design engineers are using pSemi high-speed drivers to maximize the fast switching benefits of GaN,” says Jim Cable, chief technology officer of pSemi. “Because of its rise and fall speed, the PE29101 enables the highest possible resolution imagery—something the industry needs for LiDAR to reach its fullest potential.”

LiDAR operates on the same principles as radar but instead uses pulsed lasers to precisely map surrounding areas. Traditionally used in high-resolution mapping, LiDAR is now used in advanced-driver assistance programs (ADAS) and is widely seen as an enabling technology to fully autonomous vehicles. Furthermore, solid-state LiDAR has emerged as the future leader in the commercialization of LiDAR systems, largely due to its affordability, reliability and compact size compared to mechanical sensors.

In LiDAR systems, the pulse laser’s switching speed and rise time directly impacts the measurement’s accuracy. To improve resolution, the current must switch as quickly as possible through the laser diode. GaN technology offers LiDAR systems superior resolution and a faster response time because of its very low input capacitance and its ability to switch significantly faster than metal-oxide semiconductor field-effect transistors (MOSFETs).

GaN FETs must be controlled by a very fast driver to maximize their fast-switching potential. Increasing the switching speed requires a driver with fast rise times and a low minimum output pulse width. The PE29101 offers these key performance specifications, enabling GaN technology to improve LiDAR resolution.

WIN Semiconductors Corp (TPEx:3105), the world’s largest pure-play compound semiconductor foundry, has expanded its portfolio of highly integrated GaAs technologies with the release of a new pHEMT technology. The PIH0-03 platform incorporates monolithic PIN and vertical Schottky diodes with WIN’s high performance 0.1um pseudomorphic HEMT process, PP10. This integrated technology, PIH0-03, adds a highly linear vertical Schottky diode with cut-off frequency over 600GHz, as well as multi-function PIN diodes while preserving the state-of-the-art mmWave performance of the PP10 technology. The availability of monolithic PIN and Schottky diodes with a high performance mmWave transistor enables on-chip integration of a wide range of functions, including mixers, temperature/power detecting, limiters, and high frequency switching, and supports power, low noise and optical applications through100 GHz.

This integrated technology provides users with multiple pathways to add on-chip functionality and reduce the overall die count of complex multi-chip modules used in a variety of end-markets. In addition to high frequency switching, the monolithic PIN diodes can be used for low parasitic capacitance ESD protection circuits, and as an on-chip power limiter to protect sensitive LNAs in phased array radars. The vertical Schottky diodes enable numerous detecting and mixing functions and can be combined with the PIN diodes in unique limiter applications.

“Today’s complex systems and highly competitive markets require increased mmWave performance and more functionality per chip. The PIH0-03 platform is the latest example of how WIN Semiconductors is addressing these critical market needs by offering high performance GaAs technologies with new levels of multifunction integration. To meet the ever-increasing demands of next generation mobile user equipment, wireless infrastructure, fiber optics and military applications, WIN Semiconductors continues to commercialize advanced, highly integrated GaAs solutions and provide our customers a clear technology advantage,” said David Danzilio, Senior Vice President of WIN Semiconductors Corp.

Semiconductors N.V. (NASDAQ:NXPI) has expanded its cellular infrastructure portfolio of GaN and silicon laterally diffused metal oxide semiconductor (Si-LDMOS) products that deliver industry leading performance in a compact footprint to enable next-generation 5G cellular networks.

Spectrum expansion, higher order modulation, carrier aggregation, full dimension beam forming, and other enablers of 5G connectivity will require an expanded base of technologies to support enhanced mobile broadband connectivity. With spectrum usage and network footprints, multiple-input, multiple output (MIMO) technologies from four transmit (4TX) antennas to 64 TX and higher will be employed. The future of 5G networks will depend on GaN and Si-LDMOS technologies and NXP is at the forefront in its RF power amplifier development.

“Building on the success of 25 years of LDMOS leadership — NXP released the world’s first LDMOS product in 1992 — today, we are extending our RF leadership with industry leading GaN technology, developed with the highest linear efficiency for cellular applications,” said Paul Hart, senior vice president and general manager of NXP’s RF Power business. “Backed with the best supply chain, global applications support and unparalleled design expertise in the industry, NXP is positioned to be the leading RF partner for 5G solutions.”

At IMS 2018, NXP is introducing new RF GaN wideband power transistors and expanding its Airfast third-generation Si-LDMOS portfolio of macro and outdoor small cell solutions. The new offerings include:

  • A3G22H400-04S: Ideally suited for 40 W base stations, this GaN product yields up to 56.5 percent efficiency and 15.4 dB of gain and covers cellular bands from 1800 MHz to 2200 MHz.
  • A3G35H100-04S: Providing 43.8 percent efficiency and 14 dB of gain, this GaN product enables 16 TX MIMO solutions at 3.5 GHz.
  • A3T18H400W23S: This Si-LDMOS product is leading the way to 5G at 1.8 GHz with Doherty efficiency up to 53.4 percent and gain of 17.1 dB.
  • A3T21H456W23S: Covering the full 90 MHz band from 2.11 GHz to 2.2 GHz, this solution exemplifies NXP’s best-in-class Si-LDMOS performance for efficiency, RF power and signal bandwidth.
  • A3I20D040WN: Within NXP’s family of integrated ultra-wideband LDMOS products, this solution offers peak power of 46.5 dBm with 365 MHz wideband class AB performance of 32 dB of gain, 18 percent efficiency at 10 dB OBO.
  • A2I09VD030N: This offering boasts peak power of 46 dBm with class AB performance of 34.5 dB gain, 20 percent efficiency at 10 dB OBO. The RF bandwidth for this product is 575 MHz to 960 MHz.

The breadth of the company’s RF Power technologies—which include GaN, silicon-LDMOS, SiGe, and GaAs—allows product options for 5G that span frequency and power spectrums with varying levels of integration. This wide array of options, combined with the products that NXP builds for digital computing, and baseband processing, makes NXP a unique supplier of end-to-end 5G solutions.

To learn more, visit NXP at the International Microwave Symposium (IMS 2018) June 10-15 at booth #739 or at www.nxp.com/RF.

pSemi Corporation (formerly Peregrine Semiconductor), a Murata company focused on semiconductor integration, announces the expansion of its digital step attenuator (DSA) portfolio with a family of value, high-performance DSAs. The four value DSAs feature industry-leading attenuation accuracy at an entry-level price point.

“pSemi has a long, successful history of digital step attenuator development,” says Jim Cable, CTO at pSemi. “Our team introduced the world’s first single-chip DSA in 2004, and now, we are further expanding our DSA portfolio with the introduction of the value, high-performance DSAs. These four new DSAs nicely round out our DSA portfolio and complement our RF catalog parts.”

The value DSA family—the PE43620, PE43650, PE43665 and PE43670—are offered in a 2-bit, 5-bit, 6-bit or 7-bit configuration. These high-performance DSAs have excellent attenuation accuracy, low insertion loss and high linearity. The four products are available in compact QFN packages.

For 1K-quantity orders, the PE43620 DSA (2 bit, 50-ohm) is $0.63 each; the PE43650 (5 bit, 50-hm) is $1.44 each. The PE43665 (6-bit, 75-ohm) is $1.23 each, and the PE43670 (7-bit, 50-ohm) is $2.02each. Volume-production parts, samples and evaluation kits will be available in August.

pSemi Corporation is a Murata company driving semiconductor integration. pSemi builds on Peregrine Semiconductor’s 30-year legacy of technology advancements and strong IP portfolio but with a new mission: to enhance Murata’s capabilities with high-performance RF,

The semiconductor industry is nearing a third consecutive year of record equipment spending with projected growth of 14 percent (YOY) in 2018 and 9 percent in 2019, a mark that would extend the streak to a historic fourth consecutive growth year, according to the latest update of the World Fab Forecast report published by SEMI. Over the semiconductor industry’s 71-year history, only once before – in the mid 1990s – has the industry logged four consecutive years of equipment spending growth.

Korea and China are leading the growth, with Samsung dominating global spending and ascendant China on a fast, steep rise, surging ahead of all other markets. See Figure 1.

Figure 1 equipment spending by region (includes new and refurbished)

While Samsung is expected to reduce equipment investments in 2018, the company still accounts for a dominant 70 percent of all investment in Korea. At the same time, SK Hynix is increasing its equipment spending in Korea.

China’s equipment spending is forecast to increase 65 percent in 2018 and 57 percent in 2019.  Notably, 58 percent of investments in China in 2018 and 56 percent in 2019 stem from companies with headquarters in other regions such as Intel, SK Hynix, TSMC, Samsung, and GLOBALFOUNDRIES. Domestic, Chinese-owned companies – backed by large government initiatives – are building a considerable number of new fabs that will start equipping in 2018. The companies are expected to double their equipment investments in 2018 and again in 2019.

Other regions are also ramping up investments. Japan is increasing equipment spending by 60 percent in 2018, with the largest increases by Toshiba, Sony, Renesas and Micron.

The Europe and Mideastern region will boost investments by 12 percent in 2018, with Intel, GLOBALFOUNDRIES, Infineon and STMicroelectronics the largest contributors.

Southeast Asia will boost investments by more than 30 percent in 2018, although total spending is proportionately smaller than in other regions owing to its size. The main contributors are Micron, Infineon and GLOBALFOUNDRIES, though companies including OSRAM and ams are also increasing investments.

The SEMI World Fab Forecast, which also includes information on other companies, covers data and predictions through the end of 2019, including milestones, detailed investments by quarter, product types, technology nodes and capacities down to fab and project level.

Learn more about the SEMI fab databases at:

www.semi.org/en/MarketInfo/FabDatabase and www.youtube.com/user/SEMImktstats.

Quantum bits are now easier to manipulate for devices in quantum computing, thanks to enhanced spin-orbit interaction in silicon.

A silicon quantum computer chip has the potential to hold millions of quantum bits, or qubits, for much faster information processing than with the bits of today’s computers. This translates to high-speed database searches, better cybersecurity and highly efficient simulation of materials and chemical processes.

Now, research groups from Purdue University, the Technological University of Delft, Netherlands and the University of Wisconsin-Madison have discovered that silicon has unique spin-orbit interactions that can enable the manipulation of qubits using electric fields, without the need for any artificial agents.

Researchers are taking advantage of a newly found phenomenon in silicon that makes quantum bits easier to manipulate, leading to faster and longer-lived information processing via quantum computing. Credit: (Purdue University image/Rifat Ferdous)

“Qubits encoded in the spins of electrons are especially long-lived in silicon, but they are difficult to control by electric fields. Spin-orbit interaction is an important knob for the design of qubits that was thought to be small in this material, traditionally,” said Rajib Rahman, research assistant professor in Purdue’s School of Electrical and Computer Engineering.

The strength of spin-orbit interaction, which is the interaction of an electron’s spin with its motion, is an important factor for the quality of a qubit. The researchers found more prominent spin-orbit interaction than usual at the surface of silicon where qubits are located in the form of so-called quantum dots – electrons confined in three dimensions. Rahman’s lab identified that this spin-orbit interaction is anisotropic in nature – meaning that it is dependent on the angle of an external magnetic field – and strongly affected by atomic details of the surface.

“This anisotropy can be employed to either enhance or minimize the strength of the spin-orbit interaction,” said Rifat Ferdous, lead author of this work and a Purdue graduate research assistant in electrical and computer engineering. Spin-orbit interaction then affects qubits.

“If there is a strong spin-orbit interaction, the qubit’s lifetime is shorter but you can manipulate it more easily. The opposite happens with a weak spin-orbit interaction: The qubit’s lifetime is longer, but manipulation is more difficult,” Rahman said.

The researchers published their findings on June 5 in Nature Partner Journals – Quantum Information. The Wisconsin-Madison team fabricated the silicon device, the Delft team performed the experiments and the Purdue team led the theoretical investigation of the experimental observations. This work is supported by the Army Research Office, U.S. Department of Energy, the National Science Foundation and the European Research Council.

Upcoming work in Rahman’s lab will focus on taking advantage of the anisotropic nature of spin-orbit interactions to further enhance the coherence and control of qubits, and, therefore, the scaling up of quantum computer chips.

In the field of photovoltaic technologies, silicon-based solar cells make up 90% of the market. In terms of cost, stability and efficiency (20-22% for a typical solar cell on the market), they are well ahead of the competition.

However, after decades of research and investment, silicon-based solar cells are now close to their maximum theoretical efficiency. As a result, new concepts are required to achieve a long-term reduction in solar electricity prices and allow photovoltaic technology to become a more widely adopted way of generating power.

One solution is to place two different types of solar cells on top of each other to maximize the conversion of light rays into electrical power. These “double-junction” cells are being widely researched in the scientific community, but are expensive to make. Now research teams in Neuchâtel – from EPFL’s Photovoltaics Laboratory and the CSEM PV-center – have developed an economically competitive solution. They have integrated a perovskite cell directly on top of a standard silicon-based cell, obtaining a record efficiency of 25.2%. Their production method is promising, because it would add only a few extra steps to the current silicon-cell production process, and the cost would be reasonable. Their research has been published in Nature Materials.

This scanning electron microscopy image shows Silicon’s pyramids covered with perovskite. Credit: EPFL

Perovskite-on-silicon: a nanometric sandwich

Perovskite’s unique properties have prompted a great deal of research into its use in solar cells over the last few years. In the space of nine years, the efficiency of these cells has risen by a factor of six. Perovskite allows high conversion efficiency to be achieved at a potentially limited production cost.

In tandem cells, perovskite complements silicon: it converts blue and green light more efficiently, while silicon is better at converting red and infra-red light. “By combining the two materials, we can maximize the use of the solar spectrum and increase the amount of power generated. The calculations and work we have done show that a 30% efficiency should soon be possible,” say the study’s main authors Florent Sahli and Jérémie Werner.

However, creating an effective tandem structure by superposing the two materials is no easy task. “Silicon’s surface consists of a series of pyramids measuring around 5 microns, which trap light and prevent it from being reflected. However, the surface texture makes it hard to deposit a homogeneous film of perovskite,” explains Quentin Jeangros, who co-authored the paper.

When the perovskite is deposited in liquid form, as it usually is, it accumulates in the valleys between the pyramids while leaving the peaks uncovered, leading to short circuits.

A key layer ensuring an optimal microstructure

Scientists at EPFL and CSEM have gotten around that problem by using evaporation methods to form an inorganic base layer that fully covers the pyramids. That layer is porous, enabling it to retain the liquid organic solution that is then added using a thin-film deposition technique called spin-coating. The researchers subsequently heat the substrate to a relatively low temperature of 150°C to crystallize a homogeneous film of perovskite on top of the silicon pyramids.

“Until now, the standard approach for making a perovskite/silicon tandem cell was to level off the pyramids of the silicon cell, which decreased its optical properties and therefore its performance, before depositing the perovskite cell on top of it. It also added steps to the manufacturing process,” says Florent Sahli.

Updating existing technologies

The new type of tandem cell is highly efficient and directly compatible with monocrystalline silicon-based technologies, which benefit from long-standing industrial expertise and are already being produced profitably. “We are proposing to use equipment that is already in use, just adding a few specific stages. Manufacturers won’t be adopting a whole new solar technology, but simply updating the production lines they are already using for silicon-based cells,” explains Christophe Ballif, head of EPFL’s Photovoltaics Laboratory and CSEM’s PV-Center.

At the moment, research is continuing in order to increase efficiency further and give the perovskite film more long-term stability. Although the team has made a breakthrough, there is still work to be done before their technology can be adopted commercially.

Winbond Electronics Corporation, a global supplier of semiconductor memory solutions, today announced the introduction of the W25N01JW, a high-performance, 1.8V Serial NAND Flash memory IC delivering a new high in data-transfer rates: 83MB/s via a Quad Serial Peripheral Interface (QSPI).

Winbond’s new high-performance Serial NAND technology also supports a two-chip dual quad interface which gives a maximum data transfer rate of 166MB/s.

This high-speed Read operation, some four times faster than existing serial NAND memory devices offer, means that the 1.8V W25N01JW chip can replace SPI NOR Flash memory in automotive applications such as data storage for instrument clusters or the center information displays (CIDs).

This is important for automotive OEMs because the adoption of more sophisticated graphics displays in the instrument cluster, and larger display sizes of 7 inches and above in the CID, is increasing system memory requirements to capacities of 1Gbit and higher. At these capacities, serial NAND Flash has a markedly lower unit cost than that of SPI NOR Flash, and occupies a smaller board area per megabit of storage capacity.

SPI NOR Flash has been the preferred memory technology in automotive displays for many years because of its high read speed, which supports the fast boot requirements of automotive user interfaces, and because of its high reliability and long data retention. By raising the data transfer rate of its serial NAND technology to 83MB/s – matching the read speed of automotive SPI NOR Flash – Winbond has ensured that the W25N01JW can support fast boot operation and the demanding requirements of sophisticated graphics applications.

The W25N01JW also meets strict automotive requirements for quality and reliability. Built with high-reliability single-level sell (SLC) memory technology, and implementing 1-bit error correction code (ECC) on all Read and Write operations, it complies with the endurance, retention and quality requirements of the AEC-Q100 standard and relevant JEDEC specifications.

The W25N01JW device operates from -40°C to 105°C and retains data for 10 years at 85°C after 1,000 program/erase cycles, whereas eMMC can only retain data for a fraction of that time under these conditions even when used in SLC mode, which are today widely used for data storage in the CIDs of high-end vehicles.

“Cars’ large and attractive displays need higher memory capacity, beyond the ‘sweet spot’ of SPI NOR Flash, which is good for up to 512Mbits,” said William Chen, deputy director of the Flash Product Marketing Division at Winbond. “For systems that require high-speed memory in capacities of 1Gbit or higher, Winbond’s high-performance Serial NAND Flash is the new best choice for automotive OEMs, offering a combination of lower unit cost, smaller size and excellent reliability and data retention.”

The W25N01JW is available for sampling today in a capacity of 1Gbit. A two-chip implementation in dual-quad I/O mode provides 2Gbits of memory capacity and a maximum data transfer rate of 166MB/s.

The chip is available in industrial grade and in an extended-temperature automotive grade version operating at up to 105°C. It is compatible with standard SPI NAND Flash protocols. It is housed in standard 8mm x 6mm WSON and TFBGA packages that are footprint-compatible with standard SPI NOR Flash products.

WIN Semiconductors Corp (TPEx:3105), the world’’s largest pure-play compound semiconductor foundry, has expanded its gallium nitride (GaN) process capabilities to include a 0.45?m-gate technology that supports current and future 5G applications. The NP45-11 GaN-on-SiC process allows customers to design hybrid Doherty power amplifiers used in 5G applications including massive MIMO (multiple-input and multiple-output) wireless antenna systems. Similar to macro-cell applications, MIMO base stations often combine Doherty power amplifiers with linearization techniques to meet demanding linearity and efficiency specifications of today’s wireless infrastructure.

GaN devices outperform the incumbent LDMOS technology, offering superior efficiency, instantaneous bandwidth and linearity, particularly in the higher frequency bands utilized in 5G radio access networks.

Ideal for use in sub-6 GHz 5G applications including macro-cell transmitters and MIMO access points, the NP45-11 technology supports power applications from 100 MHz through 6GHz. This discrete transistor process is environmentally rugged, incorporating advanced moisture protection and meets the JEDEC JESD22-A110 biased HAST qualification at 55 volts. Combined with WIN Semiconductors’ environmentally rugged high voltage passive technology, IP3M-01, the NP45-11 technology enables hybrid power amplifiers in a low cost plastic package.

The NP45-11 technology is fabricated on 100mm silicon carbide substrates and operates at a drain bias of 50 volts. In the 2.7GHz band, this technology provides saturated output power of 7 watts/mm with 18 dB linear gain and more than 65% power added efficiency without harmonic tuning.

“5G radio access networks create several challenges to power amplifier designs used in MIMO systems. High output power and linear efficiency are primary design objectives to meet performance specifications and lower total cost of ownership. The tradeoff between output power and linearized efficiency is significant because of the high peak-to-average power ratio employed in today’s wireless modulation schemes. This tradeoff becomes more difficult in 5G applications due to greater instantaneous bandwidth requirements and higher operating frequency,” said David Danzilio, Senior Vice President of WIN Semiconductors Corp.