Notes from The ConFab 2016 – Day 1 of The Confab 2016

By Dick James, Senior Technology Analyst, Chipworks

The ConFab 2016 kicked off June 13 in the Encore Hotel in Las Vegas, the 12th in the series, presented by Solid State Technology (part of Extension Media), which they promote as the “Premier Conference and Networking Event for the Semiconductor Manufacturing & Design Industry.”

The event started with a networking reception Sunday night, giving the early arrivals a chance to mingle with some good food and wine. A feature of The Confab is that networking lunches and receptions are a focused part of the agenda, and time is set aside for face-to-face meetings; these can be pre-arranged by the event staff. Attendance is usually limited to ~150 so that there is ample time for everyone to get together over the three days.

As usual, Pete Singer was the conference chair, and the keynote speaker opening the event was Tom Caulfield, SVP and GM of GLOBALFOUNDRIES’ (GF) Fab 8 in Malta, New York, speaking on “Unlocking the IoT Opportunity for the Next Golden Age.” He surprised me at the start by saying that “the best years of semiconductors are ahead of us, not behind us,” given that it is hard to see even five years ahead at the moment, and pessimists are predicting that leading edge technologies will price themselves out of the business.

Tom Caulfield evangelizing the Next Golden Age of Semiconductors

Tom Caulfield evangelizing the Next Golden Age of Semiconductors

Tom then made the point that the main driver for the industry through its existence has been the evolution of connectivity, and the next phase will be as well.



And of course that takes us to the Internet of Things (IoT), currently at the peak of the hype curve, but undoubtedly a real phenomenon. McKinsey & Co. have predicted that by 2019/20 the IoT semiconductor value will be $50B – $75B, and they have broken it down nicely into segments and technologies;

What this doesn’t show is that all of the things will generate vast amounts of data, which will need a 5G communications infrastructure, which Tom described as huge, and maybe the biggest opportunity, rather than the silicon in the things themselves.

Then we moved on to the need for collaboration, the business model needs to innovate as well as the technology; though my perception is that there’s a good deal of collaboration in the industry already, though maybe not as much as needed. The GF-Samsung 14 nm agreement was mentioned (though I gather that it is 14 nm only), and design/technology co-optimization, which is now essential in the foundry business – Intel has been doing it for years.

The need for cooperation goes beyond the chip industry, though, and the Albany area was used as an example, since it embodies the three “E”s – education, economy, and ecosystem, i.e. workforce development, government support, and access to the tech cluster around CNSE.

Tom finished up with a plug for the new AMD Radeon 480 GPU, fabbed on the GF 14LPP process (we have one on order!), and a wrap-up of the above.

The theme of the morning session was “The Semiconductor Industry Outlook for 2016 and Beyond”. First up was Dan Armbrust, CEO of Silicon Catalyst, the industry’s first incubator company, which is trying to fill the void of start-up funding for new chip companies. I had not realised it, but the amount of venture capital (VC) money for start-up semiconductor companies has declined to near zero, even though VC funding is itself at almost record levels. Dan and his colleagues have set up a model whereby their partners provide in-kind support, reducing the need for actual seed capital, and Silicon Catalyst will also provide mentoring, physical space, business and legal services, and “lots of pizza”! Contributing partners


include TSMC, Synopsys, Advantest, Keysight, imec, PDF Solutions, Autodesk, Open Silicon, and the
MEMS foundry imt.

The next slide summarizes the model:


So far they have had three screening events, looked at 80+ applications, and selected ten companies for incubation. Seems like a good idea!

Lode Lauwers from imec was the next speaker; he did the usual obligatory description of imec’s capabilities and ecosystem, but once he got into the technical discussion, he put up a roadmap that extended to N+5, i.e. 2 nm, which is the first that I have seen.


In terms of possible technologies, I don’t think there’s anything new, but on our (now failing) two-year process cadence, that takes us out to 2026, so imec is looking a fair way ahead, and it seems the guys in R&D will have jobs for a while.

He also showed the following graphic of where broad applications fit on the roadmaps, so one perception is that IoT will only need technology down to 14 nm – I’m not sure the FDSOI lobby will agree with that, now that some of them are talking about stretching it to 7 nm.


He finished up with some examples of the collaborations that they are doing, using different flows and products such as memory and imagers.

My new colleague Kevin Gibb of TechInsights (TechInsights and Chipworks are now merging) next reviewed recent trends in chips, showing the scaling and some of the process changes we have seen in logic, DRAM and NAND flash technologies, and touching briefly on the die stacking in the Hynix HBM and a ReRAM example.



Kevin was followed by Hughes Metras, speaking for Europe’s other semiconductor collaborative research institute, CEA-Leti, with a slightly different roadmap, including FDSOI and their Coolcube monolithic 3D-stacking.


They are also looking at other forms of 3D integration, and Hughes showed examples of 2.5D/3D interposers showed Hughes showed examples of others that also include photonics devices amongst others. Of course we had examples of IoT – CEA-Leti has found applications in everything from medical to truck tires to pipelines.

Mark Reynolds from New York Empire State Development finished the morning session, describing the incentive programs that the state has in place to attract high-tech companies with high-income jobs there.

In essence this boils down to keeping the real estate off the books of the manufacturing company, by providing ready-to-go sites, infrastructure, and workforce, even going as far as building fabs with long term leases at incredibly competitive rates (e.g. $1!), and oftentimes including tooling and equipment. In addition there are tax credits, and the state has pumped oodles of cash into their schools, community colleges and universities to ensure a world class workforce.

We all know this has worked in luring AMD to build what is now the GLOBALFOUNDRIES fab in Malta; more recent examples are the new 300-mm ams fab in Utica (which has just started construction) Solar City in Buffalo (the largest PV plant in the US), the Soraa LED fab in Syracuse, and the GE SiC operation, with the fab in Albany and the packaging operation in Utica.

One could argue that it’s cheaper and easier just to write unemployment or welfare cheques for those in need, but the key to this strategy is the high-income jobs – surveys have shown (I’m told) that one job in a plant such as Malta has a five-to-one multiplier for other jobs, due to the infrastructure and social support (e.g. anything from schools to coffee shops) needed in the local area.

After lunch the main event of the afternoon was a panel on IoT, with Kelvin Low from Samsung Foundry, Rajeev Rajan from GF (VP IoT Product), Uday Tennety of GE Digital, and Jim Hewitt from Siemens as the moderator. There were lots of questions about applications and security, but occasionally we got onto the technology needed for IoT, and how compact the devices could be.

I was curious if the different elements could be integrated into one chip, since the basics of an IoT part are sensor(s), a microcontroller to process the data, a wireless interface, maybe some memory, and power management. These at the least require a range of process technology, since RF processes are usually different from logic and power, never mind the possibly of a MEMS sensor of some sort.

So I put the question, and was mildly surprised that both the foundry guys agreed that it is becoming possible, since FDSOI, with its back bias capability, allows a wider range of voltages and frequencies, and they clearly see this as an opportunity for them to get seriously into the IoT chip market.

The panel lasted an hour and a half or so, then we had a break before another reception.

ASMC 2016 Conference Has Highest Attendance Ever, Chipworks Achieves Twelfth Paper

By Dick James, Senior Technology Analyst, Chipworks

It’s spring in the north-eastern part of North America, and that means it’s the time of year for the Advanced Semiconductor Manufacturing Conference, in the amiable ambiance of Saratoga Springs, New York. The conference took place a couple of weeks ago, on May 16 – 19.

As the name says, ASMC is an annual conference focused on the manufacturing of semiconductor devices; in this it differs from other conferences, since the emphasis is on what goes on in the wafer fab, not the R&D labs, and the papers are not research papers – some are better described as “tales from the fab”! After all, it’s the nitty-gritty of manufacturing in the fab that gets the chips out of the door, and this meeting discusses the work that pushes the yield and volumes up and keeps them there.

I always come away impressed by the quality of the engineering involved; not being a fab person myself any more, it’s easy to get disconnected from the density of effort required to equip a fab, keep it running and bring new products/processes into production. Usually the guys in the fab only get publicity if something goes wrong!

There were 96 papers spread over the three days, 60 presentations and 36 posters, and the highest attendance ever at 350+ (registration was actually closed on day 1 – we ran out of room!). In addition we had keynotes from Don O’Toole of IBM and Christine Furstoss of GE Global Research, a tutorial on Nanoscale III-V CMOS by Jesús del Alamo from MIT, and to finish the Wednesday afternoon there was a panel discussion on “Moore’s Law Wall vs. Moore’s Wallet, and Where Do We Grow From Here?”. Bob Maire of Semiconductor Advisors wrapped up the conference Thursday lunchtime with a talk on China’s effect on the semiconductor biz; “Mergers & Acquisitions in the Semiconductor Industry – Could China Cause Continued Consolidation?”

Full House at Don O’Toole’s ASMC Keynote

Full House at Don O’Toole’s ASMC Keynote

I guess it’s a reflection of the location, but 46 out of the 96 papers were from Silicon on the Hudson – GLOBALFOUNDRIES, IBM, and CNSE/G450 affiliates. Having said that, there were papers from the likes of Samsung, TSMC, and UMC, not to mention NXP, Infineon, ON Semi, and others, plus some academic and student papers. It’s always tough to get papers from far afield these days, especially with tightening travel budgets and visa requirements, not to mention the gut desire to keep internal information in-house.

And of course Chipworks usually has an offering, though we missed out last year due to personal circumstances, otherwise it would be twelve years in a row. It’s a bit of an odd fit, since we are a service company that doesn’t make anything; but we do take the leading edge chips apart, and it seems the fab guys at the conference like seeing the competition’s stuff – and their own – since, if you’re deep into running the fab, you don’t get much of a chance to look at the final product.

And, now that we’ve been presenting since 2005, our papers are actually a condensed history of the technology from the 90-nm era down to 14-nm finFETs – if you read the references at the end of the blog you’ll see we’ve covered a fair spread of technology, not just logic transistors, but also flash and DRAM memory.

The initial reason for my submitting a paper back in 2004 for the 2005 conference was that ASMC that year was co-located with Semicon Europa – and I liked the idea of a trip to Munich! We were also a growing company, and starting to flex some of our marketing muscles by presenting at, rather than just attending conferences. In that context, the 2005 conference was a success, since Tom Cheyney, editor of the now-defunct Micro magazine, invited me to write regular articles for the publication, and that led to a series of articles and blogs that is still going.

Looking back at the older presentations, they really are a trip down memory lane – remember that first Intel 90-nm transistor with embedded silicon-germanium source/drains for the PMOS? We looked at that in 2005 [1].

Fig2_Intel 1


The compressive stress given by the SiGe turned out to be a very effective tool for cranking up the strain in the channel, to the extent that PMOS and NMOS drive currents are now comparable, definitely a different design paradigm from the days of my youth.

And it turned out that the technology was transferable to high-k, metal-gate (HKMG) finFETs – witness the latest 14-nm Intel PMOS device:

Fig3_Intel 2


Taking a different tack, IBM was already using SOI, but before they used embedded stress techniques, the SOI layer was only 45 nm thick – not quite FDSOI, but thinner than their current (GLOBALFOUNDRIES) HKMG offering that uses 80-nm thick SOI.

Fig4_IBM 1


As you can see below, things are considerably more complex these days!

Fig5_IBM 2-2


When it comes to memory, 90-nm DRAM was the order of the day, and the recessed channel array transistor (RCAT) had just been introduced:

Fig6_Samsung 1


Fig7_Samsung 2


Now we have 10-nm class (likely 18-nm) 8-Gb DRAMs, though the latest I reviewed at ASMC [9] was a 26-nm 4-Gb part in 2013 – that was three generations ago!

Fig8_Samsung 3


Fig9_Samsung 4


In the meantime we have seen the introduction of buried tungsten saddle-fin transistors for the wordlines (buried wordlines – BWL), ZAZ (zirconia/alumina/zirconia) high-k capacitor dielectrics, and air-gaps; shrinking cell area by more than a factor of ten. The node definition has also moved from half the M1 pitch to half of the active silicon pitch, nothing stays the same in our business.

I didn’t talk about flash that first year, but a couple of years later [3] the leading edge was a 62-nm, 8-Gb part, and 50-nm was starting to come into production. The conference that year was in Stresa, on Lake Maggiore in Italy, one of the more exotic locations that we’ve been to.

Fig10_Samsung 5


The latest in planar flash in 2013 [9] was a 19-nm Toshiba 128-Gb device:

Fig11_Toshiba 1


We still have the conventional floating gate/control gate structure, but cell size has shrunk by an order of magnitude, we have air gaps between cells, and in order to keep effective coupling between control gate and floating gate, the aspect ratio of the floating gate has increased from ~1.3 to ~4.8.

In 2016, of course, we have planar flash down to the 15-nm generation, including the use of high-k dielectric, and we are into the third-generation vertical flash parts.

So much for then and now – this year’s conference had 15 different sessions:

  • Contamination Free Manufacturing (CFM)
  • Advanced Metrology I & II
  • Defect Inspection I & II
  • Factory Optimization I & II
  • Advanced Equipment and Materials Processes
  • Yield Enhancement & Yield Learning
  • Advanced Equipment/CFM
  • Advanced Patterning/CFM
  • Advanced Process Control (APC)
  • Yield Enhancement
  • 3D TSV

Including a poster session for shorter papers that covered all the above topics.
The subjects of individual papers ranged from improvements to chemical-mechanical planarization, through threshold voltage variations in HKMG gates due to non-uniform alloying, to ‘smart manufacturing’ in legacy 200mm fabs, and multiple papers on virtual metrology – i.e. a broad swath of the practical wafer manufacturing problems to fab loading algorithms and everything in between. The detailed schedule can be found here, and no doubt the proceedings will be available through IEEE Xplore in due course.

Next year’s ASMC will again be in Saratoga Springs, on May 15 – 18; we hope to see you there!


  • James, 2004 – The Year of 90-nm: A Review of 90 nm Devices, Proc. ASMC 2005
  • James, Low-K and Interconnect Stacks – a Status Report, Proc. ASMC 2006
  • James, Nano-Scale Flash in the Mid-Decade, Proc. ASMC 2007
  • James, From Strain to High K/Metal Gate – the 65/45 nm Transition, Proc. ASMC 2008
  • James, Design-for-Manufacturing Features in Nanometer Processes – A Reverse Engineering Perspective, Proc. ASMC 2009
  • James, Recent Innovations in DRAM Manufacturing, Proc. ASMC 2010
  • Fontaine, Recent Innovations in CMOS Image Sensors, Proc. ASMC 2011
  • James, High-k/Metal Gates in Leading Edge Silicon Devices, Proc. ASMC 2012
  • James, Recent Advances in Memory Technology, Proc. ASMC 2013
  • James, 3D ICs in the Real World, Proc. ASMC 2014
  • James, High-k/Metal Gates in the 2010s, Proc. ASMC 2014
  • James, Moore’s Law Continues into the 1x-nm Era, Proc. ASMC 2016


What to Expect in 2016 in the Chipworld

By Dick James, Senior Technology Analyst, Chipworks

It’s the time in the media world that we see a frenzy of predictions for the coming year. They are mostly business or tech trends, so I figured I might as well chip in (har! har!), and give a more detailed idea of what new semiconductor products we look forward to this year, now that we are in 2016.

This might seem to be a bit like fortune-telling, but it’s actually a compilation of the notes we’ve made from this year’s press announcements, coupled with the trends we’ve observed in our reverse engineering, and keeping an open ear at the industry events that we’ve attended.

Logic & Foundries

2016 will be a relatively quiet year when it comes to the leading-edge processes, since we do not expect to see a high-volume of 10 nm products this year. There has been a fair bit of comment that the upcoming Apple A10 processor might be on 10 nm this autumn, but to me it seems a real stretch to expect a full node advance barely 18 months after the introduction of a 14 nm product from Samsung, and a 16 nm product from TSMC, especially in the volume that Apple would require.

We do expect to see the second generation 14/16 nm processes, FinFET Plus (16FF+) from TSMC and 14LPP from Samsung and possibly their co-supplier GLOBALFOUNDRIES. The second-tier foundries such as UMC and SMIC will be ramping up their 28 nm high-k metal gate (HKMG) product, so we will be monitoring those as we get them. It appears that UMC will be skipping 20 nm and going straight to 14 nm, but that will not likely appear until 2017.

When it comes to fully depleted silicon on insulator (FD-SOI), we expect to see a mainstream 28 nm product this year, since Samsung has stated that they are producing, and have shipped more than a million wafers, with STMicroelectronics as one of their lead customers. Chipworks has already analyzed a custom 28 nm FD-SOI ASIC manufactured at STMicroelectronics, which was a simple implementation without back-bias. For the mainstream parts, we will be analyzing back-bias implementation, if found, as it is touted as one of the big advantages of FD-SOI.

GLOBALFOUNDRIES is also on the FD-SOI bandwagon, but they seem to be concentrating on their 22FDX™ processes. A number of the ASIC design houses are claiming to be designing into those, so with luck we will see some very early product by year-end.

There will also be a continued emphasis on low power variants of older generation processes, such as 40 and 55 nm, aimed at mobile/wearable devices where battery life is critical.

To finish up, another process sector where we expect to see development is radio frequency silicon on insulator (RF-SOI). We are already seeing the introduction of RF-SOI into products such as antenna switches for the RF front end of mobile phones.


This will be another year of evolution for dynamic random-access memory (DRAM), with the introduction of 1X nm generation memories by the big three (Micron, Samsung, and SK Hynix), although possibly not until year-end.

Fig 1


Micron has 1X and 1Y nm nodes in its roadmap (above), enabling 1X volume mid-2016.

Fig 2


Samsung predicts three 1X nm nodes (see above), though there is no time scale here; however, we already have their 20 nm part, which is in volume production, so it’s reasonable to expect a 1X nm part this year. We haven’t heard anything formal from SK Hynix, but again, we already have their 20 nm part, so we would expect a 1X nm device in 2016.

The other facet of the DRAM business is stacked memory using through-silicon vias (TSVs); in 2015 we saw the Samsung version, and the SK Hynix High Bandwidth Memory (HBM). We’re still waiting for the Intel/Micron Hybrid Memory Cube (HMC), and we expect to get our hands on that this year, as well as the HBM2 from SK Hynix AND Flash.

The big news in NAND flash memory is the introduction of 3D/vertical technology, with the bitcells stacked one above another instead of on the die surface. Samsung launched their V-NAND over a year ago, with 32 active layers in both multi-level cell (MLC) and tri-level cell (TLC) versions, using charge-trap storage technology. They are now shipping the third generation part (256 Gb) with 48 layers, so we should see that in the near future.

Last month, at IEDM (International Electron Devices Meeting), Intel/Micron detailed their 3D-NAND, a 32-layer device, this time using conventional floating gate charge storage. According to their investor calls, they are sampling these at the moment and should be shipping volume in the second half of this year.

SEM cross-section of Intel/Micron vertical-channel 3D-NAND structure

SEM cross-section of Intel/Micron vertical-channel 3D-NAND structure

SanDisk/Toshiba are also sampling, but their 3D-NAND is a 48-layer, 256-Gb TLC device, built using their own Bit-Cost Scalable (BiCS) charge-trap technology. They have been more cautious about the economics of launching 3D technology, but again I look forward to getting some in 2016. Last, but not necessarily least, SK Hynix claim that they are already in 3D production, and we should also see their floating-gate version this year.

SK Hynix

In parallel, all the companies are still evolving planar flash products – we will likely find 13 – 15 nm planar flash chips being launched, since the 15/16-nm ones are already here.
Emerging Memory
The highest-profile announcement this year for this memory class was from Intel/Micron, on their 3D XPoint memory; this appears to be some sort of resistive random-access memory (RRAM), using “bulk properties” to provide memory storage in a cross-point layout. Both Intel and Micron predict a big future for this product; Micron claims that the 3D XPoint business could easily be of the same order of magnitude as their DRAM businesses by 2018, and Intel sees broad applications for 3D XPoint memory (dubbed Optane), and says that it will be available this year for PC and server usage.

Fig 4


Less noticed was a similar release from SanDisk and HP, also detailing storage-class RRAM-based memory, but with no details as to launch dates. Micron and Sony also have a jointly developed RRAM, but again no dates.

Image Sensors

There has been a steady evolution in the image sensor biz, with Sony leading the pack, and culminating in the deep-trench isolation between pixels in the Apple 6s/6s Plus camera. Sony has had a two year+ lead in stacking the sensor on top of the image processor and connecting the two with custom TSVs, but we now see OmniVision and Samsung with design wins using multiple versions of its new stacked chip products.

We can, no doubt, expect to see a further-evolved camera chip in the iPhone 7, and Apple’s competitors will also be pushing the envelope, so we will be monitoring every new smartphone to see what appears.

Meanwhile, other sectors are developing fast – to name two, the push on automated driver self-assist (ADAS) and self-driving vehicles is providing a new space for lower-tech (but different specifications) image sensors, and security is likely to be a hot market given last year’s terror attacks.
Advanced Packaging

Packaging technology has been in as much ferment as any of the wafer fab technologies, with 2.5/3D stacking getting most of the press. We expect 2016 to be a busy year in that space too; TSMC is producing its Chip-on-Wafer-on-Substrate (CoWoS) silicon interposer for a limited range of products, and seems about to launch its cheaper integrated fan-out (InFO) organic substrate, possibly using it for the Apple A10 system-on-chip (SoC) this fall.

Fig 5


TSMC has TSVs in volume production, though not high-density for 2.5/3D; the new fingerprint sensor in the Apple 6s/6s Plus uses TSVs so that the wire bonds don’t get in the way of the sapphire touchpad.

Intel has a parallel “Embedded Multi-die Interconnect Bridge” (EMIB) technology (to TSMC’s InFO), and given the completed Altera deal, we may finally see a 14 nm field-programmable gate array (FPGA) product with EMIB this year.

Add in the Intel/Micron HMC, SK Hynix’s HBM2 and Wide IO2 stack, and the OSATs are also pushing the envelope and likely to ship new formats this year, so there will be plenty for us to look at.

This has been a relatively high-level review of what we expect this year, but as you can see from the above, it will be quite a hectic year – lots of new technology for us to analyze!


Intel/Micron Detail Their 3D-NAND at IEDM

By Dick James, Senior Technology Analyst, Chipworks

On the Monday afternoon at IEDM the key paper for me was the Intel/Micron talk on their 3D-NAND flash part (paper 3.3), which is currently sampling to customers. Samsung put their V-NAND flash on the market last year, but that uses charge-trap technology, whereas the Intel/Micron device has adapted conventional floating gate technology to the vertical direction.

This is the first-generation product, with 32 active tiers plus additional layers for dummy wordlines and source and drain select gates. A vertical channel surround-gate structure is used for the flash cells. The CMOS decoders and sense-amps are situated under the NAND flash array, which saves significantly on die area. It appears that this product will be a 256-Gb memory, or 384 Gb when the TLC version is introduced. Die size is 168.5 mm2, giving a bit density of 1.52 and 2.28 Gb/mm2 for the MLC and TLC devices.

Intel/Micron 3D-NAND flash die (Source: Intel/Micron/IEDM)

Intel/Micron 3D-NAND flash die (Source: Intel/Micron/IEDM)

The wordlines/control gates are horizontal polysilicon layers with an ONO inter-poly dielectric, and the floating gates are also polySi. The vertical channel and tunnel dielectric are formed in holes etched through a horizontal polySi/oxide stack.

EM cross-section of vertical-channel 3D-NAND structure  (Source: Intel/Micron/IEDM)

EM cross-section of vertical-channel 3D-NAND structure (Source: Intel/Micron/IEDM)

The process is shown below; the cell hole is first etched through the wordline tiers, and then the control gate is recessed back and the inter-poly dielectric is formed. The floating gate is then deposited, and etched back to form an isolated floating gate in each cell; the tunnel-oxide is formed, and the polySi channel is deposited to line the hole in the stack.

Process flow of vertical-channel 3D-NAND stack formation  (Source: Intel/Micron/IEDM)

Process flow of vertical-channel 3D-NAND stack formation (Source: Intel/Micron/IEDM)

M cross-section of 3D-NAND stack  (Source: Intel/Micron/IEDM)

(Source: Intel/Micron/IEDM)

An image of the full stack is shown on the right; I see 38 wordline layers, plus a thick polySi layer at top and bottom of the stack, presumably for the drain and source select transistors. There are two tungsten metal layers below the stack for the decoders and sense-amps, and also the wordline drivers; and it looks like the M3 bitline is also tungsten. There is another metal level above used for power busses and global interconnects, but we don’t know if that is copper or aluminum.

Putting the wordline drivers under the array is claimed to keep the wordlines short, but it raises some questions – how are the wordlines contacted from below? Do we have the sort of staircase at the ends of the wordlines that we saw in the Samsung V-NAND, and could it be inverted? (Can’t imagine that!)

The vertical channels contact what looks like a polySi sourceline at the base of the stack; it’s a bit clearer in this schematic:

Schematic of base of 3D-NAND stack  (Source: Intel/Micron/IEDM)

Schematic of base of 3D-NAND stack (Source: Intel/Micron/IEDM)

While the NAND cells are floating gate cells, we can see that the source and drain select devices are single gate oxide transistors.

The larger size of the cell improves the performance since it has a higher cell capacitance – more electrons can be stored, and a better natural Vt distribution (~50%) is achieved. (Note that at 20nm planar, less than 10 electrons gave 100mv Vt shift!)

Number of electrons/100mV Vt shift (left), and Vt distribution vs 20-nm planar flash  (Source: Intel/Micron/IEDM)

Number of electrons/100mV Vt shift (left), and Vt distribution vs 20-nm planar flash (Source: Intel/Micron/IEDM)

The cell geometry also means that the cell/cell interference is reduced – again, comparing to the 20nm planar chip;

Cell/cell interference of 3D-NAND vs planar NAND  (Source: Intel/Micron/IEDM)

Cell/cell interference of 3D-NAND vs planar NAND (Source: Intel/Micron/IEDM)

We will see what the commercial part looks like when we get our hands on one, likely in the first few months of next year. Unfortunately there are no scale bars on any of the images, so we have no feel for what the actual dimensions are; though probably not too different from the Samsung, which is classed as a 40nm device.

There are actually not too many features in common with the Samsung chip – vertical stacking with 32 active layers, and that’s about it. Otherwise, charge-trap technology vs floating-gate; polySi wordlines vs tungsten; metallization below the stack, vs none; and maybe a completely different way of accessing the wordlines.

For now, we wait and see!

A Look Ahead at IEDM 2015

By Dick James, Senior Technology Analyst, Chipworks

In the second week of December, the good and the great of the electron device world will make their usual pilgrimage to Washington D.C. for the 2015 IEEE International Electron Devices Meeting. To quote the conference website front page, IEDM is “is the world’s preeminent forum for reporting technological breakthroughs in the areas of semiconductor and electronic device technology, design, manufacturing, physics, and modeling. IEDM is the flagship conference for nanometer-scale CMOS transistor technology, advanced memory, displays, sensors, MEMS devices, novel quantum and nano-scale devices and phenomenology, optoelectronics, devices for power and energy harvesting, high-speed devices, as well as process technology and device modeling and simulation.”

That’s a pretty broad range of topics, but from my perspective at Chipworks, focused on the analysis of chips that have made it to production, it’s the conference where companies strut their technology, and post some of the research that may make it into real product in the next few years. Though these days process papers actually tend to be after the launch of the relevant product, such is the preoccupation with trade secrecy.

In the last few weeks I’ve gone through the advance program, and here’s my look at what’s coming up, in more or less chronological order. As usual there are overlapping sessions with interesting papers in parallel slots, but we’ll take the decision as to which to attend on the conference floor.


Again this year the conference starts on the Saturday afternoon, with a set of six 90-minute tutorials on a range of leading-edge topics:

  • Electronic Control Systems for Quantum Computation, David DiVincenzo, Aachen University
  • Advanced CMOS Device Physics for 7nm and Beyond, Scott Thompson, University of Florida
  • Thin Film Transistors for Displays and MoreTom Jackson, Penn State University
  • Nanoscale III-V Compound Semiconductor MOSFETs for Logic, Luca Selmi, University of Udine
  • RF and Analog Device TechnologiesAnthony Chou, GlobalFoundries
  • Implantable MEMS and Microsystems for Neural Interface, Eusik Yoon, University of Michigan

The first three are from 2.45 – 4.15, and the remainder from 4.30 – 6.00. This year I hope to make it to Scott Thompson’s session on 7 nm and below, and possibly the GF tutorial on RF/analog at 4.30.

On Sunday December 14th, we start with the short courses, “Emerging CMOS Technology at 5 nm and Beyond” and “Memory Technologies for Future Systems”.

Last year the process short course was “Challenges of 7nm CMOS Technology”, so I guess we’ve moved on a node; though I still need convincing that the 10-nm process architectures are locked down as yet – the launches seem to be sliding a bit, to the back end of 2017, based on the quarterly stock analyst meetings that I’ve perused.

The 7-nm course has been organized by Yuan Taur, UCal San Diego, winner of the J.J. Ebers Award three years ago. He introduces it bright and early, at 8.30 a.m.

The first session is a four-hander on Device Options and Tradeoffs, with Mark Lundstrom and Xingshu Sun, Purdue University, Dimitri Antoniadis, MIT and Shaloo Rakheja, NYU presenting – we’ll see how they sub-divide the topics, but this is the time to get the low-down on the I-V Theory of Nanotransistors, III-V MOSFETs, Nanowire FETs, Band-to-Band Tunnel FETs, and 2D channel Materials.

Second up is Bruce Doris of IBM, discussing Process Integration Challenges and to follow, his IBM colleague Takeshi Nogami goes into more detail about BEOL Process Challenges . Starting the afternoon Krishna Saraswat (Stanford University) looks at Emerging Interconnect Technologies, then we have Tony Yen from TSMC reviewing Advanced Lithography (maybe we’ll get a hint of when EUV will become real), and the last talk of the day is by Asen Asenov (Glasgow University/Gold Standard Simulations) on Variability and Design for Manufacturability.

It now seems that 10nm, and probably 7nm, will be silicon-based, so we’ll see what the guys predict for 5nm; new channel materials, nanowire transistors, and how will they integrate into a manufacturable process? What will be the effects on the performance of the basic logic blocks? What will device reliability be like with the potential new materials/structures? Hopefully we’ll find out here!

The era of big data, and the big systems that will result from the internet of things, will put huge demands on the associated memory systems, so a memory review and look-ahead is appropriate; Dirk Wouters of Aachen University has organized the parallel course on our Sunday.

We start at a more civilized 9.15, and at 9.30 Rob Aitken of ARM is up with a look at the System Requirements for Memories, setting the context for the subsequent sessions. The next two talks review conventional memories; DRAM by Changyeol Lee, from SK Hynix, and Flash by Youngwoo Park of Samsung Electronics, with lunch in between.

Emerging memory is split into ReRAM and PCM, with Daniel Ielmini, (Politecnico di Milano) instructing, and STT-MRAM, from Thibaut Devolder, a CNRS Research Associate, Universite Paris-Sud.

I would call both courses a full day, seeing as we finish at ~5.30 p.m., but it’s worth sticking around to the end.

If you have the stamina, at 6.00 Leti is hosting a Devices Workshop at the Churchill Hotel, across the street from the Hilton. 


Monday morning we have the plenary session, with three pertinent talks on the challenges of contemporary electronics:

  • Moore’s Law at 50: Are we planning For retirement?, by Greg Yeric, ARM
  • Quantum Computing in Si, by Michelle Simmons, University of New South Wales
  • Silicon for Prevention, Cure and Care: A Technology Toolbox of Wearables at the Dawn of a New Health System, by Chris Van Hoof, Imec

In keeping with IEDM’s tradition of intellectual overload, after lunch we have eight parallel sessions!

Session 2 starts a track on Nano Device Technology, in this case with papers on Ge and other Group IV Devices; six presentations, including an invited paper (2.4) by Y-C Yeo (et al.) of TSMC on “Germanium-based Transistors for Future High Performance and Low Power Logic Applications”.

Session 3 gets us into the Memory Technologies track, discussing PCRAM and Flash, with the first three papers on 3D-NAND, two on PCRAM, and one on integrated one-time programmable memory (OTP). The one most likely to draw a crowd is an invited talk (3.3) by Intel/Micron about their floating-gate 3D-NAND flash (sampling at the moment, judging by their last quarterly financial call).

Samsung started shipping their V-NAND last year, but that uses charge-trap storage, in which the electrons that make up the memory bits sit on a silicon nitride layer; the Intel/Micron device uses the conventional floating-gate method used in planar flash, where the electrons are stored on a polysilicon floating gate. It’ll be interesting to see the difference!

Plan-view TEM images of Samsung V-NAND flash array

Plan-view TEM images of Samsung V-NAND flash array

Macronix is getting into the 3D-NAND game too (3.2), but their device uses a single-gate, flat-cell thin film transistor with an ultra-thin body that they have dubbed “single-gate vertical channel” (SGVC).

The gates are horizontal, defined by layer thickness, and the channels are vertical polySi stripes; this seems to be a charge-trap device, and the theory seems to be that “the design is not as sensitive to CD variation and is said to have potentially more than four times the memory density of GAA vertical channels at the same scaling node.”

Schematic of Macronix 3D-NAND flash cells (paper 3.2)

Schematic of Macronix 3D-NAND flash cells (paper 3.2)

The other 3D-NAND paper is from imec (3.1), replacing the polySi channel with InGaAs.

The two PCRAM talks (3.5, 3.6) discuss different phase-change materials, GaSb-Ge and ALD Ge-Sb-Te; and the OTP paper (3.4) from UMC/National Chiao Tung University claims a “Newly found Dielectric Fuse Breakdown” that gives “a smallest memory cell array which can be easily integrated into state-of-the-art advanced CMOS technology”.

In session 4, we take a look at Circuit Device Interactions – this is a Focus Session on Beyond von Neumann Computing, with nine presentations discussing how neuromorphic, brain-inspired computing can be implemented in our electronic world.

Session 5 starts the Modeling and Simulation track, reviewing Physical Modeling for Advanced Devices, Power Devices, and Memories, which covers off a pretty broad swath of technology, from avalanche breakdown in diamond (5.2), through GaN-on-silicon (5.3), to RRAM (5.4) and 3D-NAND (5.5, 5.6).

Integrated Thin Film Transistors (TFTs) are discussed in session 6, as part of the Display and Imaging Systems track. I didn’t know that polysilicon nanowires counted as TFTs, but we have two papers (6.1, 6.3) on that topic, and 6.2 is a study of a finTFT. We get into more conventional areas with an invited talk (6.4) on “TFT Backplane Technologies For Advanced Array Applications”, then we have three presentations on oxide-based TFTs to finish up the session.

Reliability and Characterization of Resistive RAM and BEOL Processes is the topic in session 7. In 7.1 TSMC discusses the use of on-chip charge collectors formed from antenna-coupled floating gates, used to study the actual potential on transistor gates during plasma charging stress.

SK Hynix participates in a study (7.2) on copper diffusion in through-silicon vias (TSVs), which is timely now that their High-Bandwidth Memory (HBM) has just appeared on the market.


Optical cross-cection of SK Hynix HBM stack

Optical cross-cection of SK Hynix HBM stack

Wafer Level Chip Scale Package (WLCSP) stress on high precision mixed-signal ICs is discussed by NXP in 7.3, again timely given the number of these we are seeing in mobile phones these days.

TSMC reviews TDDB lifetime models for back-end structures (7.4), and the last three papers look at RRAM reliability.

Session 8 is the first of the Process and Manufacturing Technology track, on 3D Integration and BEOL. We have a study of a self-forming Ta-Mn-O copper barrier layer from IBM/GLOBALFOUNDRIES (GF) in 8.1, which may give us a clue as to what we might see in their 10-nm back end; followed by Xilinx/Tohoku U (8.2) looking at wafer/wafer bonding of over 100,000 3-µm electrodes.

Disco Corp impressed me a few years ago by showing that they could thin wafers down to 7 µm thick – now they are claiming 2.6 µm in 8.3. They have looked at thinning effects on DRAM performance; 5.6 µm is OK, 2.6 µm is not.

TSMC also examine performance effects with thinned wafers and backside through-vias (8.4), and STMicroelectronics contributes to the next two papers, an invited talk (8.5) on 3D integration, and at the transistor end of the scale, a low-k spacer aimed at FDSOI technology (8.6).

The last two presentations are on monolithic 3D integration, where front-end layers are stacked rather than completed wafers, an intriguing concept, but with significant thermal challenges.

Session 9 reviews Advanced Compound RF and Power Devices; I could go through these in detail, but for the sake of brevity I’ll just say that five of the six papers discuss GaN devices, and the exception is about a vanadium oxide-on sapphire RF switch with a record switching cut-off frequency of 26.5THz (9.3).

Then in the evening we have the conference reception at 6.30, through until 8 pm. This year is the International Year of Light, plus the 50th Anniversary of Moore’s Law, so IEDM is celebrating with a special laser light show – should be fun!



In the morning we have another seven parallel sessions, starting at 9 am, with session 10 on RRAM in the Memory Technology track. The first paper (10.1, from Micron) is an invited discussion of “Non Volatile Memory Evolution and Revolution”. Judging by the authors, most of the papers are research, but in 10.5 Tsing-hua U/TSMC report an RRAM that uses a FinFET transistor for the “select” gate and an adjacent FinFET’s HfO2-based dielectric film for a storage node of the RRAM cell, in a 16nm process; and 10.7 is a consideration of RRAM use for security applications from GF.

Schematics and TEM Cross-section of finFET RRAM (10.5)

Schematics and TEM Cross-section of finFET RRAM (10.5)

Session 11 is the second Circuit Device Interaction session, this time on CMOS Scaling and Circuit/Device Variability. It kicks off with a paper by Renesas on 8T SRAM design in (presumably) TSMC’s 16-nm process (11.1), followed by a discussion from GF on self-aligned double patterned (SADP) BEOL use for a sub-10nm SRAM bitcell (11.2).

Samsung is up next (11.3), with a study on BTI variation in finFET SRAMs, then TSMC discusses magnetic thin-film inductors integrated into CMOS (11.4); Cadence gives an invited talk on the simulation of the variability of reliability of IC design (11.5); imec looks at the self-heating effects of bulk FinFETs from the 14nm to the 7nm node (11.6); and layout dependent aging in HKMG devices is detailed by Peking U. (11.7, with heavy SMIC involvement) in the last paper.

Session 12 is another on Modeling and Simulation, this time on 2D and Organic Semiconductor Devices. Black phosphorus devices are covered off in the first two papers, (12.1, 12.2), and 2-D tunnel FETs are examined in 12.312.5. Paper 12.6 is an invited talk on “Charge Transport Modelling in Organic Semiconductors” from the University of Rome, and the last paper (12.7) discusses the nature of metal-graphene contacts.

We move into the medical arena in session 13, a focus session on Silicon-based Nano-Devices for Detection of Biomolecules and Cell Function, with six invited papers on bio-analysis and measurement.

Flash and Novel Device Characterization and Reliability is the subject of session 14, starting with an imec paper (14.1) on Scanning Spreading Resistance Microscopy (SSRM) of a finFET, using diamond-based probe tips to scrape off material as the surface is repeatedly scanned to create a 3-D resistance profile.

Tunnel FETs are studied in the next two talks (14.2, 14.3), then we have two Si nanowire papers, again from imec (14.4, 14.5). U Tokyo details work on the reliability of Ge gate stacks in 14.6, and NAND flash reliability is the topic of 14.7 and 14.8.
Session 15 is the second in the Process and Manufacturing Technology track, this time on Moore and More; the first three presentations (15.115.4) discuss Ge and III-V devices, then we have an invited talk (15.5) on nanocarbon interconnects. Paper 15.6 reports on a photonic BiCMOS process, and in the last paper we hear about thin RF-SOI CMOS on flexible substrates (15.7).

Diamond-shaped Ge nanowire (paper 15.4)

Diamond-shaped Ge nanowire (paper 15.4)

30µm-thick RF-SOI CMOS circuits laminated on a flexible substrate (paper 15.7)

30µm-thick RF-SOI CMOS circuits laminated on a flexible substrate (paper 15.7)

The speaker at the conference lunch will be Pat Tang, VP of Product Integrity, Amazon, presenting on “Working backwards from the Customer to Physics of Failure in Consumer Electronics Reliability”. This talk will examine a product integrity vision based on 3 technical strategies:

  1. Working backwards from the customer to physics of failure
  2. Design for reliability through simulation tools;
  3. Development of customer-use centric standards using stress-strength analysis.

Pat joined Amazon in 2010 to lead Product Integrity and is responsible for the architectural integrity and product reliability of Amazon’s Kindle Fire tablets, e-readers, Fire TV and Fire Phone products. He was at Apple before that, where he was the reliability manager responsible for the qualification of Mac products such as the Macbook Pro, Macbook Air, iMac, MacPro, AppleTV and the first prototypes of iPad, so familiar with many of the products that have changed our lives in the last decade.

Session 16 is a focus session on Advances in Wide Bandgap Power Devices, part of the Power and Compound Semiconductor Devices track. As such we have eight invited talks, beginning with a review of “State-of-the-Art GaN Vertical Power Devices” (16.1) from Toyota.

Then imec looks at “200mm GaN-on-Si Epitaxy and e-mode Device Technology” (16.2), followed by Intel (16.3) discussing high-k GaN MOS-HEMTs. Paper 16.4 also reviews GaN power devices, and in 16.5 CEA-Leti steps back a bit to examine GaN epi on silicon and packaging topologies in the context of power electronics.

In 16.6 MIT details recent work on GaN-only normally-off transistors, HRL Labs talks about increasing the switching frequency of GaN HFET converters in 16.7; and the session finishes with STMicroelectronics giving a (presumably) more commercial survey of “SiC- and GaN-based Power Devices: Technologies, Products and Applications” (16.8).

Session 17 looks at Neuromorphic Computing Techniques in the circuit/device interaction track; definitely at the academic end of the scale for me, although seven of the eight papers use phase-change or resistive memory as synapses, so if this style of computing takes off that seems to bode well for emerging memory.

M/NEMS Resonators, Sensors and Actuators are considered in Session 18, covering sensor arrays (18.2), nanowires (18.3), transducers (18.4, 18.6), resonators (18.5), and energy harvesting (18.7).

The next parallel session is another focus session, this time on Flexible Hybrid Electronics (Session 19). Mostly research reviews, but paper 19.3 details an ultrathin Si-based flexible NAND flash memory, and 19.4 describes changing advanced CMOS electronics into a flexible and stretchable form.

Transistor Ageing, Variability and the Impact on Circuit Design is the subject of Session 20; NBTI is the topic of 20.1, off-state self-heating in scaled technologies (presumably including finFETS) is dealt with in 20.2, and more specifically self-heating in ETSOI is covered in 20.3. Paper 20.4 discusses hot-carrier aging, and “Technology Scaling and Reliability: Challenges and Opportunities” is an invited review in 20.5.

Samsung surprised me with the launch of their 14-nm finFET chip in this year’s Galaxy phone, and we get a peek at the “the extensive 14nm FinFET reliability characterization work” carried out by them in paper 20.6.

Random telegraph noise vs timing data investigated in a 32nm test chip in 20.7, and Rob Aitken of ARM gives an invited talk on “Implications of Variability on Resilient Design” in the last talk of the session (20.8).

Session 21 is the third in the Process and Manufacturing Technology track, on Advanced Modules and FinFET Devices. This looked like it might be the session where we get key papers on new production processes (such as the Samsung 14-nm), but I think we’re out of luck. Having said that, we do have some interesting papers on the program.

IBM starts off with a discussion (21.1) on “Understanding and Mitigating High-k Induced Device Width and Length Dependencies for FinFET Replacement Metal Gate Technology”. In the abstract it says that the gate fill metal is important to obtain a flat Vt-W response; given that finFET gate width is quantized, I’m curious to see what they have to report.

GLOBALFOUNDRIES gives an invited paper “Variation Improvement for Manufacturable FINFET Technology” (21.2), though it considers a 90-nm CPP process, not the Samsung-based 14-nm process now ramping up, which has a CPP of 78 nm. Gate stack, junction, and gate height variation were identified as key contributors to threshold voltage variation; I think we’ve seen that demonstrated before, but GF’s take will be of interest.

In paper 21.3 Tsing Hua U. reports on a bi-layer stacked gate dielectric, claiming improved drive current due to enhanced carrier mobility, resulting from less remote scattering caused by fewer charged oxygen vacancies.

Tokyo U. tries to clarify the SiO/HfO interaction in the common HKMG gate stack in 21.4, and in 21.5 they study “Preferential Oxidation of Si in SiGe for Shaping Ge-rich SiGe Gate Stacks”, which uses the different oxidation kinetics of Si vs Ge to optimize the gate stack.

Paper 21.6 is an imec study of Ge nFETs, looking at Si surface passivation and La band engineering; and in a joint work with Samsung (21.7) we hear about the contact resistance of Si:P source/drain epi with Ge pre-amorphization and Ti silicidation.

The last talk (21.8) is another imec joint paper, this time with ASM, detailing the use of phospho-silicate glass (PSG) to dope source/drain extensions. Last year we heard about Intel using PSG for solid-phase doping the base of the fin to reduce punch-through – now another example of this decades-old doping technique being re-used!

We go from finFETs to Steep Slope Transistors in session 22, with the first four papers discussing tunnel-FETs (TFETs), starting with an invited review by Intel (22.1). SMIC tells us in 22.2 that they’ve integrated TFETs into one of their foundry processes, and Forschungszentrum Jülich and MIT report on their TFET work in 22.3 and 22.4.

Papers 22.5, 22.6 detail studies on HfZrOx FETs, and the last presentation (22.7) describes a new type of SOI FET with a super-steep subthreshold slope of less than 6 mV/decade.

That brings us to the end of the afternoon, and Applied Materials is again hosting a panel on “The Changing Face of Non-Volatile Memory” at the Omni Shoreham Hotel on Calvert Street NW from 6.15 – 7.40 pm. They usually cater us well, so once we’re sated from the hospitality we can wander back to the Hilton for the conference evening panels:

Is there a potential for a revolution in on-chip interconnect?”, and;

Emerging Devices – Will they solve the bottlenecks of CMOS?”

This year we are back to two panels, but the first is portrayed as more of an election, so we have the following candidates:

  • The incumbent: Rod Augur, GlobalFoundries
  • We can design around it: John Wilson, nVidia
  • Nano/novel materials or devices to the rescue: Azad Naeemi, Georgia Tech
  • Active Interconnect: Toshi Sakamoto, NEC
  • Monolithic 3D: Maude Vinot, CEA
  • Multilithic 3D: Paul Enquist, Ziptronix

Moderated by Paul Franzon of North Carolina State University.

The parallel panel has Heike Riel, IBM Research as moderator, with Supriyo Bandyopadhya (Virginia Commonwealth University), Wilfried Haensch, also of IBM Research, Adrian Ionescu, EPFL, Carlo Reita, CEA-LETI, Sayeef Salahudin, UC Berkeley, and Frank Schwierz, Technical University of Illmenau as the distinguished panelists.


Wednesday morning has sessions 25 – 31; S25 is another Circuit Device Interaction session, with the intriguing title of “More than Moore – Value Added Technologies”. As a session, it’s bracketed by Toshiba, with two papers – the first, 25.1, reports on the use of magnetic tunnel junction (MTJ) memory for L2 and L3 cache memory in a CPU. They claim that with this technique, CPU power and chip area can be reduced 65 % and 37 % compared to conventional SRAM based CPU.

The second Toshiba talk (25.8) describes multi-gate oxide, dual work-function (MGO- DWF)-MOSFETs, with asymmetric LV-source/HV-drain junctions. This structure is reported to have an FMAX of 150 GHz, making it a contender for a low-power RF power amplifier.

We stay with RF in 25.2, but in a completely different context, as TSMC plugs their InFO fan-out packaging scheme in “High Performance Passive Devices for Millimeter Wave System Integration on Integrated Fan-Out (InFO) Wafer Level Packaging Technology”. There has been a lot of press gossip about the possible use of InFO for next year’s Apple A10 processor, so it’s a different spin to see InFO used down at the RF end of a phone.

25.3 is also an RF paper, this time discussing a reconfigurable 3/5 GHz, 0.13 μm CMOS low-noise amplifier, flip-chip integrated with a four-terminal phase-change RF switch. Intel is co-authoring this with Carnegie Mellon U. and John Hopkins U., so could this be using their EMIB co-packaging technology?

TSV-free monolithic 3D-IC stacking is the subject of 25.4, layering ultra-thin body devices and using CO2 far-infrared laser annealing (CO2-FIR-LA) technology for dopant activation. This is claimed to avoid device degradation, and a test chip with logic circuits, 6T SRAM, ReRAM, sense amplifiers, analog amplifiers, and gas sensors was fabricated to demonstrate the practicality of the scheme.

Next we have an invited review (25.5) of “New Devices for Internet of Things: A Circuit Level Perspective”; IoT had to show up somewhere!

If you know what Physically Unclonable Functions (PUFs) are, then paper 25.6 may be of interest – a “Robust and Compact Key Generator Using Physically Unclonable Function Based on Logic-Transistor-Compatible Poly-Crystalline-Si Channel FinFET Technology”.

IBM goes opto in the penultimate paper 25.7, looking at the integration of CMOS, RF and optoelectronic devices to enable low-cost O-band datacom transceivers.

In S26 we get back to memory. The first four papers, 26.126.4 detail a variety of STT-MRAMs, a 40-nm macro in 26.1, and a 28-nm macro in 26.2, while 26.3 reports a double magnetic tunnel junction device; and 26.4 studies the size dependence of the thermal stability of perpendicular STT-MRAM.

Samsung discloses some of the technology in their 20-nm DRAM in 26.5; we’ll see if it agrees with what we found in the part that we looked at! They refer to the use of “honeycomb structure and air-spacer technology” as being capable of taking us into the 1x-nm generations.

Air spacer between bitline and storage node contact in Samsung 20-nm DRAM (26.5)

Air spacer between bitline and storage node contact in Samsung 20-nm DRAM (26.5)

Next up, SH Hynix has a talk on HKMG transistors in DRAM peripheral circuitry (26.6), something we have speculated about but not yet seen. The last paper, 26.7, reveals a one-transistor SRAM cell using a lateral MOS for access, and intrinsic vertical open-base bipolar structures for self-latch function.

Layered 2D Materials and Devices: From Growth to Applications is the topic of Session 27; from the look of the abstracts, we have seven invited talks which will give us a comprehensive review of the state of 2-D materials in electronic devices, predominantly graphene and molybdenum sulphide.

We go back into the world of Compact Modeling in session 28; the first two talks look at modeling zinc oxide and IGZO thin-film transistors (28.1, 28.2) used in displays. In 28.3 we get into finFETs with Asen Asenov’s, group, Gold Standard Simulations, examining the effects of the gate edge roughness and fin edge roughness.

Paper 28.4 looks at modeling graphene FETs for RF applications, 28.5 analyses STT-MRAMs, and the last paper (28.6) is an invited review, “Physics-based Compact Modeling of Charge Transport in Nanoscale Electronic Devices”.

Session 29 is an example of IEDM broadening its take on electron devices, since this session considers Devices for In Vitro Bioanalytics and In Vivo Monitoring. So for me there are new concepts such the field-effect control of ions in nanofluidic transistors (NFTs) in 29.1, ion-sensitive FETs (29.2, in this case fabricated in SOI-CMOS), and pH imaging sensors using CCDs (29.3).

We also have disposable ‘electronic microplates’ in 29.4 that use mechanically-flexible interconnects and TSVs to connect the electrodes on a CMOS biosensor to the electrodes on the electronic microplate, while keeping physical separation, and a CMOS-based “High Density Optrode-electrode Neural Probe” (29.5).

The last two papers report on “an ultra-thin (5um) implantable system using organic light emitting diodes and organic photodetectors in a reflectivity monitoring system suitable for hemodynamic measurement of the brain” (29.6), and a microbubble blood pressure sensor mounted on an acupuncture needle in 29.7.

A completely different world from chips in mobile phones, though I don’t doubt that app’s will be developed for some of them!

Advanced Imagers and Photodetectors are dealt with in session 30; Olympus is first up (30.1), with “Multi-storied Photodiode CMOS Image Sensor for Multiband Imaging with 3D Technology”, using two stacked imagers that function individually for optimized performance.

Olympus stacked image sensor with RGB light imager on top and IR sensor below (30.1)

Olympus stacked image sensor with RGB light imager on top and IR sensor below (30.1)

Panasonic reports on an organic photoconductive film sensor in 30.2, then we have an invited paper from NHK Labs (30.3), also on an organic photoconductive image sensor.

A “BSI Image Sensor with Stacked Grid Structure” is discussed by TSMC in 30.4, and 30.5 is a photo-detector paper demonstrating a GeSn multiple-quantum-well-on-Si avalanche photodiode. A “Selenium/CMOS Hybrid Digital X-ray Imager” is described in 30.6; another stacked sensor, this time with a CMOS image sensor overlaid with a chlorine-doped crystalline selenium photo-conversion layer, is detailed in 30.7.

We return to power and compound semiconductor devices, in the form of III-V: FETs, Photonics, Si Integration in session 31. The first paper, 31.1, is on “Gate-All-Around InGaAs Nanowire FETS”, by imec et al., then we have “Vertical InAs Nanowire MOSFETs on Si” from Lund University (31.2), followed by an MIT talk (31.3) on “Quantum-size Effects in sub 10-nm fin width InGaAs FinFETs”.

Lund University returns in 31.4 with “Single Suspended InGaAs Nanowire MOSFETs”; followed by an invited paper by U. Tokyo discussing “CMOS Photonics Technologies Based on Heterogeneous Integration of SiGe/Ge and III-V on Si” (31.5).

MIT is back with a report on an InGaSb p-channel FinFET in 31.6, and the last paper of the session (31.7) is from imec again, on an InGaAs TFET with claimed superior SS reliability over a MOSFET.

After the morning sessions, the IEDM Entrepreneurs Lunch features Abbie Gregg, President of Abbie Gregg, Inc. (AGI), an engineering consultancy specializing in microelectronics process analysis and the design, startup and operations of clean laboratories and manufacturing facilities.

We are back to the Nano Device Technology track – Beyond CMOS Technologies in S32 after lunch, and U. Texas/Austin has the first paper in 32.1, looking at 2D nanomaterials, which would seem ideal for flexible electronics, in this case fabricating RF transistors.

32.2 describes a transition-metal dichalcogenide (TMD) body FinFET with back-gate control, postulated as a possible candidate for 2-nm technology, and 32.3 details a single-layer CVD molybdenum disulphide FET.

The next talk (32.4) is not a device report; it describes a method for separating semiconducting carbon nanotubes (CNTs) from metallic CNTs. The last two papers have spintronics as the topic, with an invited talk on “Spintronic Majority Gates” (32.5), and a demonstration of spintronic switch prototypes that encode information in a magnetic domain wall (32.6).

Session 33 has Emerging Nanodevices and Nanoarrays as the subject, beginning with two papers on vacuum nanoelectronics. MIT has fabricated (33.1) nanoscale cold cathodes (tiny electron guns) built from arrays of nanowire (NW) field emitters, with a current density of >100 A/cm2. Each emitter (6-8nm tip diameter) sits atop a vertical silicon nanowire (10µm tall, 100-200nm in diameter). The nanowire acts as a current limiter to protect the emitter from possible damage from heating and arcing.

Top - schematic of field-emission array. Bottom left – SEM cross-sectional view of Si NW current limiter with gate oxide removed to show details; right, emitters are shown

Top – schematic of field-emission array. Bottom left – SEM cross-sectional view of Si NW current limiter with gate oxide removed to show details; right, emitters are shown

33.2 is a simulation study of proximity effects on transmission efficiency and crosstalk in field emission vacuum microelectronic devices.

Next up is a MEMS plasma generator (33.3) designed for use in liquids, followed by an invited review (33.4) of “Nanoarrays for Disease Detection via Volatolomics” – it appears that volatolomics is the profile analysis of volatile organic compounds which are by-products of metabolic and pathological processes, and are emitted from various body fluids including breath, skin, urine, blood, and others.

Flexible graphene Hall sensors are described in 33.5, then “Suspended AlGaN/GaN Membrane Devices … for Ultra-low-Power Air Quality Monitoring” (33.6) finishes the session.

Modeling of III-V and Ge Materials and Alternative CMOS Device Architecture are dealt with in session 34. Intel gives the first presentation, discussing “CMOS Performance Benchmarking of Si, InAs, GaAs, and Ge Nanowire n- and pMOSFETs” (34.1), hopefully a summary of some of the wide-ranging R&D they have been doing in the last few years.

The next paper (34.2, U. Tokyo) discusses the intrinsic properties of Ge films in terms of phonon and electronic structures, providing critical parameters for device modeling. In 34.3, IBM looks at replacement metal gate resistance in FinFETs, coming to the conclusion that TiN gate fill is better than TiN/W for highly scaled gate lengths.

Process variation effect (PVE), work function fluctuation (WKF), and random dopant fluctuation (RDF) in 10-nm high-k/metal gate gate-all-around silicon nanowire MOSFET devices are studied in 34.4 (National Chiao Tung U.); with the result that the NW device has greater immunity to RDF, while suffering from PVE and WKF.

Intel is back in 34.5, in a “Study of TFET Non-ideality Effects for Determination of Geometry and Defect Density Requirements for Sub-60mV/dec Ge TFET”; and finally, ETH Zurich discusses InAs-GaSb/Si heterojunction TFETs in 34.6.

The last session (numerically), session 35, covers GaN Material and Device Interactions. MIT/Synopsys start the session off (35.1) with “Design Space and Origin of Off-State Leakage in GaN Vertical Power Diodes”, in which dislocations were identified as the main off-state leakage mechanism for GaN vertical diodes on different substrates; the authors claim that “designed GaN vertical diodes demonstrate 2-4 orders of magnitude lower leakage current while supporting 3-5 times higher electric field, compared to GaN lateral, Si and SiC devices.”

The next paper (35.2, ON Semi et al.) reports on the correlation between the off-state vertical leakage of 650V rated GaN-on-Si power devices and the dynamic Ron.

In 35.3, HKUST demonstrates a power FET with photonic-ohmic drain (PODFET) using a HEMT-compatible process on a conventional AlGaN/GaN-on-Si power electronics platform. It appears that photons are synchronously generated with the switching channel current, and they pump electrons from deep surface/bulk traps, improving the device dynamic performance.

Imec/ON Semi have been looking at how different parts of AlGaN/GaN buffers on Si contribute to the observed current collapse in devices, comparing three different types of buffers, namely stepped buffers, low temperature AlN interlayer buffers, and superlattice buffers (35.4).

Infineon et al. (35.5) report on AlGaN/GaN MIS-HEMTs with a fluorine-passivated dielectric/AlGaN interface, causing a modification of the “native” surface donors, leading to a fundamentally different device and defect behavior; potentially a new direction for reducing VTh drifts or defect engineered devices.

We finish the session with an invited review of “Steep Subthreshold Swing TFETs: GaN/InN/GaN and Transition Metal Dichalcogenide Channels” (35.6, U. Notre Dame et al.).

Chronologically the last papers are due at 3.40 pm – by then a lot of attendees will have headed for home, especially West-coasters who want to get home today.

I will definitely be suffering from information overload and becoming brain-numb, but with 230 papers and an average of six parallel sessions at any one time, plus the offsite events, that’s not really surprising. On the other hand, where else do we go to get all this amazing stuff?

Time to unwind, maybe do a little holiday shopping, and go for an indulgent meal.


Apple Watch and ASE Start New Era in SiP

By Dick James, Senior Technology Analyst, Chipworks

Back in April the Apple watch appeared in our labs, and of course we pulled it apart to see its contents. That set us some challenges, since inside the case we have the S1 “chip” (as Jony Ive called it in the launch last year). As you can see, it occupies most of the space inside the case, so it’s a pretty large chip; normally only the likes of IBM or Nvidia make chips this large.



Actually, we knew that there had to be multiple chips inside the S1, because we did a pseudo-teardown last year, based on Apple’s promo video at the time. It turns out that the S1 is actually an assembly of chips on a dedicated printed circuit board (PCB) substrate, with over 30 chips plus many passive components. So it is more accurately described as a System-in Package (SiP).

This was confirmed when we took the S1 out of the case and x-rayed it;




And we identified many of them;



This gave us the teardown information that we needed to find what chips were used, but the S1 is so different from any of the other wearables that we have looked at, that we had to go in and see how it was put together. So cut it in two and then onto the polishing wheel, and we get an idea of what Apple’s assembly house has done for them.

Actually, we did two cross-sections along the lines shown here;



This is section P1AS2;


On the left is the Dialog PMU; in the centre is the Apple APU (APL0778), with an Elpida DRAM co-packaged; and at the right is the Sandisk 64-Gb flash, including the controller chip and a spacer die. There seemed to be a wide-spread assumption that the APL0778 would be in a Package-on-Package (PoP) stack with the memory, as in the iPhones, but here it is in a straight-forward two-die stacked package.

If we look closer, we can see that the S1 uses conventional assembly techniques, but once all the components are on the 4-layer PCB, the whole thing has been over-molded with more molding compound, and then plated with metal to give the stainless-steel looking finish. A close-up of the right edge shows what I mean;



There are two 32-Gb flash dies in a conventional package with its own substrate, which is flip-bonded onto the PCB, covered with the SiP over-molding, and the exterior is metallized, giving the silver finish.

Section P1AS1 has the Broadcom BCM4334 in the centre, and the AMS NFC booster chip on the right. At left is a co-axial RF test socket.


Again, if we look closely, we can see that underfill has been used across the whole PCB before the over-molding was performed. Another feature of note is the I-shaped EMI shielding on the right of the BCM die, molded into the SiP – this is the first time we have seen this in any sort of package. In the x-ray image above, it surrounds the BCM chip, separating it from the other components. Here we are in close-up;



In effect the complete S1 assembly has EMI shielding since (with the exception of the accelerometer/gyro) the whole thing has a metal coat, mostly copper with a skin of iron/chromium. Such a coating will also inhibit moisture ingress, a good thing since I’ve heard tales of folks showering while wearing a Watch, and wrists can get a bit sweaty anyway.

A big question for us is – who supplied such an innovative package? Press commentary has identified the provider as ASE (Advanced Semiconductor Engineering Inc.) of Taiwan; and I presented at an IMAPS wearables workshop back in June, and when I got to the Watch analysis the attendees from ASE shared a few knowing looks.

The last quarterly analyst call from ASE also included this graphic, which details quite nicely the SiP concept, and includes details such as the EMI shielding:



ASE has also had more revenue from SiP this quarter, “In terms of overall, the SiP revenue accounted for about 22% in the second quarter, up from 15 a quarter ago because of the EMS SiP product ramp up.” Interestingly, they are also running below break-even on the SiP product (In response to a question as to whether all SiP projects are losing money, or just the one; “Thank God it is. It’s only this particular project that is running below break-even. Other things are moving very nicely.”).

Given these comments, I’m inclined to believe the press on this one – ASE is the supplier.

Another nugget comes from perusing the transcript of the call – “What kind of application and what kind of customers you are working with for the new SiP projects?

Tien Wu replied, “I don’t think I’d comment specifically but I’m pretty sure you will find some new products that have come out pretty soon. Sorry.”

He also noted, “We promise each other will never come out specific customers. So I will give you a non-qualifying, non-specific answer. We are expanding the SiP coverage to the cellphone, to the tablet in that particular arena. Hopefully, we can report more revenue, more penetration.”

I take that to mean that we may well see this style of SiP in the new iPhone and iPad later this year – more fun!


The Confab – Semi Industry is Now Mature

By Dick James, Senior Technology Analyst, Chipworks

The Confab started on Tuesday last week, an industry get-together organised by Solid State Technology (part of Extension Media), which they promote it as the “Semiconductor Manufacturing & Design Industry’s Premier Conference and Networking Event”.

The conference portion started with an afternoon panel session, “Exploring the Edges of Semiconductor Technology and Business”; I had the pleasure of kicking it off with a presentation on the state of the art in the business, as seen by Chipworks, then we got together on the podium for the panel part of the discussion.

Pete Singer moderated, and the other panelists were my co-blogger Phil Garrou, and Gopal Rao, ex-Intel and now an independent consultant. Pete had a set of pre-prepared questions on where we thought the business was going, the progress of 3DIC, and what we thought the impact of IoT (internet of things) might be.

Left to right: Phil Garrou, Dick James, Gopal Rao

Left to right: Phil Garrou, Dick James, Gopal Rao

We did our best to answer these and other questions from the floor, but Phil brought up a point that resonated with me; in the major segments of our industry we’re now down to three players, and that’s a sign that those segments have probably consolidated as much as they can. In the same way the auto industry has consisted of three significant players in each continental market (three in North America, three in Europe, etc). It’s a bit of an arm-wavy argument, but I think that it’s at least arguable.

So in DRAM we have Samsung, Micron, and SK-Hynix; in flash we have Samsung, Micron, SK-Hynix, and Toshiba/Sandisk: and in leading-edge logic we have Samsung+GLOBALFOUNDRIES, Intel, and TSMC.

Subramani Kengeri from GLOBALFOUNDRIES gave a good illustration of this a couple of years ago in an ASMC keynote speech:



And Tom Caulfield (also GLOBALFOUNDRIES) followed it up at this year’s ASMC, specifically in the DRAM space:



This is a point also made by Bill McClean of ICInsights in recent years, but he continues the logic to argue that now we are a mature industry, the business will tend to follow the world economic cycle rather than the capacity-based boom/bust cycles that we have seen in the first few decades.

Which makes sense from the mile-high perspective – we have all seen the changes in the customer base from the defence and computer industries, through the PC era, to a largely consumer-driven set of products – Apple is now the largest buyer of silicon chips in the world, after all.

Bill bracketed the Confab sessions neatly by giving the final presentation – “Are IC industry cycles dead or just sleeping?” His conclusion was that they are likely sleeping, but the trigger has changed from chip-making overcapacity or shortage, to whether world GDP goes positive or negative. To support that contention, he showed the correlation between worldwide GDP and IC market growth is now better than 0.9, compared with 0.35 back in the eighties.



This trend is likely a result of the consolidation of companies that we’ve seen and will continue to see, combined with the move to fabless and fab-lite, and its consequent tighter control over Capex; and, last but not least, the lack of disruptive new entrants to build mega-fabs and add over-capacity. China has had its play, India does not seem to want to get into that end of the business, and the Russian economy doesn’t seem to be up to it.



So, while we will see periods of growth and recession as always happens, likely amplified for our business since we are now so tied to consumer cycles, hopefully we won’t see the disruptive/destructive ups and downs that old-stagers like me have seen every three – five years in the last four and a half decades.

Having said that, there will be challenges, and it’s hard to see beyond 2020. We are now in the 14nm era in logic processes, and in five years (assuming a two-three year gap between generations) we will be ramping up seven-nm and heading for five.

In DRAM, Samsung has three 1x-nm nodes in their roadmap, possibly spread over five years, and flash is already at 14 – 16nm and moving to vertical – but how long will that last? Theoretically, I guess v-NAND could shrink from its current ~40nm node down to ~15nm, with more layers stacked together.

(That gives us the prospect of multi-Terabits on a single die, and I guess server farms would likely love such a product. On the consumer side, it does make me wonder if there is actually a market for (say) a 16-TB MicroSD card. On the whole, it starts to make my brain hurt.)

Those thoughts left me leaving the Confab actually wondering where these mainstream products are going to be in the early twenties, or if the technology is going to run out of steam. I know we’ve had these thoughts before, mostly due to mis-perceived lithography limits, but now we’re getting to the point where there may not be enough atoms or electrons to do what we want to do using current techniques.

Of course the research consortia are busy looking at ways of getting past this apparent impasse, it’s just that there seem to be quite a few options and no clear winner at the moment. And all the above doesn’t even consider the possible introduction of EUV and/or 450mm wafers.

Time will tell, and I may be retired by then, but we do live in interesting times, and it’s not going to change..

Apple Watch Launch Confirms WiFi and NFC Inside

By Dick James, Senior Technology Analyst, Chipworks
Today (April 10) is the day that the Apple Watch becomes available for order, and of course we will be buying some to see what’s inside. We won’t be going for the gold Edition model, even so some of us here would like to; the Sport version should be quite good enough.
At the Apple event back on March 9 it was almost a case of last and least for the Apple Watch, after listening through the ResearchKit and new MacBook launches, and more Apple Pay demos. The Watch presentation was almost a case of déjà vu, since we got most of the details last year in the announcement last September.
The one new technical detail that I did pick up on was that the use of WiFi was confirmed – there was no mention of that last year (time 74.00 in the March 9 video). There was also much emphasis on the ability to use Apple Pay and make calls through the Watch, so we know that there are microphones in there, and it has NFC (near-field communications) capability, but we knew that after the initial launch last year.
The WiFi news was interesting to us, since we did a pseudo-teardown back then, based on Apple’s promo video, and we came to the conclusion that the Broadcom BCM4334 was in the Watch. But no mention of WiFi – what gives? I guess they just forgot, and even in the new launch it was just a passing reference.
We identified the BCM4334 from a layout image of the board inside the Watch that we took from a screen capture of the video, and the characteristic footprint of a flip-chip component.

Screen shot of PCB from Apple Watch – source: Apple film “Introducing Apple Watch”

Broadcom BCM4334 die and position on Apple Watch PCB


According to Broadcom, “The BCM4334 is a single-chip dual-band combo device supporting 802.11n, bluetooth 4.0+HS & FM receiver. It provides a complete wireless connectivity system with ultra-low power consumption for mass market smartphone devices. Using advanced design techniques and 40nm process technology to reduce active and idle power, the BCM4334 is designed to address the needs of highly mobile devices that require minimal power consumption and compact size while delivering dual-band Wi-Fi connectivity.”
So we have WiFi confirmed! In the meantime we’ve been looking at that board a little more, and we have also confirmed that the NFC and NFC booster chips used in the iPhone 6 and 6 Plus are also present.
Again, we looked at the footprints on the board – nothing quite as characteristic as the Broadcom chip, but knowing the size of the chip package and the solder ball array density gives us a good clue. And knowing the size of the BCM4334, we can work out the sizes of the other chips on the board.
In the iPhone 6 the NFC controller was a NXP 65V10, which contained the PN548 die, and an AMS AS3923 NFC power booster; so it’s at least a possibility that Apple will be using them in the Watch.
Below is the AS3923 from the iPhone, showing the 5 x 4 solder ball grid on the bottom of the part. Like the Broadcom chip, it is also a flip-chip-on-board (FCOB), so the die size will be characteristic, and while a 5 x 4 grid is certainly not unique, the combination of the two gives us reasonable confidence that a matching footprint on the Watch board indicates the presence of an AS3923.
Top and bottom images of AMS AS3923
Similarly with the NXP 65V10:
Top and bottom images of NXP 65V10 
Here we have a 7 x 7 array, but it and the die size coincide with a footprint on the PCB.
Lastly, a business contact pointed out that the motion sensing is likely done by the same Invensense sensor that was used in the iPhones, the MP67B (probably the MPU6700), and when we looked, again the size and solder pads match. We wrote about this after the iPhone analysis,and in its lowest power mode, it can draw less than 10 µA.

Top and bottom images of Invensense MP67B

Putting these three together, we see below:

PCB from Apple Watch showing Invensense, AMS, and NXP die positions

Come April 24 we will know what else is in there, as you can see that board is quite packed. In the meantime, we’ll be looking for some more recognizable components.


Samsung’s FinFETs ARE in the Galaxy S6!

By Dick James, Senior Technology Analyst, Chipworks

The much anticipated Samsung Galaxy S6 made an early appearance in our teardown labs last week,  thanks to the diligent skills of our trusted logistics guru. We got our hands on the 4G+ version, the SM-G920I, with what Samsung claims is the world’s first octa-core 64 bit operating system. There is a wide array of industry buzz surrounding this flagship smartphone, but from my process-oriented point of view the focal point has to be on the Exynos 7420 application processor.

Samsung Galaxy S6 Teardown
Galaxy S6 Motherboard

Samsung Exynos 7420 Application Processor

The Samsung Exynos 7420 application processor is reportedly fabbed in Samsung’s 14 nm FinFET process. This is what Samsung has shown so far.

Which is not exactly specific! To start with, here’s the package marking of the package-on-package:

The layout of this is quite unusual – normally the memory marking (SEC 507 etc.) is in lines of text above the APU marking (7420 etc.), not in a diagonally opposed block. Which leads me into the speculation that maybe the 7420 is out of GLOBALFOUNDRIES, rather than a Samsung fab in Korea or Texas. Could ALB be short for Albany (NY)? Is the G in the lot code short for GLOBALFOUNDRIES? That all seems rather unlikely, but if Samsung wants to switch on the volume quickly in anticipation of huge volumes for the S6, what better way than to use three fabs? They did sound very confident in their last quarterly analyst call, saying that they expect 14-nm to be 30% of the LSI line capacity by year end. And there are lots of rumours about Qualcomm using the Samsung 14-nm process.

The die photograph and the die mark confirm the use of the Exynos 7420:

The functional die size is ~78 mm2, which compares well with the ~118.3 mm^2 of the Snapdragon chip used in the Galaxy S5, and the 113 mm^2 size of the 20-nm Exynos 5433. If the 7420 was a straight shrink of the 5433, we’d expect it to be 55 – 60 mm^2, but the back-end metallization stack is reported to be similar to the 20-nm planar process, so a full 50% shrink is unlikely (and the analog regions never shrink as well as digital anyway). We’ll have to wait until we see the floorplan to see how much functionality the two parts have in common.

Our guys in the lab made their usual exceptional effort in enabling us to see what the process looks like – within a few hours of getting the phone in-house, we have a decapsulated part and a cross-sectional sample under the microscope.

The Exynos 7420 uses 11 layers of metal, as you can see from the die seal cross-section above. Now let’s look at the transistors:

And we do have finFETs! This section is parallel to the fins, and across the gates. The bottoms of the contacts approximately indicate the top edge of the fin, and we are seeing the gates wrapped over and further down the sidewalls of the fin than the contacts appear to go. We will need another section orthogonal to this one to see if we have the type of epi growth in the source-drains that Intel uses.

This makes Samsung the second in line to get finFETs into volume production; they have successfully taken their 20-nm, first-generation, gate-last, high-k, metal-gate stack and adapted it to a first generation fin structure. We will need more detailed images to see whether the fins have vertical or sloped sidewalls, and how close to the Intel model they are, but those will come in the fullness of time when we have completed our full analysis and published our report.

Meanwhile, keep an eye on the blog!

IEDM – Monday was FinFET Day

By Dick James, Senior Analyst, Chipworks

In my conference preview blog last week, I mentioned that session 3 on the Monday afternoon would be a hot session, with three finFET papers, by TSMC, Intel, and IBM. I was right – even though they were given in the Grand Ballroom, it was full.

Paper 3.1 from TSMC disclosed what looks like their 16FF+ 16nm finFET technology, advanced from the 16FF reported last year – although they don’t actually call it that in the paper. A 15% speed boost and 30% power reduction is claimed, or 40% speed gain and 60% power saving compared to the 20nm process.

Gossip in the industry has it that 16FF was not advanced enough for TSMC’s customers, so they did some transistor engineering and cranked up the performance; 16FF is not even mentioned on the website these days, and 16FF+ is now in risk production, with endorsements by Avago, Freescale, LG Electronics, MediaTek, Nvidia, Renesas and Xilinx.

The 48nm fin pitch and 90nm contacted gate pitch announced last year were maintained, as is the 1x metal pitch of 64nm. This level uses “advanced patterning scheme” – presumably self-aligned double patterning (SADP), whereas the other 80/90 nm pitch metals are done with single patterning. The low-k dielectric stack has been optimized relative to the 16FF process to give almost 10% capacitance  improvement, and  they have also added a planar high-k MIM capacitor (>15 fF/um2) for on-chip noise reduction.

At the transistor level, we have a dual-gate oxide process, replacement metal gate (gate-last), dual epitaxial raised source/drains, and tungsten local interconnect – but NO PICTURES! Lots of plots, but no transistor images, as in last year’s 16FF paper, and we were out of luck in the live presentation as well.

So we still have no idea of what the TSMC finFETs will look like. I guess that’s good for me and Chipworks, since we’ll have to wait until they actually show up in the real world sometime next year.

Intel gave a late news paper (3.7) describing their 14nm finFET (note – finFET, not trigate) process at 4.05 pm. Being late news, there were only 15 minutes for Sanjay Natarajan to describe what looks like a technology that is distinctly changed from the 22nm process. AND there were images!



As announced back in August, fin pitch is reduced to 42nm, contacted gate pitch to 70nm, and 1x metal to 52nm, and we confirmed these in our blog on the Broadwell chip that we pulled out of a Panasonic laptop. In addition to the fins, the gates and the minimum metal levels use SADP, making for complex front-end lithography.


The fins have been modified from the 22nm process to have a more vertical profile, slimmed down to 8nm wide, and Intel also claims a “novel sub-fin doping technique” using “solid-source doping to enable better optimization of punch-through stopper dopants.” Sanjay’s presentation revealed that the solid-source doping uses a doped glass; now it’s down to us to work out when and where it’s used for punch-through inhibition. Idsat is claimed to improve by 15% for NMOS and 41% for PMOS over 22nm, and Idlin by 30% for NMOS and 38% for PMOS.

Changes have also been made to the back end – low-k dielectrics are used in the first eight levels, and significantly we see the first use of air-gaps in the M4 and M6 levels (80 and 160nm pitch).  This is Intel’s SEM image from the paper:



And here’s a TEM image from our analysis:

Intel airgaps2

can see from the spacing of the gaps and the profile of the barrier layer over the copper that a patterned approach has been taken, as described in the IITC 2010 paper [1], using a mask step after the formation of the metal seal layer.

Intel likes to point out their history – this is the second generation finFET, fourth generation HKMG, and sixth generation strained silicon; will their 10nm be the third, fifth, and seventh generations?

I’m now inclined to think so, since at an Applied Materials event in the evening, when asked about the delay in the 14nm launch, Mark Bohr was heard to say “We won’t have similar problems at 10nm”. Mark does not make such comments lightly, so to me that implies two things – the 10nm process is pretty well locked down already, and it’s unlikely that there are huge structural changes from the 14nm generation. Indeed, the aggressive shrink from 22nm to 14nm puts them well on the way to the predicted 10-nm feature sizes.

Immediately after Intel’s talk IBM had their 15 minutes of IEDM advanced CMOS fame, describing their 14nm technology. This has their fourth generation embedded DRAM, but is the first-gen finFET, and the first-gen gate-last process (and I’ve lost count of the SOI generations).

IBM claims a “unique dual workfunction process applied to both NFETs and PFETs” and sub-20nm gate lengths, which will be the smallest we’ve seen if we ever get a sample. Being IBM, the intended product will be over 600 mm2 and have 15 metal levels, presumably their Power9 server chip.

Fin pitch is the same as Intel at 42nm, but contacted gate pitch is 80nm, and 1x metal is 64nm. Here the fins are completely isolated since they are on the buried oxide, so no punch-through implants are needed at the base of the fin as on a bulk silicon substrate.

We do have pictures – these are really fuzzy, but we can see the gate wrapped over the fin with slightly raised source/drains on either side, and some nice facets on the source/drain epi.

During the presentation there were (of course) no details of the work-function materials, but it was stated that two masks were used to make the dual work-function structure; so presumably two slightly different material sets for the different work-functions. Another tidbit was that the pass-gate transistors



In the e-DRAM had a different Vt than the logic transistors, but not achieved by a workfunction change.

I’d missed it, but the IBM alliance gave a paper at the VLSI conference back in June [2], where they describe a 10nm finFET process; this look likes the same process, backed off to 14nm and with the e-DRAM added.

The e-DRAM introduces some challenges in connecting the trench capacitor plate to the fin of the pass gate. In the planar 22nm version there is a polySi strap from the polySi in the trench to the SOI on the buried oxide; in the finFET design the polySi strap is still used, but it is formed as a plug on the trench fill connecting to the SOI layer before fin definition, and the plug is etched into a fin during the fin etch. The epi module has been tuned to minimise the strap resistance and therefore the effect on access time.

Cell size of the eDRAM is now 0.0174 μm2; and if the trench capacitors are coupled together without the select gates, they can provide on-chip decoupling capacitors with a value of 450 fF/um2.





In the back-end IBM has their fifteen layers of metal ranging from 1x – 40x, and the section shows that the 40x is seriously thick, to take the power needed to run a chip this size!



That made for an eventful afternoon, with a bit of a disappointment from TSMC; we’ll look forward to seeing both their finFET and the Power9 next year. Of course we have a suite of reports on the Intel Broadwell, for those who want a detailed analysis of the part!


[1]   H.J. Yoo et al., “Demonstration of a reliable high-performance and yielding Air gap interconnect process”, IITC 2010, pp. 1-3

[2]   K-I Seo et al., “A 10nm Platform Technology for Low Power and High Performance Application Featuring FINFET Devices with Multi Workfunction Gate Stack on Bulk and SOI”, VLSI Tech 14, pp. 12-13