ASML tips roadmap for EUV litho production platform
Martin van den Brink, executive vice president for products and technology told an audience of senior executives as the IMEC Technology Forum that "EUV is the cost effective successor of 193-nm lithography below 20-nm." and that ASML believes it can extend EUV down to sub-5nm.
The company took delivery of a new [power] source in May and is "ready to integrate a system" van den Brink said.
This will be the first example of ASML's NXE platform of EUV lithography machines; the NXE3100 capable of between 60 and 100 wafers per hour throughput. This will have a numerical aperture of 0.25 and should be capable of 28-nm resolution, the same as is being achieved on the EUV Advanced Development Tool installed in IMEC, but at a commercial throughput.
Van den Brink showed the conference a slide with the NXE 3300 and NXE 3350 addressing 22- and 16-nm resolution respectively, while the NXE 3XX0, with an NA of 0.4 would push down to 11-nm resolution. The target date for the shipment of this first preproduction machine, the NXE 3100, is Q2 2010.
http://www.eetimes.com/news/se...l;?articleID=217701736
Originally posted by: tommo123
i'm having trouble imagining this. not the tech but what it would be used for.
i mean 10-15, maybe 20 years ago when asked about 32nm, who would have guessed we'd have, or soon get 6-8 cores etc etc.
What will we have when 5nm is used! 128 cores with a few graphics cores powerful enough for all our gaming needs? or is that totally off?
whatever, it's soooo far away but so exciting
Originally posted by: Hacp
Originally posted by: tommo123
i'm having trouble imagining this. not the tech but what it would be used for.
i mean 10-15, maybe 20 years ago when asked about 32nm, who would have guessed we'd have, or soon get 6-8 cores etc etc.
What will we have when 5nm is used! 128 cores with a few graphics cores powerful enough for all our gaming needs? or is that totally off?
whatever, it's soooo far away but so exciting
..... 640K ought to be enough for anyone
Originally posted by: bgeh
Originally posted by: Hacp
Originally posted by: tommo123
i'm having trouble imagining this. not the tech but what it would be used for.
i mean 10-15, maybe 20 years ago when asked about 32nm, who would have guessed we'd have, or soon get 6-8 cores etc etc.
What will we have when 5nm is used! 128 cores with a few graphics cores powerful enough for all our gaming needs? or is that totally off?
whatever, it's soooo far away but so exciting
..... 640K ought to be enough for anyone
Actually I've given that some thought. For most desktops in offices worldwide, isn't a Pentium 4/AMD Athlon XP system already sufficient? Thing is I really can't see why many people would need PCs more powerful than Core 2s. Video encoding is still comparatively rare, and the only really taxing thing I can imagine most people will use is watching HD video, and nothing much else, and even that's been pretty much offloaded to chipsets.
Unless there's some novel use for PCs I'm absolutely missing right now which requires tons of power, for most normal consumers, I'd say tommo123's probably got it about right.
[This is probably blasphemy in a computer enthusiasts' forum though ]
Originally posted by: Hacp
Originally posted by: bgeh
Originally posted by: Hacp
Originally posted by: tommo123
i'm having trouble imagining this. not the tech but what it would be used for.
i mean 10-15, maybe 20 years ago when asked about 32nm, who would have guessed we'd have, or soon get 6-8 cores etc etc.
What will we have when 5nm is used! 128 cores with a few graphics cores powerful enough for all our gaming needs? or is that totally off?
whatever, it's soooo far away but so exciting
..... 640K ought to be enough for anyone
Actually I've given that some thought. For most desktops in offices worldwide, isn't a Pentium 4/AMD Athlon XP system already sufficient? Thing is I really can't see why many people would need PCs more powerful than Core 2s. Video encoding is still comparatively rare, and the only really taxing thing I can imagine most people will use is watching HD video, and nothing much else, and even that's been pretty much offloaded to chipsets.
Unless there's some novel use for PCs I'm absolutely missing right now which requires tons of power, for most normal consumers, I'd say tommo123's probably got it about right.
[This is probably blasphemy in a computer enthusiasts' forum though ]
The faster the processors, the more mainstream video encoding will be. How annoying is it to have 3 different copies of each video file so you can move them between devices? Fast encoding can solve that by just requiring one version, and then encoding the rest on the fly. There will be other uses to faster cpus that I haven't thought of. Basically, if the technology exists, people will find a way to use it.
Originally posted by: bgeh
Originally posted by: Hacp
Originally posted by: bgeh
Originally posted by: Hacp
Originally posted by: tommo123
i'm having trouble imagining this. not the tech but what it would be used for.
i mean 10-15, maybe 20 years ago when asked about 32nm, who would have guessed we'd have, or soon get 6-8 cores etc etc.
What will we have when 5nm is used! 128 cores with a few graphics cores powerful enough for all our gaming needs? or is that totally off?
whatever, it's soooo far away but so exciting
..... 640K ought to be enough for anyone
Actually I've given that some thought. For most desktops in offices worldwide, isn't a Pentium 4/AMD Athlon XP system already sufficient? Thing is I really can't see why many people would need PCs more powerful than Core 2s. Video encoding is still comparatively rare, and the only really taxing thing I can imagine most people will use is watching HD video, and nothing much else, and even that's been pretty much offloaded to chipsets.
Unless there's some novel use for PCs I'm absolutely missing right now which requires tons of power, for most normal consumers, I'd say tommo123's probably got it about right.
[This is probably blasphemy in a computer enthusiasts' forum though ]
The faster the processors, the more mainstream video encoding will be. How annoying is it to have 3 different copies of each video file so you can move them between devices? Fast encoding can solve that by just requiring one version, and then encoding the rest on the fly. There will be other uses to faster cpus that I haven't thought of. Basically, if the technology exists, people will find a way to use it.
Except that plenty of people I know with C2Ds and C2Qs don't do video encoding; granted, it is only a anecdote, but I'm trying to illustrate the point that it doesn't necessarily follow that the availability of increased processing power implies that widespread applications will come, as seen in the office PCs context (IMO)
/alright this is getting a little too off topic now, let's get back to topic
Originally posted by: Hacp
Actually I've given that some thought. For most desktops in offices worldwide, isn't a Pentium 4/AMD Athlon XP system already sufficient? Thing is I really can't see why many people would need PCs more powerful than Core 2s. Video encoding is still comparatively rare, and the only really taxing thing I can imagine most people will use is watching HD video, and nothing much else, and even that's been pretty much offloaded to chipsets.
This brings up the CPU vs bandwidth problem. For a lot of media devices like the PS3 and Xbox, we're moving more toward wireless networking which has way less bandwidth than wired. On a standard wired 100mb connection, I can regularly see 6-10MB/s transfers across my network, so those can get away with piss poor encoding and fall back on high bandwidth. On D-Link's so called "300mb" wireless N, I've never once seen it go higher than 2MB/s. This means our wireless media boxes require much stronger encoding and much higher CPU demand in order to transcode to them effectively. It's stuff like this that make us want better and faster processors. This endless lack of bandwidth and lack of hard drive space demands better and better realtime encoding/compression.The faster the processors, the more mainstream video encoding will be. How annoying is it to have 3 different copies of each video file so you can move them between devices? Fast encoding can solve that by just requiring one version, and then encoding the rest on the fly.
Originally posted by: bgeh
I'm quite interested in your thoughts though Idontcare: Do you forsee a slowdown in Moore's Law, due to economic constraints (increasing fab costs, etc, etc..), and shrinking numbers of fabs as the transition to 450mm wafers, 16nm, etc. etc? I think of it as highly likely, but it depends on how quickly China/India develop, and if they develop quickly the demand will be large enough to keep fabs profitable enough to push for more and more investment to smaller nodes.
I was originally in the the more power the better camp; but after reading a few articles at Ars (will look for them), I had a change of heart. It is quite interesting though; the speculation that Intel's branching out to SSDs to find more and more utilisation for its fabs in search of profit margins to fund future investment
Originally posted by: Idontcare
Originally posted by: DSF
In terms of silicon we're going to hit the limit pretty soon as far as process size goes.
I'd argue it's not quite like that. There is a limit to scaling process technology, it's impractical to do what it takes to shrink atoms just for the desktop market segment, but the limit we are approaching faster still is the financial limit.
The cost of developing successive process technology nodes is exquisitely prohibitive as you go below 45nm. For one the materials of choice become more and more exotic (as far as the industry is concerned) which means elevated risk which means elevated costs to quantify and characterize the risk, etc etc.
This is why you saw consolidation of R&D efforts in the form of the Crolles Alliance and the IBM Ecosystem (aka fab club) develop at the 90nm and 65nm nodes. The situation gets direr at 45nm and beyond.
Intel has the revenue stream to justify the R&D cost structure necessary to fund 22nm and 16nm node development. But does AMD and the associated IBM Ecosystem? Yes but not at a cadence of 2yrs/node...they will be forced to either throw in the towel (ala Texas Instruments) or reduce their process technology cadence to something that reduces the annual R&D commitment to something their revenue stream can cost justify.
The economic limitations will dominate process technology cadence for everyone but Intel going forward (beyond 45nm) more so than the technology challenges of scaling towards atoms.
That's not to say it isn't a challenge, the money is needed to afford the tools needed to solve those challenges. EUV at $180M per tool is a barrier to entry for developing 16nm process technology for any company whose annual sales volume is <$10B.
Originally posted by: Foxery
Also note that Moore?s Law is sometimes misquoted as "CPUs double in speed" every 2 yeas, but it's actually "CPUs double in transistors,"
And even that is an interpretation adopted some time after the seminal paper was published.
Gordon Moore wrote in his original article that "The complexity for minimum component costs has increased at a rate of roughly a factor of two per year."
What Moore was talking about is the cost structure of IC's and how the number of integrated components in an IC has a cost structure which has a minimum (too few components and the cost per IC is high due to fixed overhead costs of the business itself, too many components and the cost per IC is high due to reduced yields).
Mainstream media subsequently re-interpreted Moore's law and replaced the definition's use of "components" with similar (but not the same) words of transistors, clockspeed, or performance "doubling every X number of months" where X was 12 months, then 18, then 24 months as we go thru the decades since the paper's publication.
Originally posted by: Idontcare
You may not realize it but there is some irony to what you posted, see the first graph (on second page) of Moore's original paper from 1965.
<a target=_blank class=ftalternatingbarlinklarge href="ftp://download.intel.com/research/silicon/moorespaper.pdf"><a target=_blank class=ftalternatingbarlinklarge href="ftp://download.intel.com/research/silicon/moorespaper.pdf">ftp://download.intel.com.........oorespaper.pdf</a></a>
Moore's law is based on the number of components per integrated circuit which minimizes manufacturing costs. The minimum found in lines from the first graph on page two is the value that goes into the second graph on page three.
Performance is actually not a metric of Moore's law, but the implication that performance increases commensurately with increasing components per integrated circuit is what drives the association of performance doubling every 2 yrs (more or less) to Moore's law.
But the truth of it is, if you read Moore's paper, that we see Moore's law already predicts the end of Moore's law. Again referring to the first graph, what we can expect these lines to look like as we move forward in time is that the spacing between the lines for each successive year will get less and less (the rate of cost reduction is decreasing because production costs are rising faster with each node) and the curvature of the line will get more and more flatter as the fixed costs (R&D for the process tech, mask sets for each device, fab setup costs) of producing an IC in an advanced node are beginning to dwarf the total sales volume generated by the IC itself.
The effect this has on the data point that goes into the second graph on page three (the graph we typically think of when we think of Moore's Law) is that the error bars on the y-axis for the data points get really large to the downside (because we don't have good curvature to easily define the number of components per IC that bring about a minimum in the relative manufacturing cost in the first graph on the second page.
And as you can see the shallow tail on the first graph favors the low-component count side, meaning the y-axis error bars on the second graph will be longer to the downside and shorter to the upside, meaning the line on Moore's graph is expected to bend-over and flatten out horizontally as the x-axis marches onwards from left to right.
It's all there, since 1965, cost basis and all.
Samsung's Kim Claims No Limit to Scaling
The ever-optimistic Kinam Kim of Samsung Electronics noted that some people in industry say the limit of scaling is ~5 nm, to which he said, "I do not agree." In fact, he said he believes that there are various possible paths to overcome obstacles of silicon scaling to continue to grow the silicon industry far beyond the nanometer range. Kim, executive vice president and general manager of Samsung's Semiconductor R&D Center and Samsung Fellow, presented at IMEC's Technology Forum in Brussels, Belgium, this week.
His presentation of the fundamental laws of physics (based on Boltzmann distribution and Heisenberg's uncertainty principle) predicts a limit of scaling around 1.2 nm. In other words, he commented, there is no limit to scaling to several nanometers.
http://www.semiconductor.net/a...o_Limit_to_Scaling.php
EUV Reduced to an Engineering Problem
While the semiconductor industry continues to work out the finer processing details of the coming 22 nm devices, it is investigating the likely structures and manufacturing techniques for two generations out, the 10 nm half-pitch. In a presentation at the 10 nm session at IMEC's Technology Forum in Brussels, Belgium, last week, Kurt Ronse, IMEC's lithography department director, said extreme ultraviolet (EUV) lithography is the only viable printing method for the 22 nm half-pitch and beyond. The 193 nm immersion tools, even using every trick of the trade, including double patterning and complex optical proximity correction (OPC), will not deliver the ultimate resolution and process window needed for 22 nm and beyond.
http://www.semiconductor.net/a...ngineering_Problem.php
ASML's EUV customers: Intel and possibly TSMC
''ASML is on track to ship its first EUV tools in 2010,'' Muse said. ''We count six tools including one to each of the following: Samsung, Intel, Toshiba, Hynix, IMEC. We also understand TSMC may be interested in securing a tool, and may take the one EUV tool ASML was planning to keep in-house for development work.''
Intel is a somewhat suprising customer. The chip giant is also working with ASML rival Nikon Corp. in EUV. Nikon is building an EUV tool.
At the recent SPIE Advanced Lithography conference here, Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC) disclosed its lithography roadmap and said it is still backing maskless technology for future devices. TSMC has not committed to EUV--yet.
http://www.eetimes.com/news/se...l;?articleID=217800693
Originally posted by: bgeh
Well we're primarily (as hardware enthusiasts) concerned with the leading edge (well everything perhaps chipsets), and my question to you would be: Do you forsee a slowdown on the leading edge then? [Let me also throw in a few assumptions: Assume China/India continues its path of growth, so you still have the prospect of rapidly growing markets with increasing demand, vs. the US/Europe with relatively stable demand]
ISuppli: Gear costs to derail Moore's Law in 2014
Moore's Law, which has held as the benchmark for IC scaling for more than 40 years, will cease to drive semiconductor manufacturing after 2014, when the high cost of chip manufacturing equipment will make it economically unfeasible to do volume production of devices with feature sizes smaller than 18 nanometers, according to market research firm iSuppli Corp.
While further advances in shrinking process geometries can be achieved after the 20- to 18-nm nodes, the rising cost of chip-making equipment will relegate Moore's Law to the laboratory and alter the fundamental economics of the semiconductor industry, iSuppli (El Segundo, Calif.) predicted.
http://www.eetimes.com/news/la...ml?articleID=217900102
Intel pushes 193-nm litho down to 15-nm
In a possible breakthrough, Intel Corp. claims that it has pushed 193-nm immersion lithography down to 15-nm--at least in the lab.
http://www.eetimes.com/news/se...cleID=218100243&pgno=1
Polcari to Call for EUV Infrastructure
"Although we recognize that there are multiple technologies to meet the semiconductor industry's ever-growing lithography demand, we believe EUV can be the cost-effective solution," Polcari said. "A mask infrastructure must be available to address the growing demand for high-resolution, low-cost-of-ownership lithography."
http://www.semiconductor.net/a...EUV_Infrastructure.php
Originally posted by: bgeh
I honestly do think that the popularity of the Atom, and its ilk poses a threat to this rapid pace of innovation in the semiconductor industry, as it somewhat hints that a 'good enough' level does exist for many consumers [again, this is not a 640K argument, I'm not saying it'll be like that for all time, I'm saying that this 'good enough' level will last a lot longer than in the past, i.e. PCs get obsolete at a slower rate than before], and with the extremely low ASPs companies won't get enough cashflow to invest into newer processes.
Originally posted by: bgeh
ShawnD1: I know what an Athlon XP can do; I had a Athlon XP 1700+ as my primary PC till mid-2007, and only retired it mid-2008. It was then that I finally felt its limitations in 2008, when the Diablo III video stuttered. It did Youtube fine (approx 60-65 CPU util), but not HD Youtube though (don't think it existed at the time). Also, it seems that nVidia is working with Adobe to bring .flv video decoding over to the graphics card, which will reduce requirements greatly when it comes to CPU decoding for videos.
Originally posted by: VirtualLarry
Originally posted by: bgeh
I honestly do think that the popularity of the Atom, and its ilk poses a threat to this rapid pace of innovation in the semiconductor industry, as it somewhat hints that a 'good enough' level does exist for many consumers [again, this is not a 640K argument, I'm not saying it'll be like that for all time, I'm saying that this 'good enough' level will last a lot longer than in the past, i.e. PCs get obsolete at a slower rate than before], and with the extremely low ASPs companies won't get enough cashflow to invest into newer processes.
link
Polcari Touts EUV at SPIE Photomask
While acknowledging that maskless electron-beam and nanoimprint lithography (NIL) have their proponents, Polcari said he considers EUV as the likely most cost-effective solution at 22 nm and below, adding that the Sematech EUV program is aiming to ensure EUV's readiness by 2013.
"Regardless of all of these problems, EUV is inevitable," Polcari predicted.
http://www.semiconductor.net/a..._at_SPIE_Photomask.php
Intel Ramping 32 nm Manufacturing in Oregon
Intel's first use of immersion 193 nm lithography comes at the 32 nm generation. Bohr said immersion will continue to serve Intel's needs at the 22 nm generation.
Asked if EUV lithography will be ready for the 15 nm generation, Bohr said, "It is not looking like EUV will will be ready, at least initially, for 15 nm production." He added that the company is working on techniques to extend immersion 193 nm lithography to the 15 nm generation.
http://www.semiconductor.net/a...acturing_in_Oregon.php
Race to EUVL Still Depends on Photons
Jos Benschop, research vice president at toolmaker ASML (Veldhoven, Netherlands), did a good job of summarizing the overall status of EUVL technology. ASML, which has led the industry in its transition to EUVL, revealed roadmaps taking the technology to the 16 nm node. This and two other keynote talks by Tony Yen of Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC, Hsinchu, Taiwan) and Hyeong Soo Kim of Hynix Semiconductor (Icheon, Korea) reiterated the superior pattern fidelity of EUVL compared with 193 nm double patterning. This should come as no surprise, since the EUVL wavelength is effectively 15× shorter than 193 nm.
http://www.semiconductor.net/article/391449-Race_to_EUVL_Still_Depends_on_Photons-full.php