If we want to go there, the difference between experience with working in or researching a mature industry and building a nascent one is that those with actual experience in the latter don’t gauge what is full of crap and what has merits based on what’s optimal for the former. Your business conditions and what you pay attention to are going to be really different for SMIC and SMEE than they are for TSMC and ASML. It’s not like the semiconductor industry is built from a cookie cutter template. Lots of different domains and specializations and situations. One could argue the kind of geopolitical pressures China’s semis industry faces is unique from that of any other country playing in this space, and that’s going to create some pretty unique choices and developments in the process.
Chinese semiconductor industry
- Thread starter Hendrik_2000
- Start date
- Status
this is even true in a mature semiconductor sector. for example what's good for Texas Instruments is completely different than what's good for TSMC. How come TI isn't scrambling to get the latest and greatest EUV? They're still operating 150 mm regular silicon fabs on i-line... are they a stone age company or something? Of course not. The reality is that beyond leading edge logic there's a whole world out there.
No one is leaping to do ASICs and FGPAs on leading edge nodes.
I only know DARPA has been using EBL for over a decade for sensitive chip development. The fact ebeam doesn’t need photo masks allows them to cut down on the learning curves and not having to share sensitive designs with mask shops. They also work with Micron fab in Virginia for larger volume chips.
I am not certain what exactly DARPA is working on, but best guess it’s for radar or communication chips which do not need the type of resolution of leading edge logic process. I am not certain if they really need or even have produced VLSI type chips. So I will refrain or speculate.
Indeed, there are a few companies offering ebeam litho systems, there are no known use for VLSI. tsmc did look into multi beam lithography as a hedge for EUV few years ago when there were doubts if EUV would work. But when asml was able to industrialize their EUV system, they adopted it as the primary litho solution.
i only speculated ebeam could be an option. Never said it IS the option. Your own take are also highly speculative, too. I feel like you trivialize the complexity behind multipatterning. I am not aware of ArF systems with the type of performance that could make that work. The newest ASML ArF system could do it. But not aware CETC has one.
i ask if anyone know what’s going on at CETC. Rom the look of it, everyone is just speculating. perhaps one of our speculation is closer to the truth, but no one knows for sure. So, I’ll just leave it at that.
??
you are taking what I said out of context. The comment of mine you referred to is my response to the “expert” jab at me (that I’m not an expert). what I tried to convey is that what I share is based on firsthand knowledge and that I don’t need to cite articles or papers, which most are garbage anyway.
and when I share what I think, those are just opinions or my speculations…no different from yours. Is it really necessary to ridicule each other’s opinion? responses or challenges would be better served if phrased more politely instead of an outright attack.
Then I guess you went on a tangent and talked about business condition, different needs or conditions between companies. Well, I don’t disagree with you, but i don’t know what that has to do with my comment on having to cite paper and actual experience.
I disagree. Ebeam litho is not limited by capability but economics.
KrFArF multiple patterning is limited by capability. it seems most of you are only thinking about imaging/resolution. never once had I seen anyone even remotely talk about the details behind what challenges is needed to make it work. And I could understand why you all do this, it’s because patterning and resolution is all you will ever see in papers. theres more that has to be considered to make what you all think can work into reality. all equipments are not make equal. There’s more to it just wavelength of the scanner, which, again, is what you all seems to focus on.
for example, SMEE having a ArF system doesn’t mean it can make what an industrialize ASML or Nikon ArF systems could do. So, speculating using SMEE ArF and simply applying multiple patterning for 28nm, to me is more unrealistic than simply using a ebeam litho. using ebeam would be much easier to implement.
not saying I’m right, but taking into more factors into my reasoning, this is the direction I lean.
and what is your definition of ”mass production”? How many chips do you think military needs? in chip making, only a few steps requires the most advanced litho for a particular node…. For military use, ebeam can take care of that few critical layers Even if it’s really slow…the rest of the steps can be addressed with slow/old outdated KrF steppers.
As you said in your other posts, ASML scanner technology is made for large scale production and efficiency, which is not the case for low volume production, it's quite possible to make very complex chips with EBL, but I think is more inefficient than projection techniques. Even, I think "experimental" ones like a possible low volume EUVL or low volume immersion lithography, the hardware has been there for a while, maybe not enough to compete with ASML or Nikon in the commercial space, but has been there.
I'm only speculating because it's hard to understand how the PLA has stuffed their advanced weapons with chips and shown off pretty advanced weaponry. They can't go to SMIC or any commercial factory because the US would sanction the shit out of any commercial factory supplying the PLA if they found out and those equipment could be monitored. Even Naura and AMEC have to walk on eggshells not to be seen as suppliers to the PLA to avoid any disruption to their business, thus only leaving CETC to act as this semi-permeable barrier between the academic-commercial sector and the military sector. Therefore, there is a possibility that CETC has developed quite advanced chip-making equipment for low-volume production, in addition to the equipment they show for large-scale production. Again, I'm just speculating.
Man, you guys go off on a tangent so s fast that it’s hard for me to keep up with the logic behind what you say. How did it go from my comment to defend against @FairAndUnbiased jab on what an “expert” is to all your comments on “SMIC and tsmc has different needs”, “matur and advanced nodes have different needs”? I agree with that…so not sure where you guys are going with that as an attempt to push back on what I shared.
I don’t know what your intention is behind stating the obvious that has little to nothing to do with what I said. are you disagreeing what i said (if that’s the case, not sure what you said has any relevance to what I said) or just starting a new topic?
and I’ll go on step further to build on your off topic responses to my remark….every company have different needs and different priorities, but they all mostly operate/prioritizes based on profits/loss. This is what I had always said. SMIC and SMEE are not 100% subsidized by the government. I’m sure making profit is their mantra. This is true for TI as well; not buying EUV and staying on 150mm are driven by P/L.
so you are talking about bench top exposure apparatus in lab environment? yeah, you are right. I don’t doubt there such lab immersion equipment that could make a few chips here and there.
I still think to build a 28nm chip you’d still need precision litho systems, even if there’s no throughput requirement. Capability is still a must.
They could buy the laser or strip it off an older ArF machine. The stage…let’s just assume they can live with the most advanced stage available domestically today (not dual stage because it’s low volume application). But, where would they buy immersion hood and the optics? so it’s hard for me to see this happening..not logical based limited datapoints that I’m working with. I speculated incorrectly countless times before and I may be wrong on this, too.
On the impressive weaponry. is 28nm tech really needed? I mean, I thought F22 was designed and built almost two decades ago, right? (I don’t know much about military weapons). Are you saying China released weapons more superior than that? I thought J-20 and the aircraft carriers are equal to American’s?
BTW, the Americans are onto AMEC and had placed them on sanction list. their stock tanked and i lost quite a bit of money on that.
[update」 oops…I mixed up AMEC with ACM Research. Ignore my last comment.
Last edited:
- Status