New 2-head-per-platten tool from Applied Materials
Applied Materials has released a new
two-head-per-platten/two-platten tool for Cu damascene flows in memory fabs;
will logic lines try it for Cu interconnect or use three platten recipes?
I like the simpler, two platen tool, however the constraint of two head-per platen adds process capability constraints. Single wafer processing has been a key element of mantaining process control, So I think for critical application/leading nodes, this might prove to be a limiter.
Couple of considerations in transfering three platen recipes to a two platen two spindle system are slurry composition and process settings. Some fabs use different slurry dilutions for bulk Cu and final polish, so a single platen process presents complications sply with two heads which may not be exactly matched. Since the two spindles may not end point at the same time, a process setting based change is limited to down force change.
I don’t know the motivation behind introducing the Reflexion GT, but I can make some general comments regarding 2 vs. 3 platens, and 1 vs. 2 heads configurations in relation to the processes and applications that they support:
Regarding 2 vs. 3 platens, it seems that Ebara has consistently managed to capture market share from AMAT, so, fundamentally, 2 platen systems for copper and barrier do work. In fact, with copper films becoming thinner and thinner, the ‘bulk’ copper CMP application seems to slowly disappear since Step – 2 slurries do just fine in getting rid of the copper in a reasonably short amount of time. This simplifies the process and should also improve COO due to elimination of the ‘bulk’ polish step.
On the other had, there are certain nuances in the Replacement Metal Gate (RMG) CMP process that necessitate the use of 3 platens. In such applications, the GT may indeed be a limiter.
Regarding 1 vs. 2 heads per platen, the motivation is clearly that of throughput. The GT has 2 conditioning arms and 2 slurry supply arms, so the process will not put any undue pressure on preserving slurry supply or generating pad asperities. The three things that I see as potential issues are: (1) Severe platen heating due to frictional heat dissipation from 2 heads on the pad, (2) Shorter pad life due to the larger duty cycle that it has to endure and (3) Highly complex slurry flow patterns that may necessitate use of abnormally high rates.
I believe one driver for this proposed technology is reducing consumables-- slurry and pad. The hope is that the tool will maybe use a bigger pad and more slurry, but at less than twice the cost while polishing two wafers. I agree with Ara that there will be a tough qualification due to the added complexity he notes regarding chemistry, thermals, and mechanics along with queuing and logistics.
The original Strasbaugh tools on which CMP was originally developed were dual wafer machines. Speedfam & Cybeq machines polished 5 and 6 wafers at a time. I just love the way history repeats itself, don't you?
All that was old is new again...eventually. Once upon a time AMAT began selling batch reactors, then changed the game with the single-wafer chamber, and now the latest is dual-wafer processing (a.k.a. the mini-est of "minibatch").
I'm not 100% sure, but I strongly suspect, that this platform could be a 300/450 bridge-tool too (not that AMAT or most other people would want to talk about that these days, but I can't help but notice).
AMAT must be making this tool with throughput as a goal compared to slurry and other savings and other thing i would agree with Ed is it might be a 300/450 bridging tool as a 450mm tool would require a pad/platen with bigger diameter & the GT platform might quickly help to put a 450mm tool in market by AMAT (although AMAT repeatedly says that it will not commit for 450mm tool development, but no one wants to loose market share) if and whenever 450mm actually becomes a reality.
Also would be interesting to see how many of the IC manufacturers would be willing to modify their existing processes to accomodate this new tool.
We're now (January 20th, 2010) past the
official ending (the 18th) of this virtual roundtable discussion, after 4695 views of 129 replies to 18 questions. I'll
edit together Interesting discussions from most of the topics into a
summary document that will be posted to the Planarization Lounge.
We'll leave the topic posting open in case there are additional comments...but they would not be included in the summary.
I like the simpler, two platen tool, however the constraint of two head-per platen adds process capability constraints. Single wafer processing has been a key element of mantaining process control, So I think for critical application/leading nodes, this might prove to be a limiter.
Couple of considerations in transfering three platen recipes to a two platen two spindle system are slurry composition and process settings. Some fabs use different slurry dilutions for bulk Cu and final polish, so a single platen process presents complications sply with two heads which may not be exactly matched. Since the two spindles may not end point at the same time, a process setting based change is limited to down force change.
I don’t know the motivation behind introducing the Reflexion GT, but I can make some general comments regarding 2 vs. 3 platens, and 1 vs. 2 heads configurations in relation to the processes and applications that they support:
Regarding 2 vs. 3 platens, it seems that Ebara has consistently managed to capture market share from AMAT, so, fundamentally, 2 platen systems for copper and barrier do work. In fact, with copper films becoming thinner and thinner, the ‘bulk’ copper CMP application seems to slowly disappear since Step – 2 slurries do just fine in getting rid of the copper in a reasonably short amount of time. This simplifies the process and should also improve COO due to elimination of the ‘bulk’ polish step.
On the other had, there are certain nuances in the Replacement Metal Gate (RMG) CMP process that necessitate the use of 3 platens. In such applications, the GT may indeed be a limiter.Regarding 1 vs. 2 heads per platen, the motivation is clearly that of throughput. The GT has 2 conditioning arms and 2 slurry supply arms, so the process will not put any undue pressure on preserving slurry supply or generating pad asperities. The three things that I see as potential issues are: (1) Severe platen heating due to frictional heat dissipation from 2 heads on the pad, (2) Shorter pad life due to the larger duty cycle that it has to endure and (3) Highly complex slurry flow patterns that may necessitate use of abnormally high rates.
All that was old is new again...eventually. Once upon a time AMAT began selling batch reactors, then changed the game with the single-wafer chamber, and now the latest is dual-wafer processing (a.k.a. the mini-est of "minibatch").
I'm not 100% sure, but I strongly suspect, that this platform could be a 300/450 bridge-tool too (not that AMAT or most other people would want to talk about that these days, but I can't help but notice).
AMAT must be making this tool with throughput as a goal compared to slurry and other savings and other thing i would agree with Ed is it might be a 300/450 bridging tool as a 450mm tool would require a pad/platen with bigger diameter & the GT platform might quickly help to put a 450mm tool in market by AMAT (although AMAT repeatedly says that it will not commit for 450mm tool development, but no one wants to loose market share) if and whenever 450mm actually becomes a reality.
Also would be interesting to see how many of the IC manufacturers would be willing to modify their existing processes to accomodate this new tool.
We're now (January 20th, 2010) past the official ending (the 18th) of this virtual roundtable discussion, after 4695 views of 129 replies to 18 questions. I'll edit together Interesting discussions from most of the topics into a summary document that will be posted to the Planarization Lounge.
We'll leave the topic posting open in case there are additional comments...but they would not be included in the summary.
Happy planarizing.