High-k and metal gate (HKMG) transistor process flows
seem likely to include an aluminum CMP process. Are the expectations of the
pads and slurries different from interconnect and STI?
Certainly new and unique slurries are needed to get optimal polishing of aluminum, and of course Cabot Microelectronics has selective and non-selective/tunable products. We have found that conventional hard pads-- IC and D100-- work fine, but we also believe that some pad modification can significantly benefit the process.
Cliff, is there a fundamental difference in the slurry-set between early Al CMP work for dual-damscene metal (I know that IBM looked at it over 10 years ago) and current work for HKMG, since the amount of material to be removed and the final specs are rather different? Would you say that the current slurries are modifications or "extensions" of past work, or are completely new formulations needed?
Ed-- Things have come a very long way since the early IBM/Al days. To a certain extent, everything today is an extension of things we have done in the past. Our core aluminum platforms-- selective and non-selective and buff-- are each unique compositions with new combinations of particles and chemistries. And they will certainly need to be tuned for the various customer system requirements-- line spacings, thickness, dielectric, tool and pad-- as usual.
The fundamental differences among HKMG, STI & interconnect include a growing sensitivity to defects such as micro-scratches that weren't even detectable in the dark days. The specific defect modes, tolerance levels for each, and associated chemical (metal) contaminants are such a strong function of exactly where you are in the device and how easily those defects can affect the gate junction. These fundamentals will, in turn, drive the details of pad and slurry differences for these processes.
So, given such basics, can we reasonably consider that (compared to established STI and DD specs.) CMP for HKMG must add additional constraints such as:
* Much tighter chemical contamination limits,
* Much tighter on mechanical defects,
* Novel post-CMP cleaning needed.
Yes/No/Maybe? If yes, would anyone care to share any ideas on the likely numbers behind specs?
In addition to the ever tighter specs on composition, it is my understanding that the precision of Al polish in HKMG is quite demanding since the very small size of the gate, and of course its electrical function means much sharper dependence of current/voltage/capacitance/breakdown on dimensions and topography.
Thanks for all the replies. We're now (January 20th, 2010) past the official ending (the 18th) of this virtual roundtable discussion. 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.
Ed-- Things have come a very long way since the early IBM/Al days. To a certain extent, everything today is an extension of things we have done in the past. Our core aluminum platforms-- selective and non-selective and buff-- are each unique compositions with new combinations of particles and chemistries. And they will certainly need to be tuned for the various customer system requirements-- line spacings, thickness, dielectric, tool and pad-- as usual.
Regarding #2
The fundamental differences among HKMG, STI & interconnect include a growing sensitivity to defects such as micro-scratches that weren't even detectable in the dark days. The specific defect modes, tolerance levels for each, and associated chemical (metal) contaminants are such a strong function of exactly where you are in the device and how easily those defects can affect the gate junction. These fundamentals will, in turn, drive the details of pad and slurry differences for these processes.
So, given such basics, can we reasonably consider that (compared to established STI and DD specs.) CMP for HKMG must add additional constraints such as:
* Much tighter chemical contamination limits,
* Much tighter on mechanical defects,
* Novel post-CMP cleaning needed.
Yes/No/Maybe? If yes, would anyone care to share any ideas on the likely numbers behind specs?
Ed et al.-
In addition to the ever tighter specs on composition, it is my understanding that the precision of Al polish in HKMG is quite demanding since the very small size of the gate, and of course its electrical function means much sharper dependence of current/voltage/capacitance/breakdown on dimensions and topography.
Any device physicists out there to add color?
Thanks for all the replies. We're now (January 20th, 2010) past the official ending (the 18th) of this virtual roundtable discussion. 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.
PS
As of now, 4695 views of 129 replies to 18 questions.