No simple answer! I guess it all depends on defectivity requirements, choice of slurry and film stack that needs to be polished. For example, for Cu CMP there seems to be a general tendency to keep the slurry and post CMP clean pH in the same or close range to avoid any drastic pH-induced changes to polished surface. But at the end of the day, whatever combination (pad/slurry + post CMP clean) that provides the best performance will be the choice.
I have heard the term "pH shock" and often wondered what the individual meant. The pH influences zeta potential and hence the slurry suspension characteristics. Rapid changes in pH can cause agglomeration. There has been data published in refereed journals showing alkaline post-CMP cleaners performing best regardless of the slurry pH - and no evidence of pH shock. Large volumes of alkaline post-CMP cleaners are sold every year. Corrosion and inhibition mechanisms are vastly different for acidic vs. alkaline conditions. However, there has been work published showing a low-K dielectric weakness exacerbated by the pH of the slurry or cleaner - namely dielectric cracking under alkaline conditions. I don't know if definitive data on that topic has been embraced by end users since they continue to routinely rely on acidic and alkaline slurries as well as both pHs for post-CMP cleaners. In fact, most cleaning BKMs consist of an acidic and an alkaline cleaner for each wafer pass, much like the old, but still used, RCA clean process.
I have seen more usage of alkaline post-CMP cleans in the recent time for the advanced Cu/low-k applications. The trend seems to be employing a slightly acidic solution in the megasonic bath and acidic or alkaline cleans in the 2 brush boxes. Some applications employ acidic and alkaline cleans in one brush box each. The 1st generation acidic cleans, as well as the newer 2nd and 3rd generation alkaline (pH ~11) cleans seem to do a good job once properly selected and fine-tuned for specific application. In general, smaller feature Cu/low-k logic devices PCMP cleaning applications are more demanding. Again, the selection of cleans and its usage recipe strongly depends on the application and the slurry properties (mainly the abrasive type, size and concentration, and the mother liquor). We recently performed an extended period soak test of new PVA brushes in different commercially available acidic and alkaline cleans dilute solutions. The only brush which remained intact without any discoloration and significant biological growth was from the alkaline clean solution. More details of this study and PCMP cleans are available in our paper in the ICPT 2009 proceedings.
Darryl, the term Ph-shock appeared pretty long time ago, when people were not wery careful (or did not understand wee chemistry) and transfered wafer directly from the acidic media (pH=2-4) into basic cleaning media (pH=10-11). This caused agglomeration: crystallites grew over the wafer surface, some of them were dislodge by the brushes producing scratches etc. Since this was observed, end-users introduced final buff&rinse steps separating the polishing chemistry from cleaning chemistry, allowing to use the best-performing cleaning approach regardless of the CMP chemistry.
Mansour has it right -- no simple answer! in addition to all of the chemical and process integration reasons, this issue is further complicated by the IP fences that have been built around certain operating spaces. Some suppliers developed products in a specific pH range for the sole motivating reason that no one else had patented in the space. I have to believe that the effectiveness of available products in the copper PCMP space would be a bit more interesting if patent fencing (i.e. claiming operating conditions for which no commercial product is offered) were not an issue.
At SCP/ATMI we looked at various pH PCMP cleaners on challenge wafers that were rinsed with DI water and dried and then sent for evaluation. When received, they were covered with slurry and agglomerates due to the DI rinse and lack of a proper clean. Routinely, alkaline cleaners performed best, which was verified at the end users' sites. In addition, it is sometimes difficult to determine where the polish process stops and the cleaning process starts since formulated products are now routinely used on platen 3 in a buff process as a pre-clean. Lastly, some end users are doing post-post-CMP cleans as well. I know of an Asian IDM who does single wafer PPCMP cleans in high volume manufacture.
Interesting that both Darryl (#2) and Rakesh (in #3) can site studies showing basic PCMP solutions seem to have inherent advantages over acidic ones. However, with the ability to add additional clean steps (pre- or post-PCMP) to two brush stages we have a lot of complexity in realworld options.
Again, it seems that we can circle back to my earlier comment that no easy answer can be found! Rakesh is right that if you look around the industry, you will find more basic or close to neutral pH commercially available pCMP cleaners and that is not to say that they are definitely better than lower pH solutions. This could be purely due to historical reasons as well as how each supplier feels more comfortable from the IP point of view. As Mike suggested, I have also heard comments by a given supplier that "we stayed away from certain pH range due to supplier X's IP". I also agree w/ Darryl's comment that boundaries between polish and clean are getting blurry. Top surface of wafer continueously evolves as it goes through multiple platens and brush boxes. So as pads and slurries evolve, so does pCMP clean solution and final pH by itself may become less of a factor. We also see several suppliers offering both flavors to have a better chance of capturing market share.
(Following Darryl and Mansour, two good topics for future discussion: the blurring boundaries between CMP and PCMP and PPCMP, and PCMP clean chemistry strategies.)
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 have heard the term "pH shock" and often wondered what the individual meant. The pH influences zeta potential and hence the slurry suspension characteristics. Rapid changes in pH can cause agglomeration. There has been data published in refereed journals showing alkaline post-CMP cleaners performing best regardless of the slurry pH - and no evidence of pH shock. Large volumes of alkaline post-CMP cleaners are sold every year. Corrosion and inhibition mechanisms are vastly different for acidic vs. alkaline conditions. However, there has been work published showing a low-K dielectric weakness exacerbated by the pH of the slurry or cleaner - namely dielectric cracking under alkaline conditions. I don't know if definitive data on that topic has been embraced by end users since they continue to routinely rely on acidic and alkaline slurries as well as both pHs for post-CMP cleaners. In fact, most cleaning BKMs consist of an acidic and an alkaline cleaner for each wafer pass, much like the old, but still used, RCA clean process.
I have seen more usage of alkaline post-CMP cleans in the recent time for the advanced Cu/low-k applications. The trend seems to be employing a slightly acidic solution in the megasonic bath and acidic or alkaline cleans in the 2 brush boxes. Some applications employ acidic and alkaline cleans in one brush box each. The 1st generation acidic cleans, as well as the newer 2nd and 3rd generation alkaline (pH ~11) cleans seem to do a good job once properly selected and fine-tuned for specific application. In general, smaller feature Cu/low-k logic devices PCMP cleaning applications are more demanding. Again, the selection of cleans and its usage recipe strongly depends on the application and the slurry properties (mainly the abrasive type, size and concentration, and the mother liquor). We recently performed an extended period soak test of new PVA brushes in different commercially available acidic and alkaline cleans dilute solutions. The only brush which remained intact without any discoloration and significant biological growth was from the alkaline clean solution. More details of this study and PCMP cleans are available in our paper in the ICPT 2009 proceedings.
Regarding #1
Mansour has it right -- no simple answer! in addition to all of the chemical and process integration reasons, this issue is further complicated by the IP fences that have been built around certain operating spaces. Some suppliers developed products in a specific pH range for the sole motivating reason that no one else had patented in the space. I have to believe that the effectiveness of available products in the copper PCMP space would be a bit more interesting if patent fencing (i.e. claiming operating conditions for which no commercial product is offered) were not an issue.Yehiel,
At SCP/ATMI we looked at various pH PCMP cleaners on challenge wafers that were rinsed with DI water and dried and then sent for evaluation. When received, they were covered with slurry and agglomerates due to the DI rinse and lack of a proper clean. Routinely, alkaline cleaners performed best, which was verified at the end users' sites. In addition, it is sometimes difficult to determine where the polish process stops and the cleaning process starts since formulated products are now routinely used on platen 3 in a buff process as a pre-clean. Lastly, some end users are doing post-post-CMP cleans as well. I know of an Asian IDM who does single wafer PPCMP cleans in high volume manufacture.
Regarding #7
Again, it seems that we can circle back to my earlier comment that no easy answer can be found! Rakesh is right that if you look around the industry, you will find more basic or close to neutral pH commercially available pCMP cleaners and that is not to say that they are definitely better than lower pH solutions. This could be purely due to historical reasons as well as how each supplier feels more comfortable from the IP point of view. As Mike suggested, I have also heard comments by a given supplier that "we stayed away from certain pH range due to supplier X's IP". I also agree w/ Darryl's comment that boundaries between polish and clean are getting blurry. Top surface of wafer continueously evolves as it goes through multiple platens and brush boxes. So as pads and slurries evolve, so does pCMP clean solution and final pH by itself may become less of a factor. We also see several suppliers offering both flavors to have a better chance of capturing market share.(Following Darryl and Mansour, two good topics for future discussion: the blurring boundaries between CMP and PCMP and PPCMP, and PCMP clean chemistry strategies.)
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.