[labnetwork] Need RIE help

Sat Jul 30 16:30:09 EDT 2011

Hi,

I would like to point out that carbon from resist also plays a role in oxide etching in HBr and Cl2 plasmas. This is why a small flow of O2 is added to HBr during the overetch step in poly-Si etching. Without the O2 addition, you will not get the very high (100:1) oxide selectivity seen in HBr/O2. I believe this was first shown in detail by a group from Fujitsu in the late 80s.  

    Jim 

James (Jim) P. McVittie, Ph.D.          Sr. Research Scientist
Paul G. Allen Building                  Electrical Engineering
Stanford Nanofabrication Facility       jmcvittie at stanford.edu
Stanford University                     Office: (650) 725-3640

----- Original Message -----
From: "Abbie Gregg" <agregg at abbiegregg.com>
To: "Richard H. Morrison, Jr." <rmorrison at draper.com>, "Mac Hathaway" <Hathaway at cns.fas.harvard.edu>, "Mary Tang" <mtang at stanford.edu>
Cc: labnetwork at mtl.mit.edu, "Ciel Villawatkins" <cvillawatkins at abbiegregg.com>
Sent: Friday, July 29, 2011 12:38:01 PM
Subject: Re: [labnetwork] Need RIE help

Fluorine poisoning is really the culprit here. In the email I read, they used the RIE tool for nitride etch and then ran a resist strip process . Nitride etch will usually have F based gases. F based chemistries and O2 based chemistries (for resist strip) can not mix in the same chamber. 

Here are 2 real experience I had: 

LAM 

An early Lam tool, the Rainbow 4400 was released for poly, Wsi, nitride and W plug etch processes. However, we found that when mixing etch applications in the same chamber the oxide etch rates exponentially increase. 

-F based etch step then HBr/O2 over etch step always showed 20 to 1 selectivity of polysilicon to oxide 

-Cl2 based etch step then an HBr/O2 over etch step showed 100: 1 selectivity of polysilicon to oxide 

We also battled mixed chemistry issues in the early days at TSMC and Chartered. The early versions of these foundries would only buy a few etch tools and then use multiple processes in the same chamber. 

-Poly etch was usually, Cl2+He in the main etch and HBr/O2 in the over etch. HBr/O2 had a 100:1 selectivity of Poly to oxide and will only etch oxide for around 5 Angstroms per minute. 

-Nitride etch was usually SF6 or CF4 or C2F6 based plasma 

-W plug process was usually SF6 based plasma 

If we had to use F in the same chamber as a poly etch tool,( as in the native oxide punch through step at the beginning of any poly etch), we will only use F sources very small flows and for 10 seconds or less. Then , we will follow it with a pump/purge step within the process run so we immediately flush any F out of the chamber. 

If we did have F poisoning, we found that in addition to doing a complete chamber wet clean, we also needed to remove all parts that were not hard anodized aluminum, wet clean these, air dry, then bake in the oven for 45 minutes. After that, we needed to season the chamber using an O2 only process. 

Obviously, this process was not production worthy so we ended up dedicating chambers to specific chemistries only and never had the issue again. Later on, Lam released 2 additional tools…the 4420 which is for silicides and the 4700 which is for plugs and trenches. This ensured that we did not mix chemistries anymore. 

In fact, if we found that any customer ran a F chemistry on another chamber that would require high selectivity to oxide and then call us to escalate the issue due to oxide punch through, we would void the warranty for a few months and not support the tool. 

However, if they had some F poisoning due to back streaming Fomblin, this is easy to check and we will support the tool. 

All you need to do is disconnect the last segment of pump foreline closest to the chamber. If you can wipe a colorless, oily substance, then or small, white, sticky gunk, your oil backstreamed into the chamber. 

Mattson 

The Mattson tools dry strip tools had the same problem with F. The Mattson tools, however are NOT RIE tools, they are downstream ICP plasma tools. An O2 only process was the POR for resist for those tools. Each time we switched the applications lab tools to CF4 + O2 chemistry to strip post implant resist, we poisoned the chamber after just 2 wafer runs . That process was even predominantly O2. (Flow was 3000 sccm O2 + 10 sccm CF4) We just needed a little fluorine to punch through the cross linked layer of the photoresist that was hardened by the implant process. After running just 2 wafer runs and then switch to a pure O2 process, we were surprised that we literally punched through the 500 Angstroms of oxide underlayer. 

We only had 2 apps lab test tools during Mattson’s early days so we did not have the luxury of dedicating chambers. We would constantly change the configuration of those tools and prep them every night if these tools had to switch chemistries (F chemistries vs O2 only). 

After a F poisoning (usually done on purpose in our applications lab, so we can strip post implant resist) , we used to open the chambers, wet clean / wipe with isopropyl alcohol and acetone mix. We also remove anything that is not hard anodized aluminum, wet clean these parts, air dry and bake them in the oven. Then, we would run 6 hours of O2 and Nitrogen plasma cleans, as an added chamber conditioning step. 

As we increased the number of application lab tools, we also dedicated chemistries to specific tools and never had to go through those long nights in between experiments anymore due to F poisoning. 

Lastly, if the STS tool is the only tool to do these experiments in a university research atmosphere, this tool will need to run a lot of O2 only processes for conditioning, before any resist strip experiments can be done. or 

Do chamber wipe downs (at the very minimum) + a quick O2 plasma clean before switching to a resist strip process. 

I hope this helps. 

Ciel Villawatkins | Sr. Technical Staff 
Abbie Gregg, Inc. | 1130 East University Drive, Suite 105, Tempe, AZ 85281 
Main 480-446-8000 | Fax 480-446-8001 | US Mobile 480 735 9547 Abu Dhabi Mobile +971(0)56 772 9531 

Email cvillawatkins at abbiegregg.com | Web The MTL Mail Server has detected a possible fraud attempt from "exchange1.netvigour.com" claiming to be www.abbiegregg.com 

Abbie Gregg 

President 

Abbie Gregg, Inc. 

1130 East University Drive, Suite 105 

Tempe, Arizona 85281 

Phone 480 446-8000 x 107 

Cell 480-577-5083 

FAX 480-446-8001 

email agregg at abbiegregg.com 

website www.abbiegregg.com 

CONFIDENTIALITY NOTICE: All information contained in or attached to this email constitutes confidential information belonging to Abbie Gregg, Inc., its affiliates and subsidiaries and/or its clients. This email and any attachments are proprietary and/or confidential and are intended for business use of the addressee(s) only. All other uses or disclosures are strictly prohibited. If the reader is not the intended recipient you are hereby notified that the perusal, copying or dissemination of this email is strictly prohibited. If you have received this communication in error, please notify the sender, and delete all copies of this message and its attachments immediately . 

From: labnetwork-bounces at mtl.mit.edu [mailto:labnetwork-bounces at mtl.mit.edu] On Behalf Of Morrison, Richard H., Jr. 
Sent: Friday, July 29, 2011 7:47 AM 
To: Mac Hathaway; Mary Tang 
Cc: labnetwork at mtl.mit.edu 
Subject: Re: [labnetwork] Need RIE help 

Well that is great info, 

Here are my results so far, thanks to all whom have wrote replies: 

Removed the Teflon ring and the F peak on the plasma scope reduced by about 30%, the ring shows heavy etch damage, according to STS it is needed in the system to help confine the plasma 

We found out that the “o”-ring that seals the chamber cover is greased with fomblin grease, changing that out today, data to come. 

We will swap out the dry pump for a new one to see if we have an oil back-stream issue due to a bad seal. One stage on the dry pump uses Fomblin oil. 

All of you have suggested very long O2 cleans so we are going to try that after the grease and pump change. We are also going to increase the O2 flow to 125sccm from 50sccm, increase the pressure to 250mT and increase the power. 

Any more ideas out there? 

Rick 

From: Mac Hathaway [mailto:Hathaway at cns.fas.harvard.edu] 
Sent: Friday, July 29, 2011 9:45 AM 
To: Mary Tang 
Cc: Morrison, Richard H., Jr.; labnetwork at mtl.mit.edu 
Subject: Re: [labnetwork] Need RIE help 

Hey all, 

This is Mac, at Harvard CNS. Our STS has shown some symptoms of this due to polymer on the walls coming off in the O2 clean, introducing fluorine via that route. 

When you wrote "the F peak is clipped of in height during the O2 plasma clean?" Did you mean it was still there, but shorter? This would clearly indicate the F is still there, and the only question is where it's coming from. 

Perhaps a longer chamber clean, with a low-pressure step to spread out the cleaning plasma, before the stripping runs?... 

Keep in mind that an O2 resist strip can gum up your chamber, and cause instability in your regular etch processes if you're not careful. 

Mac 

Mary Tang wrote: 

Hi Richard -- 

A similar problem was observed and diagnosed on one of our O2 ashers by one of our industrial labmembers who was a plasma etch engineer at HP. The cause seems to have been trace amounts Fomblin pump fluid coming back into the chamber -- Fomblin is basically a polymerized freon, so there's your fluorine source. 

Mary 

-- Mary X. Tang, Ph.D. Stanford Nanofabrication Facility Paul G. Allen Room 136, Mail Code 4070 Stanford, CA  94305 (650)723-9980 mtang at stanford.edu http://snf.stanford.edu 

On 7/27/2011 4:58 AM, Morrison, Richard H., Jr. wrote: 

Hi Everyone, 

Our facility uses an STS RIE tool (100mm wafers), to etch Nitrides/Oxides/Ti and resist cleans. I have a big problem when we use O2 to clean off resist from wafers. It seems that just running O2 plasma we etch 1000A/min of SiO2. It should be impossible to etch SiO2 with O2 plasma. 

In trouble shooting we have done the following, change the Teflon wafer holder, disconnected the SF6, CF4 and CHF3 tanks and capped the lines, then pumped the machine down and just ran O2 and we still etched the SiO2 at 1000A/min. We have a plasma scope and the F peak is clipped of in height during the O2 plasma clean? 

Does anybody have any ideas on what may be going on. 

Thanks in advance, 

Rick 

Rick Morrison 

Senior Member Technical Staff 

Acting Group Leader Mems Fabrication 

Draper Laboratory 

555 Technology Square 

Cambridge, MA 02139 

617-258-3420 

_______________________________________________ labnetwork mailing list labnetwork at mtl.mit.edu https://www-mtl.mit.edu/mailman/listinfo.cgi/labnetwork 

_______________________________________________
labnetwork mailing list
labnetwork at mtl.mit.edu
https://www-mtl.mit.edu/mailman/listinfo.cgi/labnetwork