[labnetwork] Fwd: ALD of SiO2 and SiN with precursors 3DMAS and BTBAS

[J Provine]

Hello Dr. Diadiuk,
at stanford we have worked extensively on PEALD of SiN in the last year,
characterizing deposition both on the Fiji systems in the SNF and on a
FlexAl in a single PI laboratory. we were particularly interested in low
wet etch rate in fluorine chemistries (like HF) and electrical properties
such as leakage current and dielectric constant.

successful deposition was shown to be possible with 3DMAS, BTBAS, DCS
(dicholorsilane, which i highly recommend against, reasons below), and TSA
(trisilylamine). by far the majority of the work we have done is on 3DMAS,
which we have stocked for a long time and are familiar with since we use it
for SiO2 deposition as well. there are a couple of papers coming out soon
on this work.

some details about each that may be helpful for you
*3DMAS* - good recipes established for SiO2 and SiN. most fully
characterized precursor from this list. easily obtainable from sigma
aldrich and strem and gelleste. price is comparable to other metal amide
precursors (TDMA-Hf and TDMA-Ti). etch rate and electrical properties of
SiO2 and SiN films are both demonstrably improved over PECVD at the same
temperature. GPC ~0.1A/cycle

*BTBAS* - new precursor to us for this SiN study. not extensively studied,
but deposition definitely possible for SiO2 and SiN. GPC ~0.2A/cycle.
readily available and reasonably priced from all the typical precursor
suppliers

*TSA* - promising for SiN. we have not looked into SiO2 deposition yet. not
extensively explored. GPC ~0.25A/cycle for us, but others have reported
even higher (close to 1A/cycle) which we could not reproduce on the FlexAl.
low wet etch rate and low leakage currents, similar to what was achieved
with 3DMAS. this precursor is pretty hard to stock as we could only source
it from air liquid and it look a lot of effort/time/cost to get supply to
Stanford. not sure why that is.

*DCS* - stay away from this. hard to get in a compatible storage cylinder
for ALD (much more common to come in large volumes for distribution
throughout a facility). dangerous for sure. also, not at all clear that
this precursor provides a self limiting surface reaction, so it does not
perform ALD so much as CVD. could not achieve etch rate or electrical
properties as good as seen with 3dmas and tsa, most likely because of the
lack of self-limiting half reactions. since you are already using TiCl4,
this may not be as much of a concern, but we've stayed away from chloride
chemistries in SNF because of issues with co-deposition in the same chamber
with Pt and Ru metals by ALD.

let me comment further on your specific questions:
Does anyone have any experience w/them & with TDMAT?

we also have used TDMA-Ti for our Ti precursor since we started ALD work in
SNF. we have had great success with TiO2 and TiN deposition with this
precursor. as i mentioned above we steered clear of TiCl4 because of HCl by
products and compatibility issues since we wanted to deposition Pt and Ru
in the same chamber. is it possible this is where the concern you mentioned
about TiCl4 with metal amide precursors comes from? similar to the notes
from the folks at notre dame, i don't see a safety issue with using these
precursors on the same manifold. as an additional datapoint on this,
berkeley's nanolab had TiCl4, TMA, and TDMA-Hf all in use on their picosun
system.

Any recommendations or suggestions for precursors for SiO2 and SiN?

see above.

We'd like to assess all of these precursors and films with respect to:
1) Cost and availability of precursor

addressed above

2) Robustness of precursor  (A precursor that degrades rapidly over time
wld not be preferred)

i didn't mention this above, but we never saw any of these decay over time.
the only precursor we have stocked for a long time (meaning months to a
year) to know for sure that it has good shelf life is 3DMAS. however, TSA
at least is chemically stable and requires no heating for introduction into
the chamber, so i would surmise it is also stable.

bonus info: TDMA-Ti is quite stable on the shelf or even heated (we keep it
at 75C on the tools), and will last for months.

3) Quality of film:  Electrical data would be nice but even indices of
refraction would do

because we have papers coming out soon, i've stopped short of supplying raw
data, but gave a general level of results above. indices of refraction for
the optimized recipes are on the order of 2.0 at 655nm for 3DMAS and TSA.
closer to 1.9 at 655nm for BTBAS, and 1.8 at 655nm for DCS.

4) Any chemical compatibility issues.

nothing established in this regard. everything looks fine.

5) Existence of recipes on the Oxford ALD tool at other places.  This will
reduce the amount of optimization that we will need to do.

on our FlexAl
recipes in existence for SiN and SiO2 for 3DMAS. SiN for TSA and BTBAS, and
starting points for SiO2.

additionally a starting point recipe for SiO2 for BTBAS on a different
reactor (Cambridge Nanotech FijI)

please feel free to contact me if you would like recipe details, and i can
be in touch regarding our forthcoming papers related to this topic if you
wish.

i hope this was helpful.
dr j

On Tue, Oct 13, 2015 at 6:03 AM, Mike Young <myoung6 at nd.edu> wrote:

> Hi Vicky. See below for relevant comments from two of my esteemed
> colleagues, regarding our FlexAL and its sources.
>
> --Mike
>
> Begin forwarded message:
>
> *From: *Mark Richmond <Mark.M.Richmond.7 at nd.edu>
>
>
> Mike,
>    I was never informed of any manifold compatibility issues of precursors
> by Oxford, and Dr. Xing didn't have any objections to that either. I found
> that odd but the system is under vacuum and is purged well before and after
> recipes are executed. As you can see from Dave's screen capture we have had
> this setup for the last few years, although we do not perform many Titanium
> growths.
>
>    As for the SiO2 and SiN we use BDEAS in our recipes, Nitride growth is
> very slow though. We never purchased a BTBAS source but we did have a 3DMAS
> source. The 3DMAS source was used up in attempting to characterize the
> material and we never had a good material growth so we have not replaced
> that material. It used to be in the empty spot that is now on the machine.
>
>    Vicky should be able to contact Oxford for their information sheets on
> different film recipes per precursor or I have in the past anyway.
>
> Thanks
> Mark
>
> On Mon, Oct 12, 2015 at 2:09 PM, David Heemstra <David.Heemstra.1 at nd.edu>
> wrote:
>
>> Mike:
>> We also have two separate manifolds.  Below is a screen capture for our
>> system.
>> Maybe im missing something but - it sounds like they don't want to put
>> TiCl4 and TMA on the same manifold?  I believe that is how ours is
>> currently setup.
>> I thought we had 3DMAS or BTBAS a long time ago - but currently Si source
>> is BDEAS.
>> Possible (if we are doing something that isn't ideal) it is OK as long as
>> you use sufficient purges afterwards?
>> Not really sure that this helps.  Mark may have a better understanding.
>> Dave
>>
>>
>>
>>
>> [image: Inline image 1]
>>
>
>
> On Sun, Oct 11, 2015 at 10:26 AM, mike young <mike.young at nd.edu> wrote:
>>
>>> dave/mark, can you advise Vicky regarding this ALD question?
>>>
>>>
>>>
>>>
>>> -------- Forwarded Message -------- Subject: [labnetwork] ALD of SiO2
>>> and SiN with precursors 3DMAS and BTBAS Date: Fri, 9 Oct 2015 15:56:30
>>> -0400 From: Vicky Diadiuk <diadiuk at mit.edu> <diadiuk at mit.edu> To:
>>> labnetwork at mtl.mit.edu
>>>
>>> Hi,
>>>  We are interested in adding SiO2 and SiN deposition capabilities to at
>>> least one of our ALD systems (specifically the Oxford FlexAL).
>>> The system has two sets of independent precursor manifolds.
>>> Currently, one manifold is completely filled up with TMA, the Hf source
>>> and the W source, while the second manifold only has TiCl4 for depositing
>>> TiN and TiO2.
>>> The problem is that the TiCl4 source (and other such metal halide
>>> sources) is not compatible with metal amide sources and thus we can't put
>>> any metal amide precursors on the two open slots on the second
>>> precursor manifold.
>>>
>>> Our users are interested in depositing SiO2 and SiN by using metal amide
>>> precursors like 3DMAS and BTBAS but that requires that the TiCl4 source be
>>> replaced with a compatible Ti precursor, such as TDMAT to deposit TiN.
>>> The precursors we are considering based on what people seem to be using
>>> at other universities and a quick literature search are:
>>>
>>> *SiO2*
>>> 3DMAS/TDMAS
>>> BTBAS
>>>
>>> *SiN*
>>> 3DMAS
>>>
>>> Does anyone have any experience w/them & with TDMAT?
>>> Any recommendations or suggestions for precursors for SiO2 and SiN?
>>>
>>> We'd like to assess all of these precursors and films with respect to:
>>> 1) Cost and availability of precursor
>>> 2) Robustness of precursor  (A precursor that degrades rapidly over time
>>> wld not be preferred)
>>> 3) Quality of film:  Electrical data would be nice but even indices of
>>> refraction would do
>>> 4) Any chemical compatibility issues.
>>> 5) Existence of recipes on the Oxford ALD tool at other places.  This
>>> will reduce the amount of optimization that we will need to do.
>>>
>>> We really appreciate any comments you might have.
>>> Thx,
>>>  Vicky
>>>
>>>
>>>
>>>
>>
>
> --
> Michael P. Young                                (574) 631-3268 (office)
> Nanofabrication Specialist                      (574) 631-4393 (fax)
> Department of Electrical Engineering            (765) 637-6302 (cell)
> University of Notre Dame                         mike.young at nd.edu
> B-38 Stinson-Remick Hall
> Notre Dame, IN 46556-5637
>
>
> _______________________________________________
> labnetwork mailing list
> labnetwork at mtl.mit.edu
> https://www-mtl.mit.edu/mailman/listinfo.cgi/labnetwork
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mtl.mit.edu/pipermail/labnetwork/attachments/20151013/6d1674da/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 796272 bytes
Desc: not available
URL: <https://mtl.mit.edu/pipermail/labnetwork/attachments/20151013/6d1674da/attachment.png>