REGISTER | LOGIN
Breaking News
Blog

Inside iPhone7's SanDisk/Toshiba 48L 3D NAND

Page 1 / 2 Next >
View Comments: Newest First | Oldest First | Threaded View
Page 1 / 3   >   >>
wgt0823
User Rank
Author
Re: Generation Density
wgt0823   11/1/2016 9:08:12 PM
NO RATINGS
Thanks!  

ScottenJ
User Rank
Author
Re: 48L vs 15 nm
ScottenJ   11/1/2016 8:17:25 PM
NO RATINGS
Oh yeah, they aren't cheap.

resistion
User Rank
Author
Re: 48L vs 15 nm
resistion   11/1/2016 8:03:22 PM
NO RATINGS
I agree the ArFi cost is reduced but if the 3D-specific etches are so unconventional, it is hard to imagine them being cheap.

ScottenJ
User Rank
Author
Re: 48L vs 15 nm
ScottenJ   11/1/2016 7:27:38 PM
Yes but each SAQP is an ArFi mandrel mask and an ArFi cut mask and that happens multiple times. By my count Samsungs 16nm 2D NAND has 18 ArFi masks and their 48L 3D NAND has only 6 ArFi layers.

Stair step is multiple masks because they can only shrink the resist so many times before they have to remask but it is I-line or KrF masks and they are pretty cheap.

In the array formation the layers are deposited and channels are one mask and etch down through the whole stack, word line trench is also one mask and etch down through the whole stack. I don't see a whole lot of etches for 3D NAND although some are really slow.

resistion
User Rank
Author
Re: 48L vs 15 nm
resistion   11/1/2016 7:05:26 PM
NO RATINGS
I did not include staircase but that also counts with multiple masks. SAQP for NAND should just be one trim mask.

ScottenJ
User Rank
Author
Re: 48L vs 15 nm
ScottenJ   11/1/2016 6:51:43 PM
NO RATINGS
15nm NAND has more than 1 SAQP step and SAQP requires multiple ArFi exposures that are really expensive.

64 etches sounds high to me but I would have to go count them to be sure, there are multiple step etches, for example the stair step is etch, shrink resist etch multiple times but it is done on one tool in a sequence.

I have detailed flows for both 15nm 2D NAND and 32L 3D NAND that I base my analysis on. When I compare 32L 3D to 15nm 2D from Micron my cost savings for 3D comes out to 20% and that is exactly what they have said it is publically.

There is one SADP step in 3D NAND flows.

resistion
User Rank
Author
Re: 48L vs 15 nm
resistion   11/1/2016 5:44:16 PM
NO RATINGS
SAQP requires 6-7 etches but 32L actually requires 64 separate etches! Plus (some) 3D NAND has to use double patterning if not already.

ScottenJ
User Rank
Author
Re: Generation Density
ScottenJ   11/1/2016 2:33:37 PM
NO RATINGS
They published a paper at ISSCC 2016

wgt0823
User Rank
Author
Re: Generation Density
wgt0823   11/1/2016 2:17:30 PM
NO RATINGS
ScottenJ,

 

Thanks for the info.  Where did you see the disclosure?

 

wgt0823
User Rank
Author
Re: Generation Density
wgt0823   11/1/2016 2:17:25 PM
NO RATINGS
ScottenJ,

 

Thanks for the info.  Where did you see the disclosure?

 

Page 1 / 3   >   >>
Most Recent Comments
David Ashton
 
Smartalix
 
Sang-Yun
 
Sang-Yun
 
realjjj
 
DMcCunney
 
David Ashton
 
rick merritt
 
rick merritt
Like Us on Facebook
EE Times on Twitter
EE Times Twitter Feed