Could someone confirm density limits ? I've been trying to look through various scripts (posted here, on mpw_precheck) and got this :

diff/tap (FOM)  min=0.33 max=0.57
poly            min=0.08 max=0.38
li/m1/m2/m3/m4           max=0.60 ( met_min_ca_density.lydrc )
                         max=0.80 ( tiled )
m5                       max=0.76 ( met_min_ca_density.lydrc )

But isn't there some minimum density for metal ? I don't see that checked anywhere. (cc @David Lindley)

Hi @tnt I'm on a light work week this week. I see your question and will get back to you.

@David Lindley 👋

I haven't forgotten you @tnt. I'm working with Skywater to figure out what I can share.

@David Lindley No news on this ? 🥹

Hi @tnt. We've been working on this. There is an amendment to our NDA with SWT in signoff that will allow us to disclose the data with our customers. I think by next week everything will be signed off and we can talk.

Awesome thank you !

Hi @tnt the NDA amendment has been signed so I can now discuss manufacturing required rules with you. Here are the MR density rules. I'm going to align our rules to these but ours won't be exact matches. I have seen some slight differences between Skywater's Calibre results and our Klayout ones. I'll probably stay 1.5% off these, in general. Just want to point out Skywater won't accept GDS that fails any of these rules, even by the least bit. These are hard limits.

diff minimum chip pattern density (%) using 700um x 700um window stepped at 70um >= 28%
diff maximum chip pattern density (%) using 700um x 700um window stepped at 70um <= 62%

diff maximum chip pattern density range inside seal ring (%) using 700um x 700um window stepped at 70um <= 30%
diff maximum chip pattern density range inside seal ring (2000 square micron window stepped at 200) <= 15%

poly minimum chip density (%) using 700um x 700um window stepped at 70um >= 8%
poly maximum chip pattern density (%) using 700um x 700um window stepped at 70um <= 38%

li minimum global pattern density inside seal ring >= 33%
li maximum global pattern density inside seal ring <= 62%

met1 oxide area minimum chip pattern density (%) using 700um x 700um window stepped at 70um >= 80%
met1 inside seal ring maximum global pattern density <= 62%
met1 inside seal ring minimum global pattern density >= 33%

met2 oxide area minimum chip pattern density (%) using 700um x 700um window stepped at 70um >= 80%
met2 inside seal ring maximum global pattern density <= 62%
met2 inside seal ring minimum global pattern density >= 33%

met3 oxide area minimum chip pattern density (%) using 700um x 700um window stepped at 70um >= 80%
met3 inside seal ring maximum global pattern density <= 62%
met3 inside seal ring minimum global pattern density >= 33%

met4 oxide area minimum chip pattern density (%) using 700um x 700um window stepped at 70um >= 80%
met4 inside seal ring maximum global pattern density <= 62%
met4 inside seal ring minimum global pattern density >= 33%

met5 oxide area minimum chip pattern density (%) using 700um x 700um window stepped at 70um >= 80%
met5 inside seal ring maximum global pattern density <= 78%
met5 inside seal ring minimum global pattern density >= 43%

The metal oxide rules are not as big a deal as they seem at first glance. I think their purpose is to make sure "some" metal is placed everywhere. There is a size action on the metal before the actual check. The rule can be met by placing minimum area metal shapes spaced 1.2um apart in empty areas. Our fill algorithm output meets this rule. I only see issues when someone has placed waffledrop.

Hi @David Lindley. Great to hear.

Only got a minute and the second is the easier question. That spec number is a delta between different window results across the die. So, if the max measured number is 55% (2000x2000), the minimum allowed is 40%.

Ok, I see thanks.

@tnt Re: metX oxide area: Like I said, I think their purpose is to make sure "some" metal is placed everywhere. There is a size action on the metal before the actual check. The rule can easily be met by placing minimum area metal shapes spaced 1.2um apart in empty areas. Our fill algorithm output meets this rule. I only see issues when someone has placed waffledrop to block fill.

Thanks, what I was asking for was more : What's "metal oxide" ? Like it's not just the Metal ? But if it's the Metal + the SiO2 ... isn't that everything ?

Personally, I think it's misnamed. I think of the "oxide" label is used to signify clear area checks. So, the clear area between metal1. That's not what this check does. Hence, I think it's misnamed.

Sorry if I'm being dense, but I'm still not sure I follow. Because if you need more than 80% of clear area in windowed check, how would you ever meet a minimum global density of 33%. But then inversly, if it's not the clear area but the actual metal : If you need more than 80% of metal in each window, then the global density will be too high.

My apologies for not being clear. Let me see if I can improve. In general I consider "oxide" checks to be the same as clear area checks. They are checking empty space between layers. That is NOT what this check is. That's why I said I thought the check was misnamed. Please forget anything about clear area or oxide regarding this check. That is just its name. This is a check to make sure there is some metal everywhere. The metal layer is input and then sized up by a lot before the density is calculated. So, it doesn't take a lot of metal to meet the spec even though the spec limit is 80%. Minimum metal area shapes spaced 1.2um apart will enable you to meet this rule. An example of a passing layout is shown below. Minimum metal1 area is 0.083um^2 (0.29um x 0.29um). If you space 0.29x0.29 metal1 shapes out 1.2um apart in X and Y you will meet the metal1 oxide area rule. If you have metal1 routes that are minimum width spaced 1.2um apart you will meet this rule. It doesn't take much metal to meet the rule, but it does take some metal everywhere.

Ok, thanks, that's more clear 👍