Comparison table for capacitance per um^2 for IC processes?

(1) Why XH-035? High voltage? (2) Seems like a waste to use up that much silicon just to get a capacitor.

What I've done in "have to" situations is use gate oxide... 5V devices have, typically, 2.5-3fF/um^2. I've used them regularly as compensation capacitors for LDO's. The only caution is to make sure the "channel" is conductive to avoid large series resistance. I usually roll my own square geometry (all drain, or source, call it what you want)... that geometry dramatically drops the effective R. ...Jim Thompson

Here's a set of subcircuits that accurately reflect how this capacitor behaves, capacitance and resistance, substitute the MOS model of your choice (P or N-type), and watch the wrap....

• NSQCAP *

• PD={4*(E+1.6u)} PS=0 NRD={0.4u/E} NRS=0 M=1 M_MN3 N_4 N_1 N_3 B N33E2R L=0.6u W={0.6u/31.41m} AD=0 AS=0 PD=0 PS=0 NRD=0 NRS=0 M=1 M_MN5 N_3 N_1 N_2 B N33E2R L=0.6u W={0.6u/42.06m} AD=0 AS=0 PD=0 PS=0 NRD=0 NRS=0 M=1 M_MN7 N_2 N_1 N_5 B N33E2R L=0.6u W={0.6u/63.86m} AD=0 AS=0 PD=0 PS=0 NRD=0 NRS=0 M=1 M_MN8 N_5 G N_5 B N33E2R L={0.346*E} W={0.346*E} AD=0 AS=0 PD=0 PS=0 NRD=0 NRS=0 M=1 M_MN9 N_5 N_1 N_6 B N33E2R L=0.6u W={0.6u/137.3m} AD=0 AS=0 PD=0 PS=0 NRD=0 NRS=0 M=1 M_MN10 N_6 G N_6 B N33E2R L={0.2*E} W={0.2*E} AD=0 AS=0 PD=0 PS=0 NRD=0 NRS=0 M=1 .ENDS NSQCAP

Because that's the process I dealt with last :-)

I meant without the HV extension, we won't need that here.

Hmm, that's less than the 4fF with the XH-035. Well, only for the poly-diff cap, not the other one.

In this case it'll all be low power. I am hoping to get by with a milliwatt or two.

What does that have to do with "low power"? ...Jim Thompson

Thanks, Jim. This post came in with a delay here for some reason. Will try on the weekend, the missus called. Dinner, and she is a really good cook. Then margaritas :-)

Enjoy ;-) ...Jim Thompson

I did some poking thru my XFAB libraries... I have them all ;-)

I don't use poly-diffusion caps because of the about 50% parasitic on one end to substrate... or one end is actually AT substrate.

Excluding that, I'm finding generally that poly-poly caps are all right around 1.87fF/um^2... except on XH-035, where the oxide thickness is larger... don't use it if you don't need HV, use XC-06 (my favorite).

I see some MIM caps at 2.5fF/um^2... but that's an extra process step, and you're cheap, cheap, cheap... ;-) I almost always have that process module, because I like RPOLYH resistors to keep the power (and space) down. ...Jim Thompson

Only if it's special "thinned" oxide. Field oxide is quite thick, comparatively. MIM caps are made from thin oxide, but require extra process steps. Multi-layered ones end up having strays to substrate which can be a pain. ...Jim Thompson

Color me stupid concerning CMOS processes; if an area can be isolated (think DI), then a substrate-iso-metal-iso-metal config with top metal tied to substrate can give a hell of a lot of capacitance per square micrometer.

Field oxide is thick because you DON'T want channels forming. That is also why you have field implants.

I'm not sure if any text book covers channel stop layout techniques that you use when you make a chip operate at higher than the specd voltage.

I only have one, and only after signing some papers that were talking about some sort of tribunal if I ever ... :-)

One of the caps goes to ground, but not the 2nd one. I could compromise some parameters and make that 2nd one a bit smaller, down to 200pF possibly.

As John Wayne once put it, man's got to do what man's got to do. 2.5fF sounds a whole lot better than 1.87fF.

Low power in this case is required for various reasons but it's ok, because the TX can be wimpy and the IC doesn't have a very big job otherwise.

Multi-layer is really cool. But as Jim said, one is quite restricted with such structures on an IC.

I've had an XFAB account for generations ;-)

I take most of my "walk-ins" to XFAB processing... good bunch of people.

Why do you need such a large value... coupling audio on chip ?>:-}

...Jim Thompson

Wish they would send some Thuringian sausage with shipments. I love those.

Can't talk about that yet, but I really need some capacitance. If we have to lower it we can but it'll come with some penalties for us. It's all at a feasibility stage right now, to see if we can do this with an IC at all. The challenge is the space it has to go into, the 0.2mm width requirement is a hard limit.

Are you doing "chip-on-board", or is that the package size?

Maybe you should have a pro look at your custom I/C needs ?>:-} ...Jim Thompson

That would be a wurst case scenario ;-)

-- "For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." (Richard Feynman)

Don't knock it ;-) ...Jim Thompson

-- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | San Tan Valley, AZ 85142 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

COB, and the stated dimensions are for the bare die. Aside from passivation it may also have to be protected from the elements but that won't add much if it is a coating.

Eventually yes. I am certainly not getting myself into IC design. It may likely be IP blocks to a large extent but still, that ain't my turf.

It's knackwurst, not knockwurst :-)

We are lucky, we've got a German butcher in Sacramento. It's a bit of a drive but the air-dried Landjaeger alone is worth it. He said he can't even get them completely air-dried because they fly off the shelves. So we dry them here. Last time I bought 30, surgeon general's nightmare.