Creating a spiral inductor geometry

[eskiyerli]

I would like to understand if I could create a geometry for a spiral inductor over a dielectric such in the integrated circuit layouts.

[dsmi]

Howdy! I would definitely recommend you to use embox solver for that kind of simulations -- it is more accurate, and handles the layered dielectric better.
If you provide some details -- square/archimedian spiral, dimensions -- I can help you with the meshing and simulation setup.

[eskiyerli]

It is really a general question. RF integrated circuits use spiral inductors that are usually modelled by Momentum (from Agilent) or full 3D simulators. These simulators are expensive and are not very useful if you need to use simulators a few times a month or so. A silicon integrated circuit sits on a number of layers as show in the following figure:

On-chip spiral inductors look like this:

Usually they use at least two upper-most metal layers, but sometimes three metal layers if they have symmetrical layouts.
Most integrated circuit layout design software export their files in GDSII format, which is a text format that defines the various layers and their geometries in an integrated circuit. There are also Python libraries that can generate GDSII output that can be read to most integrated circuit design software. Thus it would be very good if the simulator can somehow read GDSII format and convert it to its internal representation.
The simulator output should be the s-parameter and its companions, i.e. z and y parameters. It could be then converted to an approximate circuit model to be used with the circuit simulators.

[dsmi]

I thought you have a particular inductor in mind, which you wanted to simulate -- that is what my question was about.
The solvers on my page are most of all my exercises in computational electrodynamics, and I doubt that the GDSII format support will be ever implemented, at least by myself.
That being said the solvers -- embox in particular -- are fully capable of simulating the structures like this.
To demonstrate, I created a script which generates a mesh of the spiral inductor, simulates it, and saves the resulting Y-parameters -- see spiralind.m Optionally it can also generate the current density plot, which is shown below.

[eskiyerli]

Dear Dmitry, Thanks a lot for your email and help. I will go through your example. I think it could be a great start. Is it possible to add dielectric layers in this structure? They could be between the inductor and ground as well as around the inductor metal. Thanks, Murat

…

On Fri, 8 May 2020 at 00:05, Dmitry Smirnov ***@***.***> wrote: I thought you have a particular inductor in mind, which you wanted to simulate -- that is what my question was about. The solvers on my page are most of all my exercises in computational electrodynamics, and I doubt that the GDSII format support will be ever implemented, at least by myself. That being said the solvers -- embox in particular -- are fully capable of simulating the structures like this. To demonstrate, I created a script which generates a mesh of the spiral inductor, simulates it, and saves the resulting Y-parameters -- see spiralind.m <https://github.com/dsmi/embox/blob/master/examples/spiralind.m> Optionally it can also generate the current density plot, which is shown below. [image: inductor_current] <https://user-images.githubusercontent.com/6131468/81348889-2a679f80-907c-11ea-860e-b74a85a3aa40.PNG> — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#1 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ALPUS7ETC4XT7BYX3ARFMGTRQMWC5ANCNFSM4MXIAG7A> .

[dsmi]

Murat, this example already includes two different dielectric layers -- between the inductor and the ground and around the inductor. More can be added as needed -- this solver processes stratified dielectric analytically, which means at no cost from the simulation time point of view.
I modified the example script to make the dielectric and conductor parameters more explicit.

[eskiyerli]

Thanks. One last question. Is it possible to use non-manhattan geometries, i.e. 45degree lines?

[eskiyerli]

No worries, it can indeed solve non-manhattan geometries. I must say I still have hard time in understanding the software as there is almost no documentation except the function helps, but I managed to create a 45-degree angle micro-strip line.

I think your software has a lot of potential but it needs at least some documentation and explanation.