metal width for 5G clk routing

[huangjw]

dear all,
i want routing a 5G clk using second metal, it length is about 250um. i want using 0.2um width, is it too narrow? does the high speed clk use the metal the wider the better?
thanks!

 

[yadavvlsi]

Clock is highest switching signal in the design. Clock nets are normally routed with double width (2x minimum pitch) and double spacing. Double width is used to avoid electromigration and double width is avoid crosstalk. Wider wires have less resistance but have high capacitance. 5Ghz may have some requirement.

 

[dick_freebird]

If you are talking 5G as in 5GHz, that's a mighty lossy
proposition. What you want to do for layout doesn't really
enter into it. Look at signal attenuation for starters, what
is your distributed RLC doing to the signal in terms of both
amplitude fading at the far end, and clock skew at various
tap-points?

I believe you want a transmission line approach, just so
you know what you're getting. This can also help with
the cross-contamination.

Heck, at 400MHz I found it necessary to space clocks
10um from other wires that would load the line, although
I was also throwing clocks across thousands of um. Use
your layout parasitic extraction (C at least) and a hand-
figured R to see how much loss you're looking at in a dumb
non-Tline wire, at various keep-away distances.