I’ve put SMT resistors film-side down, and used two 100 in parallel for edge currents, but the former does not lend itself to automated assembly. I did measure this at 67 GHz with 0201 resistors and it was about 10 dB return loss; it was done in a pinch. Though I’ve also just seen dangling traces for rat-race termination at mmWave.

I’d like to try dropping one (or multiple in parallel) vertically into a non-plated hole, which should give a better field termination. In the end, the field is what matters, like a wedge absorber in a waveguide, TEM should be similar.

Many years ago I did simulate an exponentially tapered thin-film on AlN termination that looked good; high power and broad band, but never fabbed it.

There’s just no time to experiment with these things