There are several things I can think of, but you might just be expecting too much of your instrument. The AvioMax is an ICP-OES instrument, so detection is not going to be as good as an ICP-MS at the lower end of your calibration range. Trying to get 20 ppb consistently with an OES might be an extremely frustrating exercise.

Even in vendor product brochures eg [https://www.perkinelmer.com/libraries/APP-176007-Avio550Max-ICP-OES-Elements-in-Soil ], which are usually quite optimistic (to put it politely), method detection limits did not go much below 0.1 mg/kg apart from several highly responsive elements

From what you've written in the 2nd last sentence, the preparation of standards is not in question, it is the response of the instrument itself. Did the ICV and CCV pass on the second instrument as well? Are they realistic concentrations to try and measure? - somewhere in the 0.02 to 0.1 ppm range I gather from the 2nd sentence.

If you can, perhaps relax your criteria a little, make a slightly higher concentration CCV solution and drop the 0.02 ppm standard in favour of a higher concentration (8 or 10 ppm), and extract a larger mass of soil into a smaller volume at the final dilution.

ICP-OES is fairly robust, it seems unlikely that 80 mg of metal /L (4ppm * 20 metals) would overload the plasma. Perhaps one instrument is more sensitive (newer detector, different component specifications etc). Or the lens of one instrument is dirty and not letting as much signal into the mirrors etc.

Can you examine the raw data (counts per second) for both instruments? Maybe that will tell you which one is sensitive (or losing sensitivity) for a particular element x concentration combination.