Sai,
Standards for calibration curve of cobalt and yttrium should be prepared in the same manner as for the other metals. It would be advantageous to prepare a low concentration curve first and run a sample(s) of the unknown concentration. If you have to dilute sample(s) you are testing to more than 100 times to fit it/them into the calibration curve, then it is better to establish new curve with a range of higher concentration standards of the element(s) of interest. For example, let’s assume you build a calibration curve for cobalt with standards containing 5, 10, 20, 50 ppb, and a 500 X dilution of tested sample reads 35ppb, then this sample contains close to 17.5 ppm of cobalt. The accuracy of this concentration depends on how accurate you were in making this dilution. In my opinion, rather than being “close to”, you would get more accurate result if you make new calibration curve with higher conc. standards (e.g. 2, 5, 10, 20, 50 ppm of cobalt) and rerun such sample without diluting it. The range of the standards (and control samples) for all elements of the interest is either set by the lab (most common practice) or it is arbitrary set by the analyst based on the concentration(s) range repeatedly seen at majority of tested samples.