As in any similar measurement technique in which the responses of the measuring instrument are calibrated with multiple standard values, the resulting calibration relationships can be linear or non-linear. If this is your first contact with this measurement technique and you don't have someone who knows how to help you, then you have a problem. Unless you have a lot of time and willingness to learn everything yourself. What specific (manufacturer, type) spectrometer you have and what elements and in what concentrations you want to measure?

ZJ

You can use a statistics tool in the operational software or use Data Analysis/Regression or Graph/AddTrendline in Excel.

In general (but maybe not every case), you would want to keep the calibration curve as linear as you can. This probably means you need to select an appropriate range of concentrations that form a linear relationship. Two examples:

• If you overload the detector of the ICP-OES, then it will start to give a non-linear response (at high concentrations).
• If you have an overlapping signal from a second element (that you may or may not be monitoring) that interferes with the baseline of your element of interest, you will underestimate your primary element at almost all concentrations.

The quadratic equation might help with example 1, but there will be many people advising against such a practice (replacing a worn-out detector might not please those who pay for the instrument's upkeep for one).

Example 2 may not be fixed by either linear or quadratic equations, which means rethinking the set-up of the instrument and the calibration (and possible sample preparation too).

In short, keep within a linear range as much as you can, don't overload the detector, look for interferences. If you haven't identified the interference(s) in your samples, you haven't looked hard enough.