Many things affect detection limits, such as injection volume, column diameter and length, temperature stability of the cell. even how clean your DI water is that you use to make blanks and standards. As mentioned above you will need to do a study to show the limits of the instrument, if you can spread it over a few days that is even better to show variability from day to day.
Start with something like 1ppm, then dilute by ten and inject again, then by ten again until you are not seeing a peak, that will give you a rough idea of where to begin. 1ppm to be sure you are identifying the peak, maybe even 10ppm to be sure, then 100ppb, 10ppb and 1ppb. If you see 10ppb but not 1ppb, try 5ppb. You will want a peak that will give you at least a signal to noise of 5 to have a solid detection.
The %RSD of the areas at the detection limit can be high, since this is a limit of detection not limit of quantitation. Start about 5 times higher than the limit of detection as the lowest standard of your calibration curve and work up from there. Then see how accurately you can quantify the lowest standard, should give nice reproduceability and accuracy, if not you have to go higher, if it is very very good, you can go lower.
Detection limit and quantitation limit are something you have to work at when doing it initially. Once you find the limits of the instrument using clean standards, you will need to work with blank samples and spiked blank samples to find the true limits of the method you are using, so you take into account all of the preparation steps. It takes a while to do it right. Then you also need to be sure you meet any regulatory conditions that may govern the testing as mentioned above such at EPA, FDA, ISO, ect.