Equally important as, if not more important than, your question of what curve to be fitted is the question of how to model the variability.
I don't suggest relying on SPSS for this data if you have only minimal statisical training. Consult a reliable professional statisician.