Bad fits for protein protein binding

The tail comes after blank injections. We have a general Problem with blank injections, as both in the reference cell and in the flow cell with the immobilized protein, we obtain a negative signal. After reference subtraction, we have a positive "peak" at injection start, that shows an exponential decrease over 70-80 seconds. The wash step, which is set to 200-800s shows no drift anymore. Here is the confusing part:

First, we worked with the same buffer, but used a different detergent (GDN) and we obtained sensogram as in the gF2-GDN sensogram. Then we changed GDN for another detergent, and the injected blanks (the running buffer directly taken out of the bottle attached to the pump) looked fine, however, our analytes did show adsorption. Therefore we added BSA to the buffer which improved our sensograms (gF2-11) however, the fits are horrible and this is the system that now gives a net positive sensogram upon blank injection. We do not think that we can change the detergent anymore, as the protein would not be stable anymore.

I will have to see how it looks with a different chip. Thank you for the advice

From what I learned about detergents it is better to use them above the CMC (critical micelle concentration) than below. The idea is that when you have a too low detergent concentration not all surfaces are saturated with detergent which can induce drift and shifts in the sensorgram when the detergent rearranges itself. When the concentration is above the CMC, all surfaces will be occupied and the detergent will form micelles in solution. If concentrations shift for some reason the micelles can compensate.
So it is worth to look at the CMC of the used detergents.

A link: www.dataphysics-instruments.com/knowledg...ces/surfactants-cmc/
And another: www.stevenabbott.co.uk/practical-surfactants/cmc-values.php

Arnoud