If you're passing currents larger than the range of your amplifier, you might be saturating it. I'd guess this would show up as a flat, high current in your scan. If background fitting/flattening is being performed on your current image, this constant current would be flattened to zero. Maybe other weird stuff happens with your amplifier when it saturates, yielding a zero current measurement.

Try putting a resistor in series with your sample to drop the conductivity by a few orders of magnitude, within the sensitivity of your amplifier.

How are you connecting the sampe to the AFM, silver ink, a pin? The sample or the tip are ground? what range are you using to measure? as Mr Martir says maeby you are saturating the amplifier, with an IV curve you can try to know an useful potential, as you should know, the Si and the tip can be study as a Schottky junction, just try to identify the potential before the "diode" turns on 

good luck from México

Armando 

Hi! Thank you both for your answers.

Martin: I have already put a 500MOhm resistor in series with my sample. The current maps just show some noise around a fixed current and the IV curve is really strange (raises exponentially with the applied voltage).

Armando: the sample is attached to a metallic disk with silver paint. I scratched the bottom surface of the sample before connecting it to eliminate oxide. The voltage is applied to the metallic disk (already measured on the disk with a multimeter). I tested continuity with a multimeter from the top of the conductive sample to the metallic disk and it is ok. A virtual ground (transimpedance amplifier) is set to the cantilever tip. Now that you mention it, The I-V curve could be that of a shottky junction in reverse polarization. I have until now only tried positive voltages. I'll give it a try with negative voltages. Thanks for the suggestion.

Best regards,

Sebastian

Im glad to be helpful, im have some experience with CAFM, let me know how it works if it does. 

Best regards 

Armando Barón

So, if anyone's interested I somehow solved the problem with my conductive samples:

1. Play with the amplifier gain and the set point of the cantilever. Decrease the former and increase the latter so you get a good contact. This can however reduce the lifetime of your cantilevers if you are using metal-coated probes :( The best would be to use full metal probes, but they cost 800 USD for only 10 probes O.o

2. Be careful with the velocity of the voltage ramp you apply to your sample. It will influence the noise and quality of the signal a lot.

Best regards,

Sebastian