Posted by Dan (220.127.116.11) on June 01, 2012 at 16:38:42:
In Reply to: surface finish posted by Anna on June 01, 2012 at 05:49:23:
: I'm a designer and I also perform srp measurements. I have samples with N epi growth on a N-type substrate which should have resistivity = 1 mohm*cm. I have obtained a high noisy profile at the substrate. I have measured a substrate resistivity higher or equal to 2 mOhm*cm and by reapeating the measurements I have understood that the problem is the surface finish! That is, I believe for low resistance values the surface finish is very important and can lead to totally wrong results. Based on your experience do you think I'm right? If yes, how can I control the surface finish to obtain a reliable resistance value? Thank you for your kind help.
Surface finish may be the problem. Or it may be your probe conditioning.
If you have a poor surface that has lots of scratches you will get bad readings when you run over the scratches. Sometimes there are scratches you cannot see with your eye but the probes see electrically. Lots of scratches can lead to lots of noise in the data rendering it unusable.
If you have probes that are poorly conditioned they will not make good ohmic contact with the sample regardless of the surface finish. The probe contacts need to have lots of micro-contacts that are capable of punching through native oxide but not deep into the sample. Good micro-contacts provide low noise resistance data at all ranges however they do decay after time. After conditioning and calibrating a set of probe tips we find that we can get anywhere from a few days to sometimes an occasional month before the probes wear down enough that they need to be reconditioned or replaced.
How can you tell which problem you have? Well you can examine the bevels under a microscope for scratches. Our bevels are created using diamond cutting wheels with less than a hundredth of a micron cutting edge. This allows our bevels to have a roughness of less than 5Å RMS. So they aren't much less shiny than a mirror polished wafer surface. This makes scratches pretty easy to see on our bevels. If you have scratch free surfaces you can then look at the amount of noise you see on the QTA/Pen/Low Rho samples. When probes start getting worn out, the excessive noise will usually show up at the low resistances and work it's way upward. So if only one range of the resistance curves, (most likely the low resistances,) are noisy, chances are you have probe wear problems instead of surface finish problems.