Our Kelvin probes measure Contact Potential Differences of samples with respect to a golden tip.
We have adopted the definition of CPD from IUPAC which states that it is a Volta potential difference between two metals.
So, it is a matter of convention which sign of CPD, positive or negative, will be used in Kelvin probe software.
Legacy convention
So far we have been using a CPD definition that in absence of surface dipoles$^*$ reduces to $e\mbox{CPD} =E_F^{\mbox{Au}} - E_F^\mbox{s}$, where $e$ is the elementary charge (positive!, despite its symbol suggesting it is the charge of an electron) and $E_F$'s are Fermi levels of gold (Au) and sample (S). Fermi levels are measured with respect to the vacuum level $E_{\mbox{vac}}=0$.
Lets suppose that the sample at hand is a piece of aluminum (Al). Our sample holders are made of aluminum an can double as a sample. It is known that $E_F^\mbox{Au} < E_F^\mbox{Al}$ and the CPD defined above would be negative.
This choice of CPD sign is sightly counterintuitive if one keeps in mind that the sample is positively charged with respect to the tip when in contact. The Volta potential just above the positively charged sample is also positive.
Current (intuitive) convention valid since October 2023
Indeed, since $E_F^\mbox{Au} < E_F^\mbox{Al}$ upon contact of two neutral metals Au and Al the electrons will flow from aluminum to the gold and aluminum will be positively charged as result.
Thus, it is better to us a “sample-centric” definition of CPD
$$
\mbox{CPD} \equiv V_\mbox{s} - V_\mbox{Au},
$$
where $V_\mbox{s}$ is Volta potential just above the sample and $V_{Au}$ is Volta potential just above the golden tip. This in absence of surface effects$^*$ reduces to
$$
\mbox{CPD} = \frac{E_F^{\mbox{s}} - E_F^{\mbox{Au}}}{e}.
$$
$^*$ Very much like work function, the CPD value has bulk and surface contributions. In absence of the later, the CPD is just a Fermi levels difference of the two metals when separated and uncharged.
