0
\$\begingroup\$

I am a physicist with an general EE measurement question, thanks in advance for the advice!

I am measuring the properties of a majority carrier diode that I am designing, and usually just take DC IV sweeps. I know that there is shunt conductance mechanism which I am trying to minimize. I have starting taking CV sweeps as well (1 MHz, 50 mV AC ripple) over the same DC bias range, under a C || G model. For all of these, I am essentially comparing reverse bias DC leakage to CV profiling data (also reverse bias, of course).

Everything qualitatively makes sense, the less leaky diodes show lower conductance, extracting depletion widths, doping concentrations by textbook formulas are accurate.

I basically have two questions. Since I am dealing with some relatively leaky diodes, how confident should I be in trusting the CV data? I have been told you cannot use if if you have large leakage conductance, I have been using the rough rule of 2*pifC>G, is this ok?

Second, is there a fundamental difference in trends in measured AC conductance vs. DC current? For instance, some diodes see trends in the conductance, which I might ascribe to details of the leakage mechanism. But is CV conductance data really noteworthy, or just "slop" to tell you if you can trust the capacitance? In other words, am I getting any more information than my DC IV?

Thanks again, any refs would be much appreciated!

\$\endgroup\$

1 Answer 1

1
\$\begingroup\$

Answering your second question first...

Heavier-doped junctions have less leakage current. Heavier-doped junctions also have a thinner depletion width, so they have more capacitance. So taking both measurements seems redundant.

What breaks the redundancy is when you realize that other factors that can contribute to leakage and capacitance.

Majority carrier diodes (i.e. Schottky diodes) are constructed by interfacing a metal with a semiconductor. The interface introduces crystal defects and charge traps. This causes increased leakage which doesn't correlate well with the capacitance. The defect density will depend on processing nuances, so you can't expect it to remain perfectly the same throughout the product's lifetime.

Capacitance is also introduced by interconnect. Especially for lateral devices, anode and cathode interconnect can be interleaved quite tightly to permit high currents to flow under forward bias, at the cost of a fixed additional capacitance.

To summarize, while in theory the measurements of leakage and capacitance are redundant, in practice they are not.


Answering your first question...

You're right to say that measurements of \$C\$ become less accurate when \$2\pi fC<G\$. However, the accuracy of such a measurement really depends on your test instrument.

The operation of an impedance analyzer is similar to a lock-in amplifier. An AC stimulus is applied, and two AC measurements are taken: one in-phase measurement, and one out-of-phase measurement. The in-phase measurement tells \$G\$, and the out-of-phase measurement tells \$B\$ (susceptance, from which \$C\$ is derived).

The ability of the instrument to resolve \$B\$ from \$G\$, i.e. to tell the difference between the in-phase component and the out-of-phase component, depends strongly on the details of the instrument's design. It's quite possible to get a reasonably accurate measure of \$B\$ when \$G\$ is 10, 100, or even 1000 times larger. But again, it depends on the instrument design.

The best thing to do is to refer to the instrument's user manual.

\$\endgroup\$
1
  • \$\begingroup\$ Thanks for the info for first question, I am using an Agilent B1500, so I know its quite decent. I calibrate with the low/ground pad contacted to minimize stray capacitances, noise is aF range while measurement is usually pF, ill check the manual details, I know, RTFM first! \$\endgroup\$
    – daFireman
    Apr 15, 2015 at 23:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.