I was reading this article discussing non-inverting opamps and when presenting the above single-rail opamp circuit it gave the following not regarding capacitor C2:
Capacitor C2 type: It is important that this capacitor must be a low leakage type. If not the leakage could cause the output to hit the voltage rail if there is some gain in the circuit.
Why the "low leakage" requirement? Wouldn't any leakage current simply be compensated for by OpAmp action?
Any leakage in C2 would only be corrected by the output of the opamp moving positive to supply the leakage current through R2.
If this moved the output close to the positive supply rail it would reduce the output swing capability or even saturating the opamp. The actual effect will depend upon the value of R2.
The caution seems to be somewhat exaggerated. In my experience it is rare for leakage in that capacitor to be an issue with the normal values of R1 and R2.
To avoid issues with stray capacitance R1 and R2 would usually no the higher than some 10's or 100's of kilohms. It would take many microamps of leakage in C2 for there to be a problem.
In most circuits it would be acceptable for C2 to be electrolytic or ceramic depending upon the value required.
If the circuit uses very high values of resistance and/or low supply voltages there could be more of an issue.
There is a word missing from the caution, it should read
... must be a relatively low leakage type
That is, with respect to the value of R2.
The potential divider R3/R4 sets up the bias voltage on the amplifier inputs, which establishes the voltage across C2. Any leakage current through C2 has to flow from the amplifier through R2, and so will result in a DC offset.
In all practical circumstances, if C2 is non-electrolytic, then you won't have any issues. The leakage for plastic or ceramic capacitors tends to be so low you'll never observe it as a hobbyist.
If C2 is electrolytic, then there could be a problem if C2 or R2 are given extreme values, say >1000uF or >10Mohm, or if the cap voltage rating is very close to the bias voltage on it.
Interestingly, if C2 is a new electrolytic, or has been in storage for a long time, then there's likely to be an initial problem, even for reasonable values. An electrolytic capacitor will need to reform the oxide layer on its anode, and until this happens, the leakage could be orders of magnitude above specification. If R2 is big(gish), then the current available to do this will be very small, and this could take a long time. If this happens, it's a good plan to reform the capacitor manually by charging it to its rated voltage, or at least to a voltage above the final use voltage, for a while before discharging it and soldering it in.
