What is the reverse recovery time in a diode?
If a diode is conducting in a forward condition and immediately switched to a reverse condition, the diode will conduct in a reverse condition for a short time as the forward voltage bleeds off. The current through the diode will be fairly large in a reverse direction during this small recovery time.
After the carriers have been flushed and the diode is acting as a normal blocking device in the reversed condition, the current flow should drop to leakage levels.
This is just a generic description reverse recovery time. It can affect quite a few things, depending on context, as mentioned in the comments.
A space charge within a P-N junction needs to be established before forward current can flow. (If the first sentence makes you ask why, that's really a separate question -- perhaps this can help. Let's just look at the dynamics of establishing and neutralizing that space charge.)
From zero, this space charge can be established quite quickly, because an externally applied forward bias voltage can route electrons externally around. Electrons diffuse from the n-type material into the edge of the p-type material, holes in the p-type material diffuse into the edge of the n-type material, and at the metal interfaces, new electrons are injected into the n-type end and holes are generated at the p-type end to produce free electrons that can flow in the external circuit. All of these flows are flows of majority carriers in their respective materials, so diffusion happens quickly driven by much larger concentration gradients. A space charge develops rapidly because majority carriers are flowing to turn the diode on -- electrons in the n-type material, and holes in the p-type material.
However, if the external voltage is then reversed to be a reverse bias, the space charge is attracted to itself to recombine. But this recombination only happens through the diffusion of minority carriers. This minority carrier diffusion has much smaller concentration gradients, and therefore diffuses orders of magnitude more slowly. An external circuit providing reverse bias can aid in speeding this recombination, as it can allow for faster neutralization of excess holes that migrated back to the p-type material, and removal of excess electrons that migrated back to the n-type material. This hole-electron recombination or charge neutralization is assumed to happen essentially instantaneously at the semiconductor-metal interfaces, so if the external current can supply and remove electrons under reverse bias, it will do so much faster than the "normal" hole-electron recombination rate in the bulk of the semiconductor. That's why there can be huge reverse currents during the reverse recovery time.
I put together a little simulation of reverse recovery time in a 1N4007 diode vs a 1N4148:
The demo shows the diodes being switched under a square wave, and shows that the 1N4007 takes a few microseconds to turn completely off!
(See also a PDF titled "Recombination Time in Semiconductor Diodes".)
If diode is forward biased and you want to turn it off, it takes a while to extinguish free carriers flowing across the junction (electrons have to get back to n-region and holes have to get back to p-region, then they can recombine at the anode and the cathode, respectively). This time is called "reverse recovery time" and the total current flowing across the diode is negative, because carriers flow in opposite directions with respect to forward bias. The charge flowing during reverse recovery time is called "reverse recovery charge" and the diode has to extinguish it ("recovery" from reverse-biased to neutral condition) before you can turn it on. In the end, reverse recovery phenomenon depends on silicon doping and geometry and is a parasitic effect in diodes, because energy involved in the process is lost. strong text