In a nuclear reactor, shutdown refers to the state of the reactor when it is subcritical by at least a margin defined in the reactor's technical specifications. Further requirements for being shut down may include having the reactor control key be secured and having no fuel movements or control systems maintenance in progress.
The shutdown margin is defined in terms of reactivity, frequently in units of delta-k/k (where k is taken to mean k-effective, the effective multiplication factor) or occasionally in dollars (the dollar is a unit equal to the change in reactivity needed to go from critical to prompt critical). Shutdown margin can refer either to the margin by which the reactor is subcritical when all control rods are inserted or to the margin by which the reactor would be shut down in the event of a scram. Hence, care must be taken to define shutdown margin in the most conservative way in the reactor's technical specifications; a typical research reactor will specify the margin when in the cold condition, without xenon. Under this specification, the shutdown margin can be simply calculated as the sum of the control rod worths minus the core excess.
Minimum shutdown margin can be calculated in the same way as shutdown margin, except that the negative reactivity of the most reactive control rod and non-scramable rods is ignored. This definition allows the reactor to be designed so that it remains safely shut down even if that most reactive control rod becomes stuck out of the core.
A reactor is in cold shutdown when, in addition, its coolant system is at atmospheric pressure and at a temperature below 100 °C (210 °F). This temperature is low enough that the water cooling the fuel in a light water reactor does not boil even when the reactor coolant system is de-pressurized.