It first a question of connected inertia. To stop it you have to destroy the kinetic energy in the different parts in a time coherent with your delay. It is not enough to nil the magnetic field by interruption.

Appropriate,"Regenerative breaking"......

It depends on the motor construction and its overloading capacity. To stop the whole only via the motor it has to be used as generator and the power directed either to an "accumulator" or net (regenerative) or simply transformed in heat in a power resistor network. The shorter the braking time interval the higher the dissipated power. This could lead to an overload for the commuting system (solid state or not) and destroy parts of it due for instance to too high currents.

This can be done breaking by regeneration. Because nullifying magnetic field would not be an adequate . KE is the time coherent.

aditional information needed: kinematic scheme and scheme of excitement and J

To qualify answer the additional information needed, ecpecialy J and cinematic 

The problem being of general interest  I made a trial to estimate the feasibility.

I took a motor from a known manufacturer since not all give enough data concerning the product. Using the motor as generator the generated voltage depends on the backemf coefficient Ke. If the friction losses are neglected (which can be a good assumption for motors with ball bearings) one comes to 1st degree differential equation wich leads to a time constant T= (Ri+Re)*(Jtotal reduced at motor shaft level)/Ke^2. Ri= internal motor resistance ans Re= shunt external resistor. For the example I got T=80 ms. In fact as long as the friction ( Coulomb type) is not considered the rotor will continue to turn. As a remark this time constant is different from the so called "mechanical time constant" which is valid for the active motor function and considers as well the torque constant. This the simplest way to do it but of course other can be more effective.

It will stop only when the active torque will cross the friction threshold. 

Why not just reverse the polarity, for as long as it takes to stop the motor?

Assuming we're talking about a permanent magnet DC motor, if you reverse the polarity of the voltage applied to the rotor, torque opposite that previously applied will begin to be applied to the rotating mass, and the rotation should slow down quickly. More quickly than just using regenerative braking.

http://ebldc.com/?p=34

Of course, larger currents than usual will result, so you have to make sure the motor is up to such abuse. (We used to suddenly reverse the polarity to the locomotive in our DC electric train setup, when I was a kid. Stops the train in no time. And it even spins the wheels. Fun!)

There are several methods, including the regenerative braking and reversing polarity.  Regenerative braking can be fairly fast, as long as you have a very low resistance that can handle the current surge.  The polarity reversal trick has the problem of determining when the motor has stopped, because if you keep the reverse polarity on beyond the stop point, the motor will reverse.  You can do this in a variety of ways, such as measuring the current draw and looking for the minimum.  But if I really wanted to stop it quickly, I would probably use something like a small gear and a solenoid that slammed a detent into the gear.  This has a number of problems, such as physical wear, but it will stop within one tooth.  That 6000rpm is a potential problem.  Another technique would be to use a physical brake, with a wear-resistant surface.  If by "magnetic breaker" you mean an external clutch-like mechanism, that's describing my last suggestion.  Is the .1Nm value the torque of the motor with no load, or is it the amount of energy stored in a "flywheel"?  Is the load direct or geared down?  Too many undefined parameters to suggest a proper solution.  What are the mechanics of the brushes?  1/2" graphite cylinders spring loaded, or brass springs against brass segments?  These are going to get a lot of abuse, both physically and electrically, for any regenerative braking, reversing, or other electrical approach.  Mechanical wear on non-replaceable components will destroy the motor; mechanical wear on external replaceable components may mean more frequent, but lower-cost, maintenance.

Note that "100W" tells us nothing useful.  Is it at 120VDC (< 1A current draw) or 5VDC (20A current draw). 

With the level of information we have at our disposal NO optimal suggestion can be made this is the reason I suggested a way to estimate the stopping time without tampering with the mechanical transmission. A mechanical brake implies an added rotating component thus increasing the overall inertia. The supply inversion requires a very fine control with a voltage variation according to rotating speed which can be obtained by PWM control at a frequency high enough. It has  the advantage of limiting the modifications on the control level and not touching the mechanical part.

The problem remains -as I mentioned from the beginning- the overload capability of the motor commutator whichever type it is. One should think that at inversion the voltage being the sum of the supply and the bemf can be increased by a factor of 2! And the current grows in same proportion!!! 

a very simple answer for this problem is using of mechanical or magnet breakers on your motor shaft.

reverse voltage rather than regenerative mode can be useful also.