If BOTH bearings are not properly insulated, operating on the variable frequency drive (VFD) will cause bearing damage. This is even more pronounced with bearings that take axial thrust loads, since there's now another contact surface in addition to the "normal" radial loading. Note that the damage MAY appear on the housing, rather than the bearing - it all depends on how the energy dissipates within the bearing region.
True, some of your observed temperature rise may be a result of poor conduction (e.g. coating the exterior with "sludge"), but it is also going to be a result of the reduced coolant flow through/over the motor housing. Without the proper volume and pressure of flow, the coolant isn't going to provide sufficient mass to carry away the generated heat and/or not get to the correct nooks and crannies within the machine.
Another aspect is the distance from the motor to the VFD. Depending on the length you will need a load reactor to limit the spikes created by the VFD depending on the VFD design type. Also, the pump may need an insulated bearing to stop eddy currents and it is wise to contact the pump manufacturer to get their recommendations. Watch the cable sizing and distance between motor and VFD to prevent unwanted damage (including additional harmonic heating) to motor winding insulation and/or bearing damage.
Basic affinity laws - assuming it's the same pump with no appreciable wear on the impeller or casing - give a fairly simple relationship between speed and other parameters.
(new volume) = (old volume) / (old Hz / new Hz)
(new pressure) = (old pressure) / [(old Hz / new Hz) ^2]
(new power) = (old power) / [(old Hz / new Hz) ^3]
At some point, there will not be sufficient movement of coolant. The motor manufacturer can readily determine where this occurs - and typically publishes it as part of the operation and maintenance manual or other motor documentation. This then defines the minimum operating point (in terms of speed, which also means in terms of Hz).