Use teflon-coated twizer to handle the wafer until it cools down to room temperature.

Since pyroelectric charges appears with temperature variations, I suppose you will have to use a slow temperature ramp up to the process temperature for the resist, but I guess you already try this. One other solution is to deposit a thin metal layer on top of the Niobate substrate to collect charges over a large surface, this avoiding discharges at local points. Maybe it can be integrated in the fabrication process since I suppose you will need electrodes. 

We used lithium tantalate in thin film but never experience such problems. 

I can recommend you to use low heating and cooling speed, less than 5 deg/min when temperature is low and 10 deg/min when temperature is relatively high (T>300 Celsium degreees). When such conditions are used, charges on the surface are neutralized with intrinsic carriers and dipole molecules from air.  You also can use different types of electrodes to make a short circuit between polar faces, but if heating speed is very high, it is useless because of very low intrinsic carriers concentration and mobility. 

I bake lithium niobate at 110 degree centigrade. But I did not see this problem. I dont cool the wafer immediately. I keep on a tissue paper after the bake and take out after few mins.You can also try changing the resist recipe.

Since my question, we finally find the way to avoid electric arcing within the resist layer.

As many of you suggested, we use a temperature ramp starting at 70°C up to 110°C at a rate of 8-9°/min.

No more arcs, and pretty satisfaying resist pattern afterward.

Thanks to all of you for usefull advices.