I think automap uses ggplot2. One could ggsave the graphic object to a scalable vector graphic format in a file with large dimensions specified (whether in cm, in or pixels). Alternatively look at the res(olution) option for the bitmap figure types (png, tiff, jpeg, ...). 

Hi Elfatih. It's been a while since I last used automap, but as I remember the autoKrige function which you are probably using has a "new_data" argument where you put a template raster with the extent and cell size you wish the predicted data to fill. Otherwise, the function calculates a default based on fitting around 5000 pixels into the convex hull of your input dataset.

autoKrige(formula, input_data, new_data)

new_data: A sp object containing the prediction locations. new_data can be a points set, a grid or a polygon. Must not contain NA's. If this object is not provided a default is calculated. This is done by taking the convex hull of input_data and placing around 5000 gridcells in that convex hull.

http://cran.r-project.org/web/packages/automap/automap.pdf

Thanks you so much Gerard and Robbi. I will try  your suggestions