I am not sure it would work for your purpose but hydrogen peroxide can sometimes be used instead of a plasma cleaner. An example is in the following paper: http://dx.doi.org/10.1039/B610011B
If the LiNbO3 layer is very smooth and flat, you should be able to use auto-adhesion to attach your fluidic. I have successfully pumped as much as 1.5ul/min through microfluidcs devices that were auto-adhered to various surfaces. Typically, if you can get a specular reflection from the surface, PDMS will have some auto-adhesion on that surface as long as the surface that you cast the PDMS on also had the same smoothness and flatness.
If you desire to activate the surface for irreversible bonding, you might want to try atmospheric corona discharge which may be produced with a variety of methods using common lab equipment.
The article Claudiu mentions is very interesting. However, instead of hydrogen peroxide, it uses HCl:
"The chips were soaked in 0.01 M HCl at 80 C for 1 h to make the surface more hydrophilic. The devices were completed by attaching #1.5 cover glasses to the patterned surfaces of the PDMS chips and baking them at 80 C overnight to promote adhesion between the PDMS and the glass."
Thank you for the great article suggestion Claudiu. I may try this technique.