It can be done and collimated with a simple aspheric. The work required vs. the gain on the beam data was not worth the effort. If you attemp this....never let anything touch the faces of the die itself. The divergence of the raw beam will have to be matched with the N/A of the FAC lens you are using. That will set the distance from the output face of the die. A GRIN lens is a similar story....both cases will be done with a 3 axis micro-manipulator for optimum results. While using UV cure adhesive , you can control the exact set point of the optical components. Good luck and please share your results vs. time to achieve the task. See this machine.......http://pyro-science.com/mu1.html
You are the only one that can make your dreams come true....and the only one that can stop them...A.M. Dietrich