Banner Engineering (ex:
http://www.bannerengineering.com/en-...M-QS18-Series/) makes this type of sensor as well.
You have a number of options. As Chris said, an EMITTER/RECEIVER pair generally is the most robust to different ranges, reflectances, etc. However, it requires two separate parts and (generally) ends up being more costly than other options. You also need to make sure the beam stays aligned (mount them rigidly!).
Another option is using retroreflectors, like the tape used on the vision target this year. You put a sensor on one side of the ball path and the reflective tape on the other. When no ball is present, the sensor sees its reflection. When one passes through, the ball (hopefully) is not reflective enough for the sensor to see it's beam. This only works if the ball is sufficiently more diffuse than the retroreflector; this type of sensor generally comes with a knob to tune the sensitivity to obtain the desired response. You generally also want to mount your sensor a bit further from the ball in this case.
A final option is using a diffuse sensor. Basically, your sensor shoots off into open space when no ball is present. When a ball is present, it sees its reflection off of the ball. As before, this requires some tuning to get working perfectly.
I have worked with all three configurations in FIRST, and generally they can all work. The latter two options simply require some testing and tuning to get working well.