Edwards, D. P., M. Hassall, W. J. Sutherland, and D. W. Yu. 2006. Assembling a mutualism: Ant symbionts locate their host plants by detecting volatile chemicals. Insectes Sociaux 53: 172–176. DOI: 10.1007/s00040-006-0855-z
Dáttilo, W. F. C., Izzo, T. J., Inouye, B. D., Vasconcelos, H. L. and Bruna, E. M. (2009), Recognition of Host Plant Volatiles by Pheidole minutula Mayr (Myrmicinae), an Amazonian Ant-Plant Specialist. Biotropica, 41: 642–646. doi: 10.1111/j.1744-7429.2009.00518.x
I do not know the subject, but I recently read an article that talks about two models of olfactometer that maybe you can serve. This is the link: http://revistas.ucr.ac.cr/index.php/agrocost/article/view/9965
I personally prefer T mazes to Y mazes. The arguments also stand for olefactometers There is a simple, technical explanation for this, but it is best explained with graphs. I outline my arguements in my PhD thesis, which you can download from my profile.
Remember, if using a active-flow olefactometer, that you need to very carefully control the flow on both sides, one side will overpower the bifurcation. You can visualise your flow using fake smoke. Using a passive olefactometer might be a lot easier.