Whenever I start working with new samples (multiplexed on HiSeq-2000), some samples house adapter-dimer contaminations and others dont. Samples are prepared by same person using same protocol, sequenced at same time in the same machine in same/different lanes, so why this difference. To my understanding the 45° Y shaped adapters are ligated to the DNA fragment, gel purified and then index + multiplex adapters are ligated which then are PCR amplified and after that ligated on to the cluster bed for bridge amplification. So, the adapters are also amplified and should present in all of them or may be I am missing something.
Source : http://genome.med.harvard.edu/