Before the advent of the DNA arrays it was not easy to achieve specific and efficient hybridization to several DNA probes in a single experiment. Conditions optimal to one probe favored nonspecific or no binding to another. With microchips, thousands of probes simultaneously hybridize to their targets.

How that has been made possible? Is there a free internet resource explaining this?

The assumption you have made is that hybridization conditions have been optimized for each probe. This is not actually the case. It seems that (in the case of Affy), probes are chosen such that they all have a Tm close to the 'average' in hopes of keeping the hybridization conditions constant. In reality, this is impossible, especially when they are constrained to pick probes from a very small region. The rest of the magic comes after the fact. There are many statistical algorithms that aim to identify and correct for these issues. Have a look at this recent paper for information on all the tricks that are used (on the data processing end) and how well they work:

Quin, LX et al. BMC Bioinformatics. 2006 Jan 17;7:23.

For future discussions on microarrays and other gene expression applications, please post in the Gene Expression forum (in Genomics).