A wide range of physicochemical conditions promote DNA denaturation, including temperature, pH and osmolarity. My understanding is that you are enquiring about the coupling reaction of a micro-array hybridization procedure. I found a protocol involving cDNA incubation with 0.5 M sodium bicarbonate at pH 9. Follow the link for the complete protocol.
While I am sure Fabio's method would work well, I just had one alternative suggestion to offer (depending on what your goal with the DNA is). You could always get your primer synthesis company to 5'-end tag one of your PCR primers - the one that amplifies the strand you want - with a modification to make your strands asymmetric. This could be either a bulky blocking agent (like a m7G cap, or a triphosphate or something) or a purifiable molecule, e.g. biotin.
Doing this would allow you to differentially treat your strands, either with e.g. a 5'-3' nuclease which cannot digest the end with the bulky blocking agent (and thus would only digest the strand you don't want), or to denature then purify your tagged strain.
Basically, every PCR product is amplified with two primers, each one forming the 5' end of one of the two strands that are amplified. Let's call the two strands "Watson" and "Crick", and say that you're interested in the "Crick" strand. You can synthesise your primers such that the one that primes the "Watson" strand has a normal (usually monophosphate, I think) 5' end, but the one that amplifies the "Crick" strand has a bulky adduct on the 5' end - say, a biotin. When you perform your PCR, you will now have a normal dsDNA PCR product in which the 5' end of your strand of interest - the "Crick" strand - has a biotin molecule at its 5' end.
If you were to treat this dsDNA with a dsDNA exonuclease that degrades in a 5'-->3' direction - for example, Lambda exonuclease, or T7 Exonuclease, both of which are commercially available - you would degrade the accessible "Watson" strand, but not the Biotin-blocked "Crick" strand, meaning that the "Crick" strand would remain as ssDNA. The other advantage of using a biotin addition is that biotin can be bound at very high affinity by streptavidin, meaning that you may be able to use this interaction to purify the ssDNA "Crick" strand - if you can find conditions under which the DNA is denatured but the biotin-streptavidin interaction is not.
There are of course some optimisation/clean-up steps required in both of these protocols, but hopefully they give you some ideas about how you may approach things.