I had a student ask me about this and my google-fu let me down. He asked if individual nucleotides (not in a nucleic acid) base-pair with their complementary nucleotides, essentially forming many dinucleotides. I would think this would occur, but have no evidence one way or the other. If it does, how does this impact the usage of nucleotides in nucleic acid synthesis? I'd assume they have to be broken apart first...
The base-pairing in solution only provides 2-3 hydrogen bonds, which the nucleotides could also form with water itself. There are no stacking interactions between bases in water like in a DNA helix. So I strongly doubt that the nucleotides would form a significant amount of dimers in water.
I found a paper confirming my assumption:
Short nucleotide fragments such as mono- and dinucleotides are generally unable to form stable hydrogen-bonded base pairs or duplexes in water.
An easy way to confirm this would be to observe the proton NMR spectra of free nucleotides. Base-pairing in nucleic acids shows characteristic signals in the imino proton region (around 10-15 ppm), which are not visible in unpaired nucleobases due to chemical exchange.
In a DNA strand, there are additional energetic advantages of the hydrophobic bases being internalized to the core the DNA double strand and the polar phosphate backbone being more uniformly exposed to solvent.
Thermodynamically, the formation of the two individual molecules of base monomers forming a pair the solvent will still be exposed to the hydrophobic bases. So such a single base pair would be driven by the energy of the two or three hydrogen bonds in the base pair formed.
This is why you will not see such a complex forming in water; the hydrogen bonds with water are at least as strong as the hydrogen bonds with the bases. But such base pairings can in other circumstances be made to form.
If there are no exchangeable protons for hydrogen bonding in the solvent itself, there will be an energetic advantage to the complex. If there are no exchangeable protons (as with DMSO or carbon tetrachloride) such complexes can be detected by NMR. Even in cases with single hydrogen bondable protons on the solvent, these bimolecular complexes are observed.