The R2S=O case is closer to a trigonal planar geometry, the other silicon is tetrahedral. The silicon-silicon distances for different pairs of adjacent molecule types will be different. In a very very rough forcefield optimization I see about 3% difference. I don’t think this one will work out structurally because the chains will become unable to pair after a short length as the chain will not have the flexibility to create the O–H bond without adding too much strain.

Image

But, that’s just one thing. You then need to consider how to actually selectively place/remove the hydrogen atoms, how to avoid the molecule from chemically reacting, and how to read out the data.

So, yes, eventually it would be nice to have a fully orthogonal system. There are already several synthetic DNA base pairs that can be used instead of the naturally present bases. But these would still be susceptible to DNAses or RNAses.

The way I see it is that the chemistry of living things is currently centuries ahead of human tech. A large portion of the techniques used in biochemistry rely on using living things to produce the components, and then we purify those components and use them. It makes a lot of sense to make use of that toolkit because the amount of challenges that need to be solved to create this system from scratch is massive.

Your proposal of your silicon chain reminds of the Ferroelectric RAM, where the state is encoded by the polarity of a cell that is changed by moving a zirconium or titanium cation:

Image

This does work, but it works because the crystal is contained within a semiconductor scaffold, and this is something that we do have a good handle on.