There is something anomalous in your code. If I evaluate Normal[CF] I get a matrix with mysterious variables Dss and Dtt:

{{2 Dss - Dtt, 0, 0, 0, 0, 0, 0, 0, 0, 0}, {0, 2 Dss - Dtt, 0, 0, 0, 
  0, 0, 0, 0, 0}, {0, 0, -Dss + 4 Dtt, 0, 0, 0, 0, 0, 0, 0}, {0, 0, 
  0, -Dss + 4 Dtt, 0, 0, 0, 0, 0, 0}, {0, 0, 0, 0, -2 Dss - 6 Dtt, 0, 
  0, 0, 0, 0}, {0, 0, 0, 0, 0, -2 Dss - 6 Dtt, 0, 0, 0, 0}, {0, 0, 0, 
  0, 0, 0, -Dss + 4 Dtt, 0, 0, 0}, {0, 0, 0, 0, 0, 0, 0, -Dss + 4 Dtt,
   0, 0}, {0, 0, 0, 0, 0, 0, 0, 0, 2 Dss - Dtt, 0}, {0, 0, 0, 0, 0, 0,
   0, 0, 0, 2 Dss - Dtt}}

Similarly with InputForm[CF].

I think you should talk with your collaborator.

I have seen that SparseArray does not work well with Simplify, and I presume also with other functions. In your case I would use either Normal[SparseArray[]] or DiagonalMatrix.

By the way, I suspect that f[x_] := Evaluate[CF] is equivalent to f[x_] = CF. Does anyone know for sure?

I can't speak to whether this is an intentional change in behavior. However, I would suggest that this manner of defining things is fraught. I actually wouldn't have expected SparseArray to work as a sort of "template" anyway--it's a special wrapper to represent a data structure with certain optimizations, and so I would expect to be rather opaque.

But there is probably a better way to get what you want. If you could provide some context for how you're defining CF and the various things that it depends on, then it would be easier to describe an alternate approach. In the meantime one thing you could try is to create all of the structure without actually applying SparseArray until the very last step.