The arguments leading to the Quantum Hamming Bound for non-degenerate $[[n,k,d = 2t+1]]$ codes $$\sum_{j=0}^t 3^j \binom{n}{j} ≤ 2^{n-k}$$ can be manipulated to find a stronger related bound for non-degenerate CSS codes, $$\left(\sum_{j=0}^t \binom{n}{j}\right)^2 ≤ 2^{n-k}.$$ Note that this is simply the classical Hamming bound with the left-hand side squared.
Examples:
- the quantum Hamming codes $[[2^r-1, 2^r - 2r - 1, 3]]$ fall short of the normal QHB but would saturate this one, as LHS = RHS = $2^{2r}$,
- Shor $[[9,1,3]]$ code has a strict inequality $100 < 256$,
- Gottesman $[[2^r, 2^r - r - 2, 3]]$ codes violate it but they aren't CSS,
surface codes violate it but they are degenerate.(example disputed in comments)
I think this is pretty straightforward, but I can't seem to find it anywhere to cite it. Does it have a name? Or perhaps it's wrong, can you help me find a counterexample? Or did I just discover something?