In section 3.7 of his book Introduction to Superconductivity (2nd Ed.), Tinkham states that
[...] we note that S has the eigenvalue $e^{i\varphi}$ in a BCS state in which the the phase of $\Delta$ [...] is $\varphi$.
where S is defined simply as an operator
[...] which annihilates a Cooper pair [...].
I'm trying to prove the above statement. Since Tinkham does not give a explicit definition of this S operator, nor do the references which he cites (Josephson and Bardeen), I am supposing that he meant something in the likings of \begin{align} \hat{S} = \sum_{k} \Phi_k^* \hat{c}_{k \uparrow} \hat{c}_{-k\downarrow}. \end{align}
With this definition, the BCS groundstate can be written, up to normalization, as \begin{align} \vert BCS \rangle &\propto \exp(\hat{S}^{\dagger}) \vert 0 \rangle, \end{align} where $\vert 0 \rangle$ is the vacuum. By applying the Cooper pair destruction operator and after some calculation, however, I find that \begin{align} \hat{S}\vert BCS \rangle &= \exp(\hat{S}^{\dagger}) \sum_k \vert \Phi_k \vert^2 \left( 1 - \Phi_k \hat{c}^{\dagger}_{k \uparrow} \hat{c}^{\dagger}_{-k \downarrow} \right)\vert 0 \rangle. \end{align}
The first term inside the parenthesis gives origin to a term parallel to the BCS state, which is what we wanted. The second term, however, does not seem to do so. This means that the overall state is not parallel to the BCS state and thus the BCS state is not an eigenstate of $\hat{S}$. I will be happy to share any part of the calculations that seem necessary. In particular, the second term that I found seems to stem from the fact that the Cooper pairs are not legitimate bosons, that is they don't obey the usual bosonic commutation relation \begin{align} [\hat{b}, \hat{b}^{\dagger}] &= 1, \end{align} but a modified version \begin{align} [\hat{S}, \hat{S}^{\dagger}] &= \sum_k \vert \Phi_k \vert^2 \left( 1 - \hat{n}_{k \uparrow} - \hat{n}_{k \downarrow} \right). \end{align}
I have checked multiple times my calculations and was not able to find a mistake, which leaves me to suspect of two options:
- Either the definition of $\hat{S}$ I used is not the one used by Tinkham;
or
- There is an approximation that should be made somewhere so that the second term vanishes.
Maybe both, who knows. My question is then:
Do any and which of the above statements apply? Please, provide sources or own derivation.