Superstring NS tachyon vertex operator After reading some confusing chapter of various string theory book I'm trying to construct the Tachyon vertex operator for superstring theory. I know that this is removed after GSO projection, but for the moment I would like to construct this would-be vertex. It seems very strange to me that I couldn't find a systematic and precise treatment of this argument. However it seems to me rather natural (thought I didn't understand quite well why) to put an operator of the form
$$
e^{-\phi}
$$
where $\phi$ is the field used to bosonize $\beta\gamma$ system. This operator has conformal weight $h=1/2$, so it must not be the end of the story. However this starting point is confirmed by Polchinski (vol. 2 eq (10.4.22)). Now what else shoud I add? I was thinking some momentum 
$$
e^{ikX(0,0)}
$$ 
as it is done for the bosonic tachyon. But this operator has conformal weight $h=1$ itself and so I would get total conformal weight $h= 1 + 1/2 = 3/2$.
Then on the book by Blumenhagen et. al. I found this sentence:
"The ground state of NS sector is thus $e^{-\phi}c(0)|0\rangle.$" Which seems to me even wrong because the conformal weight of $c$ is $h=-1$
I know that I'm very confused about these superstring vertex operator. This is because I didn't find any book in which there is a comprehensible treatment.
It would be of great help if someone provide a solution to this construction.
 A: The tachyon vertex operator is 
$$
c\,e^{-\phi}\tilde{c}e^{-\tilde{\phi}}\,e^{ik.X}
$$
where the $z$-frame ghost number is $(-1,-1)$, the picture number is $(-1,-1)$ and the conformal weight $(h,\tilde{h})$ is 
$$
h=\tilde{h}=\left(\frac{\alpha 'k^2}{4}\right)_{X}+\left(\frac{1}{2}\right)_{\phi}+\left(-1\right)_{c}=\frac{1}{2}\left(\frac{\alpha 'k^2}{2}-1\right)
$$
The BRST-closed vertex operator with picture number $(-1,-1)$ and $z$-frame ghost number $(-1,-1)$ should be of the form 
$$
c\,e^{-\phi}\tilde{c}e^{-\tilde{\phi}}\mathcal{V}
$$
where $\mathcal{V}$ is a $(1/2,1/2)$ superconformal tensor. In particular, the total conformal weights should be $(0,0)$, since $-1+1/2+1/2=0$. This determines the value of $k^2$ to be:
$$
-m^2=k^2=\frac{2}{\alpha'}
$$
a tachyon of mass $m^2=-2/\alpha'$.
If you change to the picture $(0,0)$, there will be no $e^{-\phi}e^{-\tilde{\phi}}$ insertion, and the tachyon vertex operator will be:
$$
\frac{1}{4}(\alpha ')^2 k_{\mu}k_{\nu}\,\,c\psi^{\mu}\tilde{c}\tilde{\psi}^{\nu}e^{ik.X}
$$
A $(+1,+1)$ picture tachyon vertex operator will be
$$
e^{+\phi}e^{+\tilde{\phi}}\frac{1}{4}(\alpha ')^2 k_{\mu}k_{\nu}\,\,c\left(i\partial X^{\mu}+\frac{1}{2}\alpha' k_{\rho}\psi^{\rho}\psi^{\mu}\right)\tilde{c}\left(i\bar{\partial} X^{\mu}+\frac{1}{2}\alpha' k_{\sigma}\tilde{\psi}^{\sigma}\tilde{\psi}^{\nu}\right)e^{ik.X}
$$
and so on. Always with $(0,0)$ conformal weight. This is not a problem because the ghost breaks unitary, so non-trivial local operators with zero conformal weight are allowed.
Note also that the world-sheet fermion number $(-1)^{F}$ is always the same since both $e^{n\phi}$ and $\psi$ carry a fermion number $-1$, assuming that $n$ is an integer (that is the case in the NS sector). The picture changing always substitute one $e^{n\phi}$ for one $\psi$ or vice versa. 
