I am trying to reconstruct something and I would appreciate someone helping me filling the gaps. To motivate my question let's first consider chiral perturbation theory with up and down quarks. This is the lowest order Lagrangian
$$\begin{equation*}
\mathcal{L}_2=\frac{f^2}{4}\langle{}\partial_{\mu}U^{\dagger}\partial^{\mu}U\rangle
\end{equation*}
$$
where $f$ is some parameter with mass dimensions and $U$ is defined via
$$U=e^{it^a\phi_a/f}$$
where $t^a$ are the Pauli matrices and $\phi_a$ are complex scalar fields. For some reason I don't know, people take this matrix to be unitary. So my first question is, why does $U$ have to be unitary?
In any case, if we demand $U$ to be unitary
$$UU^{\dagger}=1$$
$$e^{it^a\phi_a/f}e^{-it^a\phi_a^*/f}=1$$
and introducing $e^{-it^a\phi_a/f}$ on both sides
$$e^{-it^a\phi_a^*/f}=e^{-it^a\phi_a/f}$$
$$t^a\phi_a^*=t^a\phi_a$$
$$\begin{pmatrix}
\pi_3&\pi_1-i\pi_2\\
\pi_1+i\pi_2&-\pi_3
\end{pmatrix}=
\begin{pmatrix}
\pi_3^*&\pi_1^*-i\pi_2^*\\
\pi_1^*+i\pi_2^*&-\pi_3^*
\end{pmatrix}$$
now, redefining $\pi_3\equiv\pi_0$, $\pi_1+i\pi_2\equiv\sqrt{2}\pi_-$ and $\pi_1-i\pi_2\equiv\sqrt{2}\pi_+$ we see that the unitarity of $U$ imposes that $\pi_0$ is a real field after all and $\pi_-^*=\pi_+$
This up till now is nothing but the usual pion theory. I want to supersymmetrize this theory and this is where most of my doubts arise.
I have been told that I can embed the Lagrangian considered so far in
$$\mathcal{L}=f^2\int{}d^4\theta\langle\mathcal{U}^{\dagger}\mathcal{U}\rangle$$
where
$$\mathcal{U}=e^{it^a\Phi_a/f}$$
where $t^a$ are still the Pauli matrices and $\Phi_a$ are chiral superfields. Now my second question. In the nonsupersymmetric theory we imposed a unitarity constraint in the matrix $U$. I have been told that there is an analogous constraint on $t^a\Phi_a$ but I don't know which it is, let alone where it comes from. So which constraint must I apply and why? My third question would be about how I can relate the pion fields of the nonsupersymetric theory with the superfields I have just introduced, but of course I cannot properly state this question without first applying the constraint on $t^a\Phi_a$ (If the post gets anwsered this will go in a follow up question).