Thank you, naima.
For "free" field, you are right.
But now in chap. 9 we consider "interacting" field, and ##\phi (0)## is interacting one.
Thus, I do not think we can use eqs. in chap. 3 to prove ##<p| \phi (0) |0>## is Lorentz-invariant.
How about following one?
For ##t=\pm \infty##...