I am stuck at a seemingly simple assertion: In a $2$-to-$2$ scattering, Mandelstam invariant $t<0$. Here we use $(+,-,-,-)$ convention. I know that by definition, $t=(p_1-p_3)^2=(E_1-E_3)^2-(\vec{p_1}-\vec{p_3})^2$. If I can set $E_1=E_3$, then the result follows. I can set that for the case that all the masses are equal, for example. But apparently, this is valid even when the masses are unequal. How?
