In Carroll's Spacetime and geometry section 7.5, it is deriving metric $\bar h_{\mu \nu}$ with Lorenz gauge in terms of quadrupole moment of energy density.
In Eq.[7.135],
\begin{equation} \int d^3y \tilde T^{ij} = \int \partial_k(y^i T^{kj}) d^3y - \int y^i (\partial_k T^{kj}) d^3y \end{equation}
It says "The first term is a surface integral which will vanish since the source is isolated". But I could not see the form of surface integral for the first term. Is it divergence theorem? I know divergence is $\partial_k T^k$, how does it relate to $\partial_k(y^i T^{kj})$? It seems to me that $y^i$ is redundant. If it is $\partial_k(T^{kj})$, it works and converts as $\int T^{kj}n^kd^2y$.
Moreover, to my understanding, the source is emitting energy through gravitational wave. How could it said to be isolated?