The most natural method for this task would be to use oblivious transfer: on input a bit $x$ for Alice and a bit $y$ for Bob, Alice plays the role of the sender in an oblivious transfer with input $(x_0,x_1) = (0,x)$, and Bob plays the role of the receiver with input $y$, and learns $x_y = x\cdot y$. By repeating this method in parallel for all the bits of their input, Alice and Bob get the bitwise multiplication.
This is probably the most natural and efficient method to perform this task. An alternative is to use ElGamal over a group $\mathbb{G}$ with generator $g$ and public key $h$ for which Alice knows the corresponding secret key.
Alice sends an encryption $(c_0,c_1) = (g^r, h^rg^{x})$ of $g^x$. Bob computes $(c'_0,c'_1) = (c_0g^{r'},c_1^y)h^{r'})$, that is, a rerandomized encryption of $g^{xy}$, and sends it back to Alice, who decrypts and get $g^{xy}$. As $xy \in \{0,1\}$, she easily recover $xy$ from that. Again, by repeating this method in parallel for all the bits of their input, Alice and Bob get the bitwise multiplication.