Give an example of an $R$-module $M$ such that $M=M_1 \oplus M_2$, $M=\operatorname{span}_R(S)$ and $S \cap M_1 = S \cap M_2 = \emptyset$.
I'm not sure if this is a very well known fact, but I can't see why $\mathbb{Z}_6$ works here. If you take $S=\{1\}$ then the intersection with $\mathbb{Z}_2$ and $\mathbb{Z}_3$ is not empty, is it? I'm very confused.
Here $R = \mathbb{Z}$, that is, $\mathbb{Z}_2$ and $\mathbb{Z}_3$ are $\mathbb{Z}$-modules, and $\mathbb{Z}_6 = \mathbb{Z}_2\oplus\mathbb{Z}_3$ as $\mathbb{Z}$-modules (or equivalently, abelian groups). You want $S$ to be a subset of $\mathbb{Z}_2\oplus\mathbb{Z}_3$ (otherwise $\operatorname{span}_{\mathbb{Z}}S$ is not a subset of $\mathbb{Z}_2\oplus\mathbb{Z}_3$), so in order to make sense of the intersections $S\cap \mathbb{Z}_2$ and $S\cap \mathbb{Z}_3$, we identify $\mathbb{Z}_2$ with $\mathbb{Z}_2\oplus\{0\} \subset \mathbb{Z}_2\oplus\mathbb{Z}_3$ and $\mathbb{Z}_3$ with $\{0\}\oplus\mathbb{Z}_3$.
If you take $1 \in S$ to mean $1 + 0 \in \mathbb{Z}_2\oplus\mathbb{Z}_3$, then you do have $S\cap\mathbb{Z}_2 \neq \emptyset$ as claimed, but $\operatorname{span}_{\mathbb{Z}}S = \mathbb{Z}_2\oplus \{0\} \neq \mathbb{Z}_2\oplus\mathbb{Z}_3$. Likewise, if you take $1$ to mean $0 + 1 \in \mathbb{Z}_2\oplus\mathbb{Z}_3$, you do have $S\cap\mathbb{Z}_3 \neq \emptyset$, but again $\operatorname{span}_{\mathbb{Z}}S = \{0\} \oplus\mathbb{Z}_3 \neq \mathbb{Z}_2\oplus\mathbb{Z}_3$.
To see that there is a set $S \subseteq \mathbb{Z}_2\oplus\mathbb{Z}_3$ satisfying $\operatorname{span}_{\mathbb{Z}}S = \mathbb{Z}_2\oplus\mathbb{Z}_3$, and $S\cap\mathbb{Z}_2 = S\cap\mathbb{Z}_3 = \emptyset$, consider the element $a + b \in S$ (clearly $S \neq \emptyset$). If $a = 0$, then the element belongs to the intersection $S\cap\mathbb{Z}_3$; likewise, if $b = 0$, the element belongs to the intersection $S\cap \mathbb{Z}_2$. Therefore all the elements of $S$ must be of the form $a + b$ with $a, b \neq 0$. This gives three possibilities for $S$, all of which have the desired properties.