# Embedding of a proper scheme into a smooth one

Let $\mathbb{K}$ be any field and let $X$ be a proper $\mathbb{K}$-scheme. Does it exist a smooth, proper $\mathbb{K}$-scheme $Y$ and a closed immersion from $X$ to $Y$? This is tautological if $X$ is projective, but what about the general case?

• Francesco's answer is completely correct. It does raise another question, that has been much studied: does there exist an immersion from $X$ into a smooth algebraic space? There are still counterexamples if $X$ is allowed to be non-separated, cf. Edidin-Hassett-Kresch-Vistoli. However, to the best of my knowledge, this question is open for separated $X$ (there is a lovely article of Totaro on this question). Sep 30 '15 at 14:40
• Also, this new question has strong motivation: such immersions are used in the main construction of Todd homomorphisms in the singular Grothendieck-Riemann-Roch theorem of Baum-Fulton-MacPherson. Sep 30 '15 at 14:42

For a counterexample over $\mathbb{C}$, it sufficies to take a proper scheme $X$ with trivial Picard group (there are examples among toric varieties).
Such a scheme cannot be embedded in any smooth variety $Y$; in particular, $X$ is necessarily singular.
In fact, assume the contrary and take a dense affine subset $U$ of $Y$. Then we can choose an effective divisor $D_U$ on $U$, whose closure would give a non-trivial Cartier (since $Y$ is smooth) divisor $D_Y$ on $Y$, hence a non-trivial Cartier divisor $D_X$ on $X$, against the assumption $\textrm{Pic}\,X=\{\mathcal{O}_X \}$.
• How does $D_Y$ give a non-trivial divisor on $X$ ? Couldn't $D_Y$ be disjoint from $X$ ? Mar 17 '16 at 9:40
• Well, we can always take an affine chart $U$ of $Y$ containing a point $x \in X$ and an effective divisor $D_U$ passing through $x$. Then the closure of $D_U$ in $X$ cannot be trivial. I'm missing something? Mar 17 '16 at 11:05