Let RR be a , and let AA be a commutative RR-algebra.

An element aAa\in A is over RR if it satisfies a monic polynomial equation

an+rn1an1++r1a+r0=0with riR.a^n + r_{n-1}a^{n-1}+\cdots + r_1 a + r_0 = 0 \quad \text{with } r_i\in R.

The integral closure of RR in AA is

RA  =  {aA:a is integral over R}.\overline{R}^{\,A} \;=\; \{\, a\in A : a \text{ is integral over } R \,\}.

It is a subring of AA containing the image of RR.

When RR is a domain with fraction field KK, the integral closure of RR in KK is often called the normalization of RR. The domain RR is precisely when its integral closure in its fraction field equals RR.

Properties
  • If RAR\subseteq A is an , then every element of AA is integral over RR, hence RA=A\overline{R}^{\,A}=A.
  • If BAB\subseteq A is any subring containing RR and consisting of elements integral over RR, then BRAB\subseteq \overline{R}^{\,A} (maximality of the integral closure).
Examples
  1. Integers inside rationals. Take R=ZR=\mathbb{Z} and A=QA=\mathbb{Q}. If a/bQa/b\in \mathbb{Q} (in lowest terms) is integral over Z\mathbb{Z}, then it satisfies a monic polynomial with integer coefficients, forcing b=±1b=\pm 1. Hence ZQ=Z\overline{\mathbb{Z}}^{\,\mathbb{Q}}=\mathbb{Z}.
  1. A non-normal affine subring. Let kk be a and consider R=k[x2,x3]A=k(x)R=k[x^2,x^3]\subseteq A=k(x) (the rational function field in xx). The element xk(x)x\in k(x) satisfies the monic equation T2x2=0T^2-x^2=0 with x2Rx^2\in R, so xx is integral over RR. Thus the integral closure contains k[x]k[x]. In fact one checks Rk(x)=k[x]\overline{R}^{\,k(x)}=k[x].
  1. Localization. If RR is a domain, SRS\subseteq R is a of nonzero elements, and R\overline R is the integral closure of RR in its fraction field, then the integral closure of S1RS^{-1}R in that field is S1RS^{-1}\overline R. Thus integral closure commutes with .