Zero Divisors and Integral Domains

Integral domains are the backbone of algebraic number theory, where rings of algebraic integers (like $\mathbb{Z}[\sqrt{-5}]$) are studied precisely because they are domains, and unique factorization questions (do they behave like $\mathbb{Z}$, or fail as in $\mathbb{Z}[\sqrt{-5}]$ where $6 = 2\cdot 3 = (1+\sqrt{-5})(1-\sqrt{-5})$?) drive the theory of ideals and class groups. In algebraic geometry, the coordinate ring of an irreducible affine variety is exactly an integral domain, and zero divisors detect reducibility — a variety splits into pieces precisely when its coordinate ring acquires zero divisors. In cryptography, the arithmetic of $\mathbb{Z}/n\mathbb{Z}$ for composite $n$ (as in RSA) is deliberately not a domain, and the zero divisors correspond exactly to elements sharing a factor with $n$; detecting such zero divisors is equivalent to factoring $n$, which is the hardness assumption underlying RSA's security. In numerical linear algebra and computer science, singular (non-invertible) matrices are zero divisors in matrix rings, and understanding when products of matrices vanish informs rank computations and error analysis. In coding theory, cyclic codes are ideals in quotient rings $\mathbb{F}_q[x]/(x^n-1)$, and zero divisors in this ring directly correspond to the code's algebraic structure and factorization of $x^n - 1$.

:::quiz[In $R = \mathbb{Z}[x]/(x^2-4)$, why does $(\alpha-2)(\alpha+2) = 0$ fail to make $R$ a domain, even though $\mathbb{Z}[x]$ itself is a domain?]

• Being a domain is not preserved by quotienting by every ideal; here $x^2-4$ factors nontrivially ✓
• $\mathbb{Z}[x]$ was never a domain to begin with
• The quotient map is not well-defined
• $\alpha$ is not a real element so the computation is invalid

:::faq[Is the polynomial ring $R[x]$ over an integral domain $R$ always an integral domain?] Yes. If $f, g \in R[x]$ are nonzero polynomials with leading coefficients $a_m \neq 0$ and $b_n \neq 0$ (degrees $m,n$ respectively), then the leading coefficient of $fg$ is $a_m b_n$. Since $R$ is a domain and $a_m, b_n$ are both nonzero, $a_m b_n \neq 0$, so $fg$ has degree exactly $m+n$ and in particular $fg \neq 0$. This argument crucially uses that $R$ itself has no zero divisors; it is precisely why $\mathbb{Z}[x]$ is a domain but, for instance, $(\mathbb{Z}/4\mathbb{Z})[x]$ is not (since $\mathbb{Z}/4\mathbb{Z}$ already has the zero divisor $2$).

Ask the tutor: Why does an integral domain always embed into a field, and how does the field of fractions construction actually work?

Think of a ring as a universe of "numbers" where addition and multiplication make sense, but where the multiplicative structure can be far stranger than in ordinary arithmetic. In the integers, multiplying two nonzero numbers always produces a nonzero number — you cannot "cancel out" nonzero quantities against each other. This is such a deeply ingrained fact that it is easy to forget it is a hypothesis, not a theorem true of all rings. The clock arithmetic of $\mathbb{Z}/6\mathbb{Z}$ reveals the failure vividly. On a clock with $6$ hours, $2$ "hours" multiplied by $3$ "hours" lands exactly on $0$. Neither factor is zero, yet their product collapses to nothing. This happens precisely because $6 = 2\times 3$ is composite: the factors of $n$ conspire to produce zero divisors in $\mathbb{Z}/n\mathbb{Z}$. When $n$ is prime, no such factorization is available, and the ring becomes an integral domain — in fact, a field. Zero divisors also explain why polynomial equations can behave unexpectedly modulo composite numbers. Over $\mathbb{Z}/6\mathbb{Z}$, the equation $x(x-3) = 0$ has four solutions ($x = 0, 2, 3, 5$), not the usual two you'd expect from a degree-2 polynomial over a domain. Once zero divisors enter the picture, the elementary link between "degree" and "number of roots" — a link that relies on unique factorization into linear factors — breaks down.

Note the proof technique in each direction is direct verification from the axioms: the forward direction uses only distributivity and the definition of "no zero divisors"; the reverse direction uses the elementary lemma $a \cdot 0 = 0$ (itself proved directly from distributivity, not assumed) together with the hypothesis. Neither direction presupposes properties of fields or unique factorization, so the argument is fully self-contained.

Example 1: $\mathbb{Z}/12\mathbb{Z}$ — locating all zero divisors. Let $R = \mathbb{Z}/12\mathbb{Z}$, the integers mod $12$, a commutative ring with $1$. We determine exactly which nonzero elements are zero divisors, using elementary number theory (no assumption about domains is used — we compute directly). Claim: $a \in \{1, \dots, 11\}$ is a zero divisor in $\mathbb{Z}/12\mathbb{Z}$ if and only if $\gcd(a, 12) > 1$. Proof of claim. Suppose $d = \gcd(a,12) > 1$. Let $b = 12/d$. Since $d > 1$, we have $1 \le b < 12$, so $b \not\equiv 0 \pmod {12}$, i.e. $b \neq 0$ in $R$. Also $a \cdot b = a \cdot (12/d)$. Since $d \mid a$, write $a = d a'$; then $ab = d a' \cdot (12/d) = a' \cdot 12 \equiv 0 \pmod{12}$. So $ab = 0$ in $R$ with $b \neq 0$: $a$ is a zero divisor. Conversely, suppose $\gcd(a,12) = 1$. We show $a$ is not a zero divisor, in fact that $a$ is a unit, which certainly implies it is not a zero divisor (if $a$ is a unit with inverse $a^{-1}$ and $ab=0$, then multiplying both sides by $a^{-1}$ gives $b = 0$). By Bézout's identity, since $\gcd(a,12)=1$, there exist integers $x,y$ with $ax + 12y = 1$. Reducing mod 12, $ax \equiv 1 \pmod{12}$, so $a$ has a multiplicative inverse $x$ in $R$, hence $a$ is a unit and not a zero divisor. This proves the claim. Applying this: the integers from $1$ to $11$ with $\gcd(a,12) > 1$ are $2,3,4,6,8,9,10$ (each shares a factor $2$ or $3$ with $12$), and these are precisely the zero divisors. The remaining elements $1, 5, 7, 11$ are units (each coprime to $12$). For instance, check directly: $4 \cdot 3 = 12 \equiv 0$, confirming $4$ and $3$ are zero divisors; and $5 \cdot 5 = 25 \equiv 1 \pmod{12}$, confirming $5$ is its own inverse, hence a unit. This example shows $\mathbb{Z}/12\mathbb{Z}$ is not an integral domain (it has zero divisors, e.g. $3\cdot 4=0$), and moreover illustrates the general fact (foreshadowing Cancellation and Characteristic) that in $\mathbb{Z}/n\mathbb{Z}$, an element is a unit iff it's coprime to $n$, and every nonzero non-unit is a zero divisor.

Example 2: The ring $R = \mathbb{Z}[x]/(x^2 - 4)$ — a non-domain disguised as a polynomial ring. Let $R = \mathbb{Z}[x]/(x^2-4)$. Write $\alpha$ for the image of $x$ in $R$, so $\alpha^2 = 4$ in $R$, and every element of $R$ has a unique representative of the form $a + b\alpha$ with $a, b \in \mathbb{Z}$ (this is because dividing any integer polynomial by the monic polynomial $x^2-4$ leaves a remainder of degree $<2$, i.e., of the form $a+bx$, and the division algorithm gives uniqueness). Consider the two elements

Both are nonzero in $R$: if $\alpha - 2 = 0$ then $\alpha = 2$ as elements of $R$, but the coset representative of $\alpha$ is $0 + 1\cdot \alpha$ while $2$'s representative is $2 + 0\cdot \alpha$, and these differ (by uniqueness of representatives, since $1 \neq 0$ and $-2 \ne 0$ don't match), so $\alpha \ne 2$ in $R$; similarly $\alpha \neq -2$. Now compute their product using $\alpha^2 = 4$:

So $u, v$ are both nonzero, yet $uv = 0$: $R$ has zero divisors, and is not an integral domain — even though it is built as a quotient of the polynomial ring $\mathbb{Z}[x]$, which is an integral domain. This example is instructive: the property of being an integral domain is not automatically inherited by quotient rings. (It survives quotienting by a prime ideal, a fact developed alongside Field of Fractions theory, but $(x^2-4) = (x-2)(x+2)$ is not prime in $\mathbb{Z}[x]$ since it factors nontrivially — consistent with what we just found.) By contrast, $\mathbb{Z}[x]/(x^2+1)$ (the Gaussian integers $\mathbb{Z}[i]$) is an integral domain, because $x^2+1$ has no roots in $\mathbb{Z}$ and remains irreducible; the difference between $x^2-4$ and $x^2+1$ — reducible versus irreducible — is exactly what controls the presence of zero divisors in the quotient.