Deriving the Secant Substitution for x²−a²

Why a Secant Substitution Fits This Radical

The expression \(x^2-a^2\) has a difference-of-squares structure. The trigonometric identity \(\sec^2\theta-1=\tan^2\theta\) has the same structure, so it can convert the radical into a multiple of a tangent. The substitution is useful only when its angle range is chosen to match the real domain and sign of \(x\).

Throughout, assume \(a>0\). Since the square root denotes the nonnegative root, \(\sqrt{x^2-a^2}\) is real exactly when \(x^2-a^2\geq 0\), or \(|x|\geq a\). The open intervals \(x>a\) and \(x<-a\) are the two regions where the radicand is positive. At \(x=\pm a\), the radical is zero; these boundary points may be included when an integral is defined there, but they are not interior points of either substitution branch.

Deriving the Relationships from a Right Triangle

Begin with \(x>a>0\). Imagine a right triangle whose hypotenuse has length \(x\), whose adjacent leg has length \(a\), and whose opposite leg has length \(\sqrt{x^2-a^2}\). The Pythagorean theorem verifies that these lengths fit:

Let \(\theta\) be the acute angle between the adjacent leg and the hypotenuse. By the right-triangle definitions of secant and tangent, hypotenuse divided by adjacent side is \(\sec\theta\), and opposite side divided by adjacent side is \(\tan\theta\). Thus

Rearranging gives \(x=a\sec\theta\) and \(\sqrt{x^2-a^2}=a\tan\theta\). The triangle determines the positive signs here: all three side lengths are positive, and \(\theta\) is acute.

The same result follows directly from the identity \(\sec^2\theta-1=\tan^2\theta\). Substituting \(x=a\sec\theta\) gives

Taking square roots requires attention to signs. In general, \(\sqrt{a^2\tan^2\theta}=a|\tan\theta|\), because \(a>0\) and the principal square root is nonnegative. On the acute branch, \(\tan\theta>0\), so this is \(a\tan\theta\), as claimed. The absolute value is not optional if the angle can lie in another quadrant.

Deriving the Differential

Differentiate \(x=a\sec\theta\) with respect to \(\theta\). The constant \(a\) remains unchanged, and the derivative of \(\sec\theta\) is \(\sec\theta\tan\theta\). Therefore

This derivative is positive on \(0<\theta<\pi/2\). Consequently \(x=a\sec\theta\) increases from \(a\) toward infinity as \(\theta\) increases from \(0\) toward \(\pi/2\). Each \(x>a\) corresponds to exactly one angle in this interval, so the substitution does not fold the positive domain back over itself.

Worked Examples

Two Valid Ways to Handle the Negative Branch

For \(x<-a\), the radical is real, but the acute-angle formulas cannot be used unchanged with \(x=a\sec\theta\): an acute angle has positive secant, whereas \(x/a<0\). There are two consistent choices.

• Keep \(x=a\sec\theta\): take \(\pi/2<\theta<\pi\). Then \(\sec\theta<0\), \(\tan\theta<0\), and \(x=a\sec\theta\). Since the radical must be nonnegative, \(\sqrt{x^2-a^2}=-a\tan\theta\). The derivative remains \(dx=a\sec\theta\tan\theta\,d\theta\), which is positive because both trigonometric factors are negative.
• Use an acute angle: write \(x=-a\sec\phi\) with \(0<\phi<\pi/2\). Then \(\sqrt{x^2-a^2}=a\tan\phi\) and \(dx=-a\sec\phi\tan\phi\,d\phi\). Here \(x\) decreases as \(\phi\) increases, which is reflected by the negative differential.

Either choice describes the negative component of the real domain, \((-\infty,-a)\). Choose one convention for a calculation and use it consistently for the variable, radical, differential, and angle limits. A substitution that is valid on one component should not be assumed valid across the gap \([-a,a]\), where the radical is not real except at its endpoints.

Common Sign Error

A frequent mistake is to write \(\sqrt{x^2-a^2}=a\tan\theta\) without checking the sign of \(\tan\theta\). For the example \(a=3\), \(x=-5\), and \(\theta\) in quadrant II, this incorrect calculation gives

But a principal square root cannot equal \(-4\); direct evaluation gives \(\sqrt{(-5)^2-3^2}=4\). The correction is to use \(\sqrt{x^2-a^2}=a|\tan\theta|\), or, on this specific quadrant-II range, \(\sqrt{x^2-a^2}=-a\tan\theta=4\). On the positive branch, the tangent is positive, so the familiar formula \(a\tan\theta\) is correct.

When the Derivation Is Useful

The secant substitution is useful when an expression involving \(\sqrt{x^2-a^2}\) appears and \(x\) lies in one of the real intervals outside \([-a,a]\). It converts the quadratic difference into a trigonometric square, and its differential supplies the factors needed to rewrite an integrand. The important work is not merely replacing \(x\): the domain, angle range, sign of the radical, and differential all belong to the substitution.