← A-level

Further Core Pure

Improper Integrals

A definite integral is called improper when one or both limits of integration are infinite, or when the integrand has a singularity — a point where it blows up — inside or at the boundary of the interval. Neither situation is a dead end. The strategy in both cases is the same: replace the problematic point with a parameter and ask whether the integral tends to a finite limit as the parameter approaches the problem.

Improper integrals sit in compulsory Further Pure at all four main boards. Evaluating the limits that arise often requires L'Hôpital's rule — the two topics appear together frequently.

The core idea: replace, integrate, take the limit

Replace the problematic endpoint (∞, −∞, or a singularity) with a finite parameter t. Integrate normally. Then take the limit as t approaches the problem point.

If the limit exists and is finite, the integral converges to that value. If the limit is ±∞ or does not exist, the integral diverges.

Type 1 — infinite limits of integration

A-level Further Maths — Further Pure (compulsory)Edexcel (Core Pure 2)AQA (Further Pure 2)OCR (Pure Core 2)OCR MEI (Core Pure B)
Example 1 — a convergent integral

The integral converges to 1.

Try it yourself

Evaluate ∫1 x−3 dx.

Show answer
Example 2 — a divergent integral

The limit is infinite, so the integral diverges — even though 1/x becomes arbitrarily small. The function decreases, but not fast enough for the area to remain finite.

Try it yourself

Does ∫1 x−1/2 dx converge or diverge? Use the p-integral result to justify your answer.

Show answer

This is the p-integral with p = ½. Since p = ½ ≤ 1, the integral diverges.

The p-integral — a benchmark result

The single family of integrals ∫1 1/xp dx neatly separates convergence from divergence:

For p = 1 the antiderivative is ln x (as in Example 2), which diverges separately. For p < 1, x1−p grows without bound as t → ∞. The borderline p = 1 is worth remembering: the harmonic series and its integral both diverge, despite the terms tending to zero.

Type 2 — singularities in the integrand

A-level Further Maths — Further Pure (compulsory)Edexcel (Core Pure 2)AQA (Further Pure 2)OCR (Pure Core 2)OCR MEI (Core Pure B)

When f(x) → ±∞ as x → b (a boundary of the interval), replace b with a parameter approaching from inside the interval:

Example 3 — singularity at the lower limit

Although 1/√x → ∞ as x → 0, the singularity is mild enough that the area remains finite. The integral converges to 2.

Try it yourself

Evaluate ∫01 x−1/3 dx.

Show answer

p = 1/3 < 1, so the integral converges — consistent with the p-integral result.

Example 4 — a divergent singularity

The singularity is too strong; the integral diverges.

Try it yourself

Does ∫01 x−3/2 dx converge? Justify using the p-integral result.

Show answer

p = 3/2 ≥ 1, so the integral diverges.

The p-integral near zero — the mirror result

Notice the condition flips relative to the infinite-limit case: near zero, mild singularities (p < 1) are integrable; near infinity, functions must decay faster than 1/x (p > 1) to be integrable.

Where L'Hôpital's rule comes in

After integrating, you are left with a limit to evaluate. When the antiderivative involves a product of a polynomial and an exponential or logarithm, the limit as t → ∞ is often an indeterminate form — exactly the situation L'Hôpital's rule is designed for.

Example 5 — integration by parts then L'Hôpital

Evaluate ∫0 x e−x dx.

Integrate by parts with u = x, dv/dx = e−x:

Now take the limit as t → ∞. The term −e−t → 0. For −te−t:

Therefore:

Try it yourself

Evaluate ∫0 x2e−x dx. (Hint: integrate by parts twice, or use the result from Example 5.)

Show answer

Integrate by parts with u = x², dv = e−x dx:

As t → ∞, all terms with e−t vanish (L'Hôpital):

Example 6 — logarithm growth killed by a power

Evaluate ∫1 ln(x)/x2 dx.

Integrate by parts with u = ln x, dv/dx = x−2:

As t → ∞, −1/t → 0. For the logarithm term, L'Hôpital gives:

The general pattern: any power of t grows slower than et, and any power of ln t grows slower than any positive power of t. L'Hôpital makes these comparisons precise with a single differentiation.

Try it yourself

Evaluate ∫1 ln(x) / x3 dx.

Show answer

Integrate by parts with u = ln x, dv = x−3 dx:

As t → ∞, ln(t)/t² → 0 (L'Hôpital) and 1/t² → 0, so:

Integrals with two problems — split first

When an integral has problems at both ends, or at an interior point, it must be split into separate improper integrals at a convenient intermediate point. Both parts must converge independently for the whole integral to converge. It is not valid to let a single limit simultaneously handle both ends.

Integral over all of ℝ

Each half is a separate improper integral. Both must converge.

Example 7 — an integral that appears to cancel but diverges

Consider ∫−11 1/x dx. The integrand is odd, so one might expect the positive and negative areas to cancel, giving 0. They do not.

The left half already diverges. The integral does not exist — the apparent cancellation is an illusion. Symmetry does not rescue a divergent integral.

Try it yourself

Does ∫−11 x−1/3 dx converge? If so, evaluate it. Split at x = 0 and treat each half separately.

Show answer

Singularity at x = 0. For each half, the exponent 1/3 < 1, so both halves converge.

The second half is negative because x−1/3 is negative for x < 0. Unlike 1/x, both halves converge and symmetry does give the total integral = 0.

Example 8 — an interior singularity

03 1/(x−1)2/3 dx has a singularity at x = 1, inside the interval. Split there:

The antiderivative is 3(x−1)1/3. Evaluating each half:

The singularity at x = 1 looks alarming but the exponent 2/3 < 1 means it is mild enough to integrate — directly analogous to the 1/√x case in Example 3.

Try it yourself

Evaluate ∫−11 x−2/3 dx. The integrand has a singularity at x = 0; split there.

Show answer

p = 2/3 < 1, so both halves converge. The antiderivative of x−2/3 is 3x1/3.

Summary

TypeCondition for convergenceWatch out for
1 1/xp dxp > 1p = 1 is the critical borderline — diverges
01 1/xp dxp < 1Condition flips relative to the infinite-limit case
∫ polynomial × e−xAlways converges on [0, ∞)Boundary term tne−t → 0: use L'Hôpital
∫ ln(x)/xp, p > 0Converges on [1, ∞)Boundary term ln(t)/tp → 0: use L'Hôpital
Two problem pointsBoth halves converge independentlyDo not let symmetry replace a proper limit calculation
Interior singularityBoth halves converge independentlySplit at the singularity — do not integrate across it blindly
Ready to start?
Book an online lesson
Book a lesson