X-ray polarimetry measures polarization information that complements the energy, arrival time and position of detected X-rays. In astronomy, that information helps test models of radiation and the geometry of energetic sources.
The gold standard separates the detector’s angular response from the inferred polarization and the physical source model. Begin with a calibrated modulation, follow it into Stokes quantities and retain uncertainty when interpreting a weak signal.
eduKateSG · X-ray Polarimetry
Find your next learning step
Understand how angular data constrain polarization, calculate simple examples and test the limits of a combined observation.
Full chapter index · Practice and transfer · Education root · How X Works
Full chapter index
Define the evidence · Chapters 1–2
Calculate the response · Chapters 3–4
Combine and check · Chapters 5–7
Interpret and transfer · Chapters 8–10
Choose further reading · Chapters 11–12
CHAPTER 1 OF 12 · Define the evidence
1. Polarization adds a property of the radiation
For linear polarization, the degree describes the strength of an ordered orientation component and the position angle describes its orientation relative to a stated reference. These quantities add information beyond source brightness.
An angle must name its coordinate convention. Linear polarization has an orientation rather than a directed arrow: angles separated by 180° represent the same orientation.
Begin with the question the observation is meant to test. A model might predict a different polarization across energy, time or position. Define those selections before interpreting one combined number as the complete behaviour of a source.
CHAPTER 2 OF 12 · Define the evidence
2. Photoelectron directions provide a statistical observation
IXPE’s gas pixel detectors record tracks produced when incoming X-rays interact in a gas. The distribution of reconstructed initial photoelectron directions contains information about the incident linear polarization.
An individual track is an event with measurement uncertainty. The polarization estimate comes from a statistical collection, together with a calibrated response. One track does not establish the source’s polarization angle.
Retain the detector, energy selection, time interval and source region. Those choices determine which collection of radiation the result describes and which response information belongs with it.
CHAPTER 3 OF 12 · Calculate the response
3. Worked example: modulation is calibrated by response
In an idealised angular-response model, N(φ) = A[1 + μp cos2(φ − ψ)], where p is linear polarization degree, ψ its angle and μ the modulation factor for fully polarized radiation under the stated conditions.
The fractional modulation is m = (Nmax − Nmin)/(Nmax + Nmin) = μp. For an invented smooth expected curve with maximum 560 and minimum 440, m = 120/1,000 = 0.12. If μ = 0.40, then p = 0.12/0.40 = 0.30, or 30%.
These are model expectations, not a recommended estimator based on the largest and smallest noisy bins. A real analysis uses the event statistics and appropriate response. The instrument modulation factor is essential to the conversion.
CHAPTER 4 OF 12 · Calculate the response
4. Worked example: calibrated Stokes quantities
Define response-corrected, normalised linear Stokes quantities q = p cos2ψ and u = p sin2ψ in a common reference frame. Then p = √(q² + u²) and ψ = ½atan2(u,q), modulo 180°.
Take q = 0.12 and u = 0.16. The degree is √(0.0144 + 0.0256) = 0.20, or 20%. The angle is half of atan2(0.16,0.12), about 26.6°.
The two-argument atan2 function keeps the quadrant. These q and u values are already calibrated source quantities for this teaching example. Raw event summaries, weighted products and instrumental pseudo-Stokes parameters need their own documented transformations.
CHAPTER 5 OF 12 · Combine and check
5. Combine Stokes components before deriving an angle
For independent incoherent radiation components expressed in the same frame, the intensity-weighted Stokes components add. Polarization degrees and angles should not simply be averaged as ordinary scalar values.
Consider two equal-intensity components, each 50% linearly polarized. One has ψ = 0° and the other ψ = 90°. Their normalised q values are +0.5 and −0.5, while both u values are zero. The combined q and u are zero.
The combined linear polarization is therefore zero, even though each component is polarized. The arithmetic mean angle of 45° would give a misleading picture. Spatial or temporal averaging can conceal structured polarized emission.
In a simple model with unpolarized background, let f be the fraction of accepted counts coming from the source. With a constant modulation factor and compatible selections, the observed modulation is m = μpf.
For m = 0.08, μ = 0.40 and f = 0.80, the inferred source degree in this ideal model is p = 0.08/(0.40 × 0.80) = 0.25. Omitting the dilution would give 0.20.
The example assumes a uniform unpolarized background and a known source fraction. A real extraction needs suitable background scaling and response treatment. A background region is evidence to model, not a quantity that can be subtracted without accounting for its selection.
The detector’s response varies with energy, and the source’s polarization can vary too. Combining a broad energy band therefore requires appropriate response weighting or a spectral model. One modulation factor chosen without that context may be insufficient.
Similarly, changing a sky region can change the mixture of source components and background. Changing a time interval can combine different physical states.
A useful report keeps the selection and response beside the result. When comparing two published numbers, inspect those definitions before treating their difference as source variability.
| Reported quantity | Required context | Check |
|---|---|---|
| Polarization degree | Response and statistical treatment | Is a positive estimate significant? |
| Position angle | Coordinate frame and convention | Were orientations combined correctly? |
| Broadband estimate | Energy range and response weighting | Could different components cancel? |
| Regional estimate | Source selection and background | Which radiation mixture was measured? |
CHAPTER 8 OF 12 · Interpret and transfer
8. Sensitivity is different from a measured degree
For an ideal source-dominated counting case with constant effective modulation factor μ, the 99% minimum detectable polarization is approximately MDP99 = 4.292/(μ√N), with N the appropriate source counts.
For N = 10,000 and μ = 0.40, MDP99 ≈ 0.1073, or 10.73%. With 40,000 otherwise comparable counts, it falls to about 5.37%. Four times the counts halves this sensitivity threshold.
MDP describes a detection criterion under the unpolarized null model; it is not the measured degree or its error bar. Background, weighting and systematic effects require the fuller analysis. Use the correct effective counts rather than assuming every recorded event supplies equal information.
CHAPTER 9 OF 12 · Interpret and transfer
9. A weak estimate needs its statistical boundary
The degree √(q² + u²) is nonnegative. Noise can therefore yield a positive estimate even when the true linear polarization is zero. A positive plotted value alone does not establish a detection.
Near a non-detection, the angle can be poorly constrained. Report the applicable confidence information or limit and the analysis method, rather than assigning physical meaning to an unstable angle.
For astrophysical interpretation, connect the polarization result with a radiation and geometry model. The relation between an observed angle and a magnetic-field orientation depends on the emission and propagation physics; one universal arrow rule is insufficient.
First, use a model modulation m = 0.06 with μ = 0.30 and negligible background. Second, calculate p and ψ for calibrated q = 0 and u = 0.10. Third, predict the change in the ideal MDP when comparable source counts increase by a factor of nine.
Checks: p = 0.20 in the first case. The second has p = 0.10 and ψ = 45° in the stated frame. The ideal MDP falls by a factor of three because √9 = 3.
Now combine equal-intensity polarized components whose angles differ by 90°. Explain why a low combined degree cannot alone prove disordered emission everywhere. The missing information may be the spatial or temporal separation of components.
CHAPTER 11 OF 12 · Choose further reading
11. Connected astronomy and detector guides
Place polarization in cosmic evidence
- X-ray Astronomy — connect counts, spectra, backgrounds and source interpretation.
- X-ray Science — place polarization within the wider set of X-ray measurements.
Follow the instrument response
- Predicting detector channels from a mixed-energy response — practise mapping source radiation into what a detector records.
- X-ray Spectroscopy — compare energy-dependent measurements and their calibration requirements.
Choose the astronomy route when the question concerns what a source model predicts. Choose the detector-response route when the difficulty is how measured channels or event collections relate to the incident radiation.
CHAPTER 12 OF 12 · Choose further reading
12. Primary references and the next learning step
- NASA IXPE: polarization detection — photoelectron tracks and angular-distribution measurements.
- NASA IXPE: sensitivity calculations — modulation factor, effective counts and minimum detectable polarization.
- IXPE Quick Start Data Analysis Guide — Stokes analysis, responses, selections and statistical interpretation.
All numerical values are invented for teaching. They describe ideal models rather than a new observation of an astronomical source. The equations do not replace the calibrated analysis for an actual dataset.
The learning checkpoint is to connect the detected angular information, calibrated polarization and physical model without skipping uncertainty. Continue through the Education root to practise that reasoning when the detector or scientific question changes.