Rietveld refinement uses a structural and instrument model to explain an entire powder X-ray diffraction pattern. The calculation can provide detailed crystalline information, but its result remains conditional on the phases and parameters included.
The gold standard is a reproducible whole-pattern refinement that keeps calibration, background, profile shape, chemical plausibility and independent tests visible alongside the agreement statistics.
eduKateSG · X-ray Diffraction Rietveld Refinement
Find your next learning step
Start with the diffraction pattern, calculate Rwp and phase fractions, then challenge the assumptions that make a fitted model credible.
Full chapter index · Practice and transfer · Education root · How X Works
Full chapter index
Define whole-pattern evidence · Chapters 1–2
Calculate example results · Chapters 3–4
Control model assumptions · Chapters 5–7
Validate the refinement · Chapters 8–10
Continue X-ray learning · Chapters 11–12
CHAPTER 1 OF 12 · Define whole-pattern evidence
1. Rietveld refinement is a model fit to the full powder pattern
Powder X-ray diffraction combines contributions from many crystallite orientations into intensity measurements as a function of scattering angle or reciprocal-space coordinate. Peaks may overlap strongly, so interpreting the complete measured profile can be more informative than measuring isolated peak heights.
Rietveld refinement calculates a powder pattern from candidate crystal structures, instrument response and other contributions, then adjusts selected parameters to reduce disagreement with the observed data. It is a model-based inference, not a machine that automatically discovers a unique crystal structure.
The method can estimate unit-cell dimensions, relative crystalline phase amounts and selected structural parameters when data and starting models are appropriate. Its capabilities depend on diffraction quality, sample preparation, available structural information and parameter correlations.
The International Union of Crystallography’s 1999 Rietveld refinement guidelines emphasize thoughtful data collection, background handling, peak shapes, uncertainty and interpretation of residual factors. That remains a useful antidote to treating refinement software as a black box.
The gold standard states which phases were included, which quantities were measured directly, which parameters were fitted, the validity range of the model and what independent evidence supports the final materials claim.
CHAPTER 2 OF 12 · Define whole-pattern evidence
2. Pattern calculation has a structure term and an instrument term
A simplified calculated diffraction profile sums contributions from the reflections of every included crystalline phase plus a background. A reflection’s position is related to its lattice spacings, while intensity depends on its structure factor, multiplicity, scattering geometry and phase abundance.
A profile function describes how an ideal reflection is broadened or shaped by the instrument and sample. Peak width and asymmetry are not merely decorative settings: they influence overlaps and hence the fitted scale and lattice parameters.
Preferred crystallographic orientation changes relative intensities compared with an ideal randomly oriented powder. An accurate-looking fit that uses incorrect orientation assumptions can bias inferred phase fractions.
Instrument zero errors or sample displacement can move peaks together, complicating the interpretation of a refined lattice parameter. A reliable analysis therefore needs calibrated geometry and reference materials where appropriate.
NIST’s GSAS tutorial explains how calculated intensity combines structural factors, profile functions, orientation terms, absorption and background. Every part of this expression can affect the fit to observed intensities.
CHAPTER 3 OF 12 · Calculate example results
3. Worked example: Bragg’s law before the refinement
Consider a fictional first-order reflection recorded using monochromatic X-rays of wavelength λ = 1.5406 Å. Its measured peak is centered at total scattering angle 2θ = 40 degrees, so the Bragg half-angle is θ = 20 degrees.
Bragg’s law is λ = 2d sinθ for first order. Therefore d = λ/(2sinθ) = 1.5406/(2sin20°) Å. Since sin20° ≈ 0.34202, d ≈ 1.5406/0.68404 ≈ 2.252 Å.
The result is a lattice-plane spacing under the selected wavelength and reflection model. It is not particle size, grain size or proof of which phase produced the peak.
In a real mixture, peaks from different phases may overlap near one angle. Rietveld refinement uses the full model to distribute calculated intensity across many reflections, which can help when single-peak assignment is ambiguous.
If the instrument’s zero point or wavelength is wrong, the d spacing is systematically wrong as well. Collecting more intensity counts alone cannot repair the geometric calibration.
CHAPTER 4 OF 12 · Calculate example results
4. Worked example: calculate a weighted profile R factor
A commonly reported residual measure is the weighted-profile R factor Rwp = √[Σ wi(yi,obs − yi,calc)² / Σ wi yi,obs²], where yi are profile intensities and wi are suitable weights.
Use three fictitious observed intensities [100, 80, 60] and calculated intensities [98, 83, 59], with equal weights wi = 1 purely for teaching.
Residuals are [2, −3, 1]. Their squared sum is 4 + 9 + 1 = 14. The observed intensity squared sum is 100² + 80² + 60² = 10,000 + 6,400 + 3,600 = 20,000.
Thus Rwp = √(14/20,000) = √0.0007 ≈ 0.02646, or 2.65%. This is an intentionally tiny example, not a realistic full-pattern refinement or proof that the crystal model is correct.
Proper statistical weights often depend on the expected uncertainties of measured intensities. The International Tables for Crystallography warn that Rwp has no universal absolute quality threshold: its numerical value depends on counting statistics, background and the data.
Inspect the residual plotted against scattering angle. A low aggregate number can hide an important unfitted reflection, and a large number may partly reflect extremely precise data rather than a wrong phase interpretation.
CHAPTER 5 OF 12 · Control model assumptions
5. Worked example: what quantitative crystalline phase fractions mean
In an appropriate multiphase Rietveld quantitative phase analysis, the relative weight fraction for an included crystalline phase can be calculated from its refined scale factor S and factors Z, M and V describing formula units, formula mass and unit-cell volume.
A common expression under the stated assumptions is Wp = Sp(ZMV)p / Σi Si(ZMV)i. The units and scale conventions must be those of the chosen refinement method, not arbitrary peak-height ratios.
For a pure classroom example, suppose the two dimensionally comparable products S×ZMV for phases A and B are 2 and 3 in the same adopted proportional units. Their normalized crystalline proportions are A = 2/(2+3) = 40% and B = 3/5 = 60%.
These values are conditional on the identified crystalline phases being complete and correctly modelled. They are not necessarily the fractions of the entire sample if amorphous material, unidentified phases or systematic intensity errors are present.
Hill and Howard’s IUCr paper established this scale-factor approach for quantitative phase analysis under appropriate assumptions. The IUCr’s round-robin study later illustrated how incomplete models and refinement errors can bias published phase quantification.
Do not compute phase percentages directly from isolated peak heights. Multiplicity, atomic scattering, absorption, geometry and overlapping reflections make the raw peak comparison a different quantity.
CHAPTER 6 OF 12 · Control model assumptions
6. Unknown or amorphous material changes the denominator
If the included crystalline phases account for only part of a specimen, normalized phase percentages from those phases describe their relative crystalline contributions, not automatically percentages of total specimen mass.
Imagine independent validated evidence shows a hypothetical sample contains 20% amorphous mass and the remaining 80% consists entirely of crystalline phases A and B in the illustrative 40:60 proportion.
The total-mass fractions would then be A = 0.80×0.40 = 32%, B = 0.80×0.60 = 48%, and amorphous = 20%. These add to 100%, but the 20% must come from appropriate independent evidence or a validated internal-standard method.
Without that evidence, a refinement that forces A and B to sum to 100% may mask unidentified amorphous material or minor unmodelled phases. The neat sum is bookkeeping rather than proof of completeness.
International Tables explicitly notes the tendency to overestimate modeled crystalline fractions when unknown or amorphous components are excluded. Report whether percentages are relative crystalline fractions or genuinely total-sample fractions.
CHAPTER 7 OF 12 · Control model assumptions
7. Background and specimen preparation influence intensity
Raw diffractograms include background from the sample holder, fluorescence, air scatter, diffuse specimen contributions and detector or electronic effects. A fitted background can compete with broad peaks and change the inferred amount of poorly crystalline material.
A powder mount should represent the bulk sample in an appropriate geometry. Grain size, transparency, surface flatness and sample displacement can affect peak shapes and relative intensities, particularly in some reflection geometries.
Uncorrected low-angle footprint changes may suppress expected reflection intensities. IUCr’s guidelines describe this issue for certain diffraction instrument arrangements.
Background should be constrained by physics and data quality rather than distorted until unwanted reflections disappear. A smooth background fit is not a substitute for an accurate structural model.
Retain an independent instrument reference, preparation notes and unprocessed counts. A reproducible refinement begins before any software parameter is adjusted.
CHAPTER 8 OF 12 · Validate the refinement
8. Peak width must separate sample and instrument effects
Observed diffraction-peak width includes broadening from the instrument and sometimes crystallite size, microstrain, defects and distributions of lattice parameters. These effects are not all interchangeable.
Under a toy model in which both contributions are approximately Gaussian and independent, variances can add in quadrature. Suppose an observed full-width measure is 0.20° and the instrumental width measured at a suitable reference is 0.12°.
The simplified specimen contribution is √(0.20² − 0.12²) = √(0.0400 − 0.0144) = √0.0256 = 0.16°. The calculation requires comparable definitions of width and suitable profiles.
The result is a width in degrees, not a crystallite size. Turning it into nanometres requires appropriate size-broadening physics, wavelength, angle, model and, where relevant, separation from microstrain effects.
Many real diffraction profiles are better described by non-Gaussian or mixed functions, so blind quadrature subtraction can be invalid. A reference material and a well-supported line-profile model are essential.
Report the peak-shape function and which parameters were refined. Improvements in fit obtained by adding flexible profile terms need to be checked for physical plausibility and overfitting.
CHAPTER 9 OF 12 · Validate the refinement
9. Texture, strain and peak overlap can distort the structure fit
Preferred orientation may make certain reflection intensities stronger than a randomly oriented powder model predicts. Adding a correction can improve agreement, but it must not become a generic knob used to conceal wrong phases.
Anisotropic peak broadening may arise from differing microstructure or strain along crystallographic directions. A single isotropic width model can then fail systematically across reflections.
Overlapping phases create correlations among scale factors, background and profile widths. A small increase in one phase’s fitted amount may compensate for an incorrectly broadened peak of another.
Refinement constraints and restraints should be physically justified. The 2025 IUCr good-practice guidance for molecular organic crystal structures warns that unconstrained adjustment of many coordinates can improve Rwp while degrading chemically plausible geometry.
Check whether the number of adjustable parameters is supported by the information content. A complicated model with many flexible parameters can match the pattern without uniquely determining a structure.
Compare independent reflection regions, appropriate structural references and, where available, complementary evidence such as spectroscopy, chemical analysis or other diffraction conditions.
CHAPTER 10 OF 12 · Validate the refinement
10. Refinement order and uncertainty are part of reproducibility
A stable workflow commonly begins with instrument and background considerations, then fits phase scales and lattice parameters before gradually introducing appropriate peak profiles and more detailed structural variables. Exact order depends on the specimen and method.
Changing too many correlated parameters at once can conceal which adjustment produced apparent improvement. Save intermediate results, bounds, constraints and reasons for adding model complexity.
Numerical parameter standard errors produced by least-squares software may not include uncertainty from the starting structural model, missing phases, absorption assumptions or specimen preparation.
A rigorous evaluation examines the difference curve, reflection locations, residual structure and chemically plausible final parameters. IUCr’s 1999 guidelines and International Tables make clear that agreement factors are aids to judgement rather than certificates of correctness.
Use calibrated reference samples or interlaboratory comparisons where appropriate. A model that accurately refines a known standard supports some instrument assumptions but cannot guarantee all unknown sample phases were found.
CHAPTER 11 OF 12 · Continue X-ray learning
11. Practice, FAQs and connected diffraction routes
Practice A: For λ = 1.5406 Å and 2θ = 40°, compute d ≈ 2.252 Å. State which angular and wavelength calibration quantities are necessary.
Practice B: With observed [100,80,60] and calculated [98,83,59] at equal weights, verify Rwp ≈ 2.65% and explain why it is not proof of a correct phase assignment.
Practice C: If S×ZMV products are 2 and 3 for complete crystalline phases, calculate 40% and 60% relative crystalline fractions.
Practice D: If independent evidence finds 20% amorphous mass, translate those relative fractions into 32%, 48% and 20% of the total sample.
FAQ — Is Rietveld refinement the same as phase discovery? No. It refines a model; missing or incorrect candidate structures can invalidate conclusions.
FAQ — Does a low Rwp prove correct chemistry? No. Peak and background models can compensate for physically wrong structural assumptions.
FAQ — Does a refinement always quantify amorphous matter? Not automatically. Appropriate additional methods or standards are needed.
Move next from a single clean crystalline phase to a textured multi-phase mixture. List the assumptions that change before relying on reported phase fractions.
Verified eduKateSG routes for related diffraction and inferential methods:
- The Gold Standard of X-ray Diffraction — Bragg physics and peak meaning
- The Gold Standard of X-ray Crystallography — crystal models and symmetry
- The Gold Standard of X-ray Total Scattering and Pair Distribution Function Analysis — a different local-structure inverse problem
- The Gold Standard of X-ray Small-Angle Scattering — distinct scattering scales and size inference
- The Gold Standard of X-ray Diffraction Residual Stress Measurement — specific lattice-strain measurements
- The Gold Standard of Spectroscopy — complementary chemical information
- The Gold Standard of Statistics — fit uncertainty and model testing
CHAPTER 12 OF 12 · Continue X-ray learning
12. Primary sources, scope and the next refinement notebook
International Tables for Crystallography, the IUCr refinement guidelines and NIST tutorials explain the whole-pattern model, quantitative phase analysis and uncertainty problems. Hill and Howard’s phase-fraction relationship and the IUCr round-robin study show why model completeness matters.
All intensities, peaks, widths and scale products above are fictional teaching data. Their arithmetic does not establish any measured crystal, phase percentage, engineering compliance or instrument calibration.
A reproducible refinement notebook should retain raw diffraction data, reference calibrations, phases considered, starting structural models, parameter constraints, background and peak profiles, residual plots, scale-factor definitions and independent validation.
Primary crystallographic and metrological sources:
- IUCr Commission — Rietveld Refinement Guidelines — data, profiles, residuals and reporting
- NIST — GSAS Rietveld Refinement Tutorial — calculated-pattern components and analysis
- International Tables — Rietveld Refinement — weighted-profile agreement and refinement strategy
- IUCr — Hill & Howard Quantitative Phase Analysis — phase weight fraction formula
- IUCr — Quantitative Phase Analysis Round Robin — risks of incomplete refinements
- IUCr (2025) — Organic Structure Refinement Good Practice — restraints and chemical plausibility
The immutable Clementi Secondary 1 Mathematics reference illustrates instructional sequencing, not crystallographic certification.