X-ray Fluorescence (XRF) Spectroscopy: How to Prepare Samples, Analyze Data, and Troubleshooting

X-ray fluorescence (XRF) spectroscopy is one of the most versatile elemental analysis techniques available to a modern laboratory. It examines solids, powders, liquids, slurries, and thin films without the acid digestion that many other elemental methods demand, and it reports elements from beryllium through americium across a concentration range that runs from major components at tens of percent down to trace levels in the low parts-per-million region. Because measurements are rapid and, in many cases, leave the sample intact, XRF has become a routine tool in materials research, geology, metallurgy, polymer science, and pharmaceutical development. This article explains how the technique works, how to prepare samples that produce trustworthy numbers, how to analyse and interpret the resulting spectra, and how to troubleshoot the problems that most often lead to questionable data.

Understanding X-ray Fluorescence Spectroscopy

Principles of X-ray Fluorescence

When a sample is irradiated with a beam of high-energy X-rays, atoms near the surface absorb photons and eject electrons from their inner shells. The resulting vacancy is unstable, and an electron from a higher-energy shell drops down to fill it. The energy released in that transition leaves the atom as a secondary, or fluorescent, X-ray photon whose energy equals the difference between the two shells involved. Because every element has a unique set of shell energies, the emitted photons carry a characteristic fingerprint — a set of energies that identifies the element unambiguously. This relationship is captured via Moseley's law, which links the frequency of a characteristic line to the atomic number of the emitting element.

The intensity of each characteristic line, meanwhile, is proportional to how much of that element is present, although the proportionality is never a simple constant. Before fluorescent photons escape the sample they can be absorbed by other atoms, and they can themselves excite further atoms to produce secondary fluorescence and enhancement. These absorption and enhancement phenomena, collectively called matrix effects, are the central analytical challenge in XRF and the reason why sample preparation and data correction matter so much. A laboratory that controls the matrix — by making the sample uniform and by choosing a suitable calibration — obtains accurate results, while one that ignores it merely reports plausible-looking numbers.

Instrument Configurations: ED-XRF vs. WD-XRF

Two families of instrumentation dominate the field, and they differ mainly in how the fluorescent signal is separated into its component energies. Energy-dispersive XRF (ED-XRF) uses a semiconductor detector that sorts incoming photons by their energy directly, producing a complete spectrum during a single acquisition. Wavelength-dispersive XRF (WD-XRF) instead relies on analysing crystals that diffract photons of different wavelengths at different angles, which are counted sequentially by a detector. The choice between them shapes sample throughput, spectral resolution, and the range of applications a laboratory can serve.

In practice, ED-XRF is compact, economical, and fast, which makes it well suited to screening, portable field work, mapping of heterogeneous surfaces, and rapid multi-element surveys. WD-XRF offers markedly better energy resolution and lower background, so it becomes the preferred platform when closely spaced lines must be resolved, when light elements need careful quantification, or when the highest precision is required for major-component analysis. The two are complementary rather than competing, and many full-service laboratories operate both so that each measurement can be assigned to the instrument that suits it best.

Table.1 Energy-Dispersive and Wavelength-Dispersive XRF Compared.

Elemental Range, Detection Limits, and Information Depth

XRF covers nearly the entire periodic table, from beryllium to americium in most configurations, with detection limits that span an enormous range. Major and minor elements present at percent levels are measured with excellent precision, while trace elements can be detected in the low parts-per-million (10^-6) range with longer acquisition times or preconcentration. The technique does not reach the parts-per-billion or parts-per-trillion levels that some other elemental methods provide, and it reports the total concentration of each element without distinguishing between oxidation states or mineral forms. Where phase identification rather than elemental quantification is the goal, XRD testing supplies the complementary information.

A second practical property is information depth. Because fluorescent photons are attenuated on their way out of the sample, the depth from which signal originates depends on photon energy and on sample density and composition. For light elements the information depth may be only a few micrometres, whereas for heavier elements in a light matrix it can extend to a millimetre or more. Light elements such as sodium, magnesium, and fluorine produce very low-energy photons, so their measurement demands a vacuum path or a helium purge to avoid absorption by air. These facts shape sample preparation directly, because a measurement that samples only the top few micrometres is exquisitely sensitive to surface contamination, oxidation, and polishing quality.

XRF Data Analysis Workflows

Once a spectrum has been acquired, the journey from raw counts to a defensible number passes through four stages, and each one answers a different question. XRF is one of several spectroscopy testing tools in a modern analytical laboratory, and its data workflow begins with identifying which elements are present, moves through quantifying them, and ends with an honest statement of how much confidence the numbers deserve. Skipping a stage rarely produces an obvious error; it produces a result that looks precise and cannot be defended.

Spectrum Interpretation and Peak Identification

An XRF spectrum is a plot of detected photon intensity against energy, and its peaks are the signature of the elements present. Each peak sits at an energy characteristic of a specific electron transition, so identification is fundamentally a matching exercise. Three checks turn a plausible match into a confident identification:

• Peak energy: the measured centroid must agree with the tabulated line energy within the instrument's energy resolution, which is why the calibration is verified before any peak is assigned.
• Full line set: an element is only credible when all of its expected lines appear together, since a single matching peak may simply be a minor line of another element.
• Intensity ratios: the K_α and K_β lines of an element normally appear in a characteristic ratio, and the L-series lines follow their own pattern, so a distorted ratio points to a mismatch or a hidden overlap.

Heavier elements generate richer spectra with multiple lines spread across the energy range, whereas light elements produce few lines at low energy and are therefore easier to miss when the measurement conditions do not suit them. Judging the spectrum as a whole, rather than peak by peak, is what separates a confident identification from a guess.

Because every identification rests on matching a measured energy against a known one, the line positions themselves are the working reference of the technique, and a single measurement can report dozens of elements at once precisely because their lines fall at different points of the spectrum. Those positions are quoted by energy in kiloelectron volts (keV), while wavelength-dispersive instruments describe the same photons by their equivalent wavelength in angstroms (Å), the two simply being reciprocal expressions of one photon energy: a higher energy always means a shorter wavelength and a more penetrating photon. Two line families do most of the work in practice. The K lines are the most intense available lines and serve elements up to roughly antimony, while the L lines take over for heavier elements whose K lines would require excitation energies beyond the range of routine instruments. Table.3 lists the lines normally used for a representative set of elements, and their positions already hint at which measurements will be straightforward and which will demand extra care.

Table.3 Characteristic X-ray Lines of Representative Elements.

The table also explains everyday laboratory habits. The lightest elements emit photons so weakly that air alone absorbs a large share of them, which is why sodium and magnesium are measured under vacuum or in a helium atmosphere and why the condition of the analysed surface can change their result. Heavier elements emit higher-energy photons that travel further through the sample and through air with little loss, so they tolerate simpler preparation and are generally easier to detect at low concentration. The same reasoning explains the overlapped pairs marked above: when two elements produce lines within a few tens of electron volts of each other, the analyst has to decide in advance which line to use for each.

Qualitative Screening of Unknown Samples

Qualitative analysis asks a simple question: what is in this sample? For a genuinely unknown material the answer is built by scanning the entire spectrum for peaks, matching each to a plausible element, and then cross-checking that the complete line set for every candidate is present. The approach is fast and needs no calibration, but its boundaries should be stated as clearly as its findings:

• Evidence strength: a set of peaks at the correct energies and relative intensities is strong evidence, whereas a single matching peak is a hypothesis that still needs confirmation.
• No concentrations: screening reports presence or absence only, and any figure quoted from it should be treated as an estimate rather than a measurement.
• Blind spots: elements whose lines fall outside the measured energy range, sit below the detection limit, or hide beneath a major peak will not appear, so a null result is not proof of absence.

Used with those limits in mind, screening is a rapid first pass that decides whether a material is what a supplier claims and whether a more rigorous measurement is warranted.

Quantitative Analysis with Calibration Strategies

Quantitative analysis converts peak intensities into concentrations, and the conversion depends entirely on calibration. The most direct strategy builds an empirical relationship by measuring reference materials of known composition that closely match the unknown samples, then fitting measured intensity to known concentration. Accuracy rises sharply when the standards match the samples in matrix, particle size, and preparation, because the physical and chemical effects are then common to both and largely cancel. A dedicated element analysis programme is built on exactly this discipline of matched standards and verified recovery. The strategies below cover most analytical situations, and the choice between them is driven by how closely a standard can be made to resemble the sample.

The arithmetic that turns intensity into a number is worth following step by step, because it explains both why XRF performs so well and where it goes wrong. Every element travels the same path from raw counts to a reported concentration, and each link in that path has a physical meaning:

• Net intensity: the detector registers a number of counts in the region of the spectrum where a line appears, and the counts from the background beneath that peak are subtracted from it. The remainder, divided by the acquisition time, is the net intensity in counts per second, and this is the only quantity that carries analytical information about the element.
• Why a ratio is used rather than a raw count rate: the absolute intensity of a line depends on tube voltage, tube current, filter selection, detector efficiency, and measurement geometry just as strongly as on concentration. Dividing the sample intensity by a reference intensity measured under the same conditions cancels those common factors, and the resulting ratio is far more stable than either signal on its own.
• The proportionality itself: within a fixed matrix, net intensity is directly proportional to concentration, so a plot of intensity against concentration is a straight line. The slope of that line is the sensitivity, and a steep slope means the element produces many counts per unit of concentration, which is what makes its detection limit low.
• Why the line bends: absorption and enhancement change the slope as the composition of the sample changes. A calibration established on one matrix therefore drifts when it is applied to another, and matrix coefficients — measured empirically or calculated from physics — are what restore the proportionality across a wider range of compositions.
• Normalisation and iteration: the correction coefficients depend on the concentrations, and the concentrations depend on the coefficients, so the calculation is solved repeatedly until the values stop changing. Most methods then normalise the elemental totals, typically to 100 percent for an oxide or metal sum, which absorbs small systematic errors and closes the mass balance of the analysis.

In short, the reported concentration is the net intensity scaled by the calibration slope and then adjusted by the matrix coefficients, with the whole calculation repeated until it converges. Every requirement XRF places on the analyst — matched standards, careful preparation, verified corrections — exists to keep that chain intact.

Table.4 XRF Calibration and Quantification Strategies Compared.

When suitable matched standards are unavailable, theoretical approaches fill the gap. Methods based on fundamental parameters calculate expected intensities from physical constants and the measured spectrum, allowing concentrations to be estimated even for materials with no dedicated standards. Because these calculations rest on physical models rather than on measurements alone, they are transparent and adaptable, and modern software implements them routinely, though their accuracy still benefits from verification against at least a few reference materials. Where the highest confidence is required, XRF results are frequently cross-checked against an independent elemental method such as ICP testing, so that any systematic bias in either approach becomes visible.

Matrix Correction and Spectral Overlap Resolution

Even a well-calibrated method must account for the ways in which a sample influences its own signal. Three effects explain most of the deviation between measured intensity and true concentration:

• Absorption: fluorescent photons are attenuated before they escape the sample, and the effect is strongest when heavy elements are present alongside the light elements being measured.
• Enhancement: a strong emitter can excite a neighbouring element and inflate its apparent concentration, working in the opposite direction to absorption.
• Spectral overlap: when two elements produce lines at nearly the same energy their peaks merge, and the software must apportion the combined area between them.

Correction models treat the first two effects as mathematical relationships between the elements present, using either empirical coefficients derived from measurements or theoretical coefficients derived from physics. Overlap is a separate but related problem whose severity depends on detector resolution and on the relative concentrations involved, and a trace element sitting beneath the tail of a major peak is the hardest case of all. Deconvolution algorithms and the selection of interference-free alternative lines both help, and the choice between them should be justified by the composition actually expected in the sample rather than applied by default.

Troubleshooting XRF Measurements

Every laboratory eventually meets a result that does not make sense. The value drifts, the check standard fails, an element appears that should not be there, or a duplicate disagrees with its twin. Troubleshooting is most efficient when it proceeds from the general to the specific: first confirm the instrument is healthy, then examine the sample, then interrogate the spectrum, and only then question the method itself. This order stops an analyst from chasing a sample problem through the software, or a software setting through the sample.

Verifying Instrument Health with Check Standards

The first question to answer is whether the instrument is behaving as it did when the method was established. Measure a stable check standard and compare the result with its expected value and with previous measurements. If the check passes, the instrument is producing trustworthy data and the problem lies with the sample or the method. If it fails, the fault is internal, and the next steps are to examine the measurement conditions, allow the instrument to reach thermal equilibrium if it has recently been moved, and confirm that the detector window is intact and free of contamination. A torn or dirty window degrades sensitivity, especially for light elements, and is a frequent and easily missed cause of poor performance.

Resolving Matrix and Particle-Size Issues

When results are biased rather than merely noisy, the sample itself is usually responsible. Particle-size effects are the classic culprit in powders: if the grains are too coarse or unevenly ground, the measured intensities depend on how the particles happen to be arranged, and lighter elements are affected most. Mineralogical effects compound this when a target element is concentrated in one particularly hard or soft mineral phase. Inhomogeneity then shows up as poor agreement between replicate measurements or between different spots on the same specimen. Where the relationship between composition and microstructure matters, SEM-EDS analysis can localise elements on the scale at which the heterogeneity occurs.

The remedies follow directly from the causes. Grind more finely and more uniformly, mix thoroughly, and press or fuse the material so that every particle presents the same geometry to the beam. Measure more than one spot and average the results, which converts an unreliable single reading into a defensible mean. Where surface condition is the issue, re-polish or clean the specimen. Where the sample is simply too thin to be measured reliably, add backing material or increase the sample mass. In every case the goal is the same: to make the measured surface a fair representative of the material as a whole.

Correcting Spectral Overlaps and Background Errors

When one specific element reads implausibly high or low, an overlap or a background problem is often to blame. Overlaps inflate the apparent concentration of the element whose line is overlapped, so the symptom is usually an anomalously high value for a minor element that shares its line energy with a major one. A first response is to change the analytical line, since selecting a different, interference-free line for the same element frequently resolves the issue outright. Where no clean line exists, deconvolution or an empirical correction derived from standards is the appropriate tool.

Background errors produce a different signature. If the background window is poorly placed, a trace result may drift with matrix changes, because the subtracted baseline no longer reflects the true continuum. Re-examining the background positions on spectra from standards and samples, and confirming that the model behaves consistently across the energy range, restores reliability. It is worth remembering that a correction is only as good as the assumption behind it, and that an uncorrected overlap is often preferable to a correction applied blindly.

Managing Drift, Contamination, and Counting Statistics

Three slower-acting problems degrade data over time rather than all at once. Drift is the gradual change in instrument response caused by component ageing and environmental variation, and it is managed by measuring monitor samples at regular intervals and applying a drift correction when the response departs from its reference value. Contamination accumulates from the samples themselves, as dust, flakes, and residues transfer to the analysis window or chamber, and it is prevented by cleaning between samples, handling specimens carefully, and replacing the window when it becomes soiled. Counting statistics set the floor on precision: a peak measured over few counts is inherently uncertain, so longer acquisition times or higher count rates improve the signal-to-noise ratio and tighten the confidence interval around the result. Where contamination from heavy elements is a recurring concern, a dedicated heavy metal analysis programme can quantify and trace the source.

Table.6 Common XRF Problems, Likely Causes, and Corrective Actions.

BOC Sciences XRF Spectroscopy Solutions

BOC Sciences provides X-ray fluorescence testing and elemental characterisation as part of a broad analytical platform. Our laboratory supports clients across materials science, geology, metallurgy, polymers, and pharmaceutical development, and treats each sample matrix as a distinct analytical problem rather than forcing it onto a generic method. Whether the need is a routine elemental screen, a quantitative assay of major and minor components, or a difficult measurement in a matrix that resists standard preparation, projects are designed around the question being asked, and the reporting level is agreed before any data are generated.

XRF Testing Services

Our core XRF service covers qualitative, semi-quantitative, and quantitative analysis of solids, powders, liquids, and thin films. Samples are prepared using the route best suited to the material — pressed pellets or fused beads for powders, polished surfaces for metals, and film-supported cups for liquids — so that the reported values reflect the true composition of the material rather than the condition of its surface. Standard element panels are available with rapid turnaround, while custom panels extend coverage to the specific elements of interest. Each report is accompanied by the calibration and quality-check information on which the results rest, so that the numbers can be used with confidence. Broader programmes in impurity profiling combine XRF results with complementary techniques when a fuller elemental picture is required.

Method Development for Complex Matrices

Some materials defeat routine preparation. Highly heterogeneous powders, very small sample masses, organic matrices, and specimens that cannot be heated or pressed all call for method development rather than an off-the-shelf approach. Our scientists begin from the analytical question — which elements, at what levels, in what matrix — and then work systematically through preparation options, calibration design, and correction strategy until the method delivers results that stand up to scrutiny. Where a sample is genuinely unsuitable for XRF, we say so and recommend a complementary technique instead of forcing a weak measurement. This commitment to fit-for-purpose work is the foundation of our challenging sample analytical method development service, and it is reinforced by a wider capability in analytical testing and release for programmes that require consolidated data packages.

Integrated Elemental and Material Characterization

Elemental composition is often only one part of the picture, and our laboratory is structured to combine XRF with the rest of the analytical toolkit. XRF results can be paired with diffraction to identify mineral phases, with electron microscopy and energy-dispersive analysis to relate composition to microstructure, and with wet-chemical elemental methods to cross-check the numbers. For development programmes, this integration means a single point of coordination and a consolidated data package rather than a scatter of disconnected reports, which in turn shortens the path from measurement to decision.

Table.7 XRF Related Services at BOC Sciences.

Expert Services Supporting Material Analysis

Expert Services Supporting Spectroscopy Testing