X-Ray Spectrometric Methods

A forensic chemist points a handheld gun at a smear of paint from a hit-and-run scene. In ninety seconds, they read off relative peaks of lead, chromium, and titanium. This happens without dissolving a single flake. That instrument is doing X-ray fluorescence spectrometry. It is quietly one of the most consequential analytical tools of the twentieth and twenty-first centuries. It is used to certify the alloy in a jet turbine blade, screen toys for banned cadmium, and solve the structure of hemoglobin. X-ray spectrometric methods exploit a key fact. Core electrons in atoms have binding energies in the kiloelectronvolt range. These energies are element-specific and largely insensitive to chemical environment. They are also directly addressable with X-ray photons. This gives elemental analysis a rare combination. It is nondestructive, and applicable to solids, liquids, and powders without dissolution. It is simultaneously multi-elemental across almost the entire periodic table. The same physics also gives crystallography its central tool. This is coherent scattering of X-rays by ordered planes of atoms, known as X-ray diffraction. It lets chemists read the three-dimensional arrangement of atoms in a crystal. They do this from the angles and intensities of scattered beams. The big picture is that X-ray spectrometry sits at an intersection. Three related but distinct measurements are built on one shared foundation. That foundation is the interaction of X-ray photons (roughly 100 eV to 100 keV, wavelengths of 0.01–10 nm) with the electrons of matter. Emission methods (XRF) measure characteristic photons emitted when a core hole is filled. Absorption methods (XAS, XANES, EXAFS) measure how strongly a sample attenuates a beam. This is done as photon energy is scanned through an absorption edge. This reveals oxidation state and local coordination. Diffraction methods (XRD) measure the angles at which X-rays constructively interfere off periodic lattices. This reveals unit cell dimensions, phase identity, and crystallite size. Each method rests on the same quantum picture. This picture involves core-electron transitions and photon-matter interaction, established for Absorption and Emission Processes. It is now pushed to the energy regime where core, not valence, electrons participate. This article builds the physical picture of characteristic X-ray production and Bragg diffraction. It derives Moseley's law and the Bragg equation from first principles. Full quantitative worked examples are included. It works through problems ranging widely. These include identifying an unknown metal by its Kα energy. They also include computing crystallite size from diffraction peak broadening. It closes with the instrumentation and the practicing scientists who built the field. Finally, it covers the extraordinary reach of X-ray methods today. This spans from pharmaceutical polymorph screening to synchrotron protein crystallography.

Key takeaways

X-rays (0.01–10 nm, 0.1–100 keV) probe core electrons; their energies are element-specific and nearly independent of chemical bonding.
Moseley's law, $\sqrt{\nu} = k(Z - \sigma)$ , lets you identify an element from its characteristic emission line energy — the historical proof that atomic number, not mass, orders the periodic table.
XRF (X-ray fluorescence) measures emitted characteristic photons for nondestructive, multi-element quantitative analysis of solids and liquids.
XAS/XANES/EXAFS scan photon energy across an absorption edge to extract oxidation state and local bond distances, even in amorphous or dilute samples.
XRD uses Bragg's law, $n\lambda = 2d\sin\theta$ , to determine crystal structure, phase identity, and crystallite size from diffraction angles and peak widths.
Matrix effects (absorption, enhancement) and instrument geometry govern quantitative accuracy in all three techniques, demanding calibration standards or fundamental-parameters correction.
Key idea: core-electron energy levels are atomic fingerprints, and X-ray photons are simultaneously the probe that ejects them, the messenger that reports them, and the wave that diffracts off the lattice they belong to.

Where this lives today

Portable and handheld XRF analyzers, no larger than a power drill, are now standard tools in mining (ore grade determination in the field), recycling (sorting scrap metal alloys by composition in seconds), regulatory compliance (screening consumer products for restricted heavy metals like lead and cadmium under RoHS regulations), and art conservation (nondestructively mapping pigments in paintings to detect forgeries or study an artist's technique, since lead white, vermilion, and modern titanium white pigments have distinct elemental signatures). Wavelength-dispersive XRF, with its superior resolution, remains the gold standard in cement and steel production for rapid, precise quality control of major and trace elements batch by batch. X-ray absorption spectroscopy (XANES and EXAFS), almost always performed at synchrotron light sources that provide the tunable, intense, monochromatic X-ray beams required, is indispensable wherever a sample lacks long-range crystalline order — dilute metal centers in metalloenzymes, catalytic nanoparticles under real reaction conditions (operando spectroscopy), amorphous battery electrode materials, and environmental samples like soil-bound arsenic or chromium, where oxidation state controls toxicity and mobility. The chemical shift in the absorption edge position, a few eV, distinguishes Fe(II) from Fe(III) or Cr(III) from the far more toxic and mobile Cr(VI) even in complex, non-crystalline environmental matrices. X-ray diffraction underpins pharmaceutical manufacturing, where polymorph screening (distinguishing different crystal forms of the same drug molecule, which can have dramatically different solubility and bioavailability) is a regulatory requirement for every new drug product; materials science, where phase identification and crystallite/grain size measurement guide the design of battery cathodes, catalysts, and semiconductor thin films; and structural biology, where single-crystal X-ray diffraction at synchrotron beamlines remains the dominant method for determining atomic-resolution structures of proteins, viruses, and drug-target complexes — the technique that revealed the double helix, ATP synthase, and the ribosome. The same Bragg geometry that the two Braggs worked out for rock salt in 1913 now resolves structures with tens of thousands of atoms per unit cell.

Ask the tutor: How does synchrotron radiation make EXAFS and XANES measurements possible in ways a laboratory X-ray tube cannot?

The physical picture — thorough prose

An X-ray photon carries enough energy to eject an electron from an inner shell (K, L, M) of a heavy atom, something a visible or ultraviolet photon cannot do because valence electrons, not core electrons, absorb those lower energies. When a K-shell electron (principal quantum number $n=1$ ) is ejected, the atom is left in a highly excited, unstable ionic state with a vacancy deep in its electronic structure. An electron from a higher shell (L, $n=2$ , or M, $n=3$ ) falls into that vacancy within about $10^{-15}$ – $10^{-14}$ s, and the energy difference is released either as a photon (X-ray fluorescence) or transferred nonradiatively to another electron, which is ejected (the Auger effect). The emitted photon has an energy fixed by the difference between two atomic core-orbital binding energies — a difference that depends overwhelmingly on the nuclear charge $Z$ screened by the other electrons, and only weakly on the chemical environment of the atom. This is why X-ray emission lines are called characteristic lines: Kα for an L→K transition, Kβ for an M→K transition, Lα for an M→L transition, each labeled by shell of origin. Because the core levels are so tightly bound and so well shielded from the valence electrons that do the bonding, the energy of, say, Fe Kα (6.40 keV) is nearly the same whether the iron is metallic, in Fe₂O₃, or in a hemoglobin molecule — a huge practical advantage for quantitative elemental analysis, since one does not need to know the compound to identify and quantify the element. Small shifts (a few eV) do occur with oxidation state and coordination, and these chemical shifts are exactly what X-ray absorption near-edge structure (XANES) exploits to determine oxidation state nondestructively. The complementary process is absorption: as a beam of X-rays of intensity $I_0$ passes through a sample, its intensity decays exponentially,

I = I_0 e^{-\mu x}

where $\mu$ (units cm⁻¹, or $\mu/\rho$ in cm² g⁻¹ as the mass absorption coefficient) rises smoothly with $Z^3$ to $Z^4$ and falls smoothly with photon energy — except for sudden, dramatic jumps called absorption edges at exactly the binding energy of a core shell. Just above the K-edge energy, a photon has just enough energy to eject a K electron, and $\mu$ jumps by a factor of five to ten. Scanning photon energy through an edge and recording $\mu(E)$ finely resolved reveals oscillations in absorption above the edge (EXAFS, extended X-ray absorption fine structure) that arise from interference between the outgoing photoelectron wave and waves backscattered from neighboring atoms — encoding bond distances to within 0.01 Å even in glasses, solutions, and catalytic surfaces with no long-range order at all. The third pillar, diffraction, arises because a crystal is a three-dimensional periodic array of atoms with spacings (a few ångströms) comparable to X-ray wavelengths. When X-rays strike a set of parallel lattice planes, each plane reflects a small fraction of the beam; these reflected wavelets interfere constructively only at specific angles determined by the plane spacing $d$ and wavelength $\lambda$ , given by Bragg's law. Because different crystalline phases have different unit cells and therefore different sets of $d$ -spacings, a diffraction pattern (intensity vs. $2\theta$ ) is as distinctive as a fingerprint — quartz, calcite, and NaCl each give an unmistakable, reproducible pattern regardless of how the sample was prepared, as long as it is crystalline.

Figure

Characteristic X-ray emission: core-hole creation and Kα/Kβ/Lα transitions

Figure

Bragg diffraction geometry: path-length difference between reflected rays

The core ideas — governing equations

Moseley's law relates the frequency of a characteristic X-ray line to atomic number:

\sqrt{\nu} = k(Z - \sigma)

where:

$\nu$ is the frequency of the characteristic line (Hz), related to photon energy by $E = h\nu$
$Z$ is the atomic number of the emitting element
$\sigma$ is a screening constant (≈ 1 for Kα lines, since one other K electron partially shields the nuclear charge)
$k$ is a proportionality constant specific to the line series (Kα, Kβ, Lα, …), incorporating the Rydberg constant and shell quantum numbers

Reading: a plot of $\sqrt{\nu}$ (or $\sqrt{E}$ ) against $Z$ is a straight line — this was the direct experimental proof that atomic number, not atomic mass, is the correct ordering variable of the periodic table, and it is the working principle behind identifying unknown elements from XRF spectra today.

Bragg's law governs diffraction of X-rays by a crystal lattice:

n\lambda = 2d\sin\theta

where:

$n$ is the order of diffraction (a positive integer, 1, 2, 3, …)
$\lambda$ is the X-ray wavelength (Å), fixed by the source (e.g., Cu Kα = 1.5406 Å)
$d$ is the interplanar spacing of the reflecting lattice planes (Å)
$\theta$ is the angle between the incident beam and the lattice plane (the Bragg angle; the detector sits at $2\theta$ from the incident beam)

Reading: for a given $d$ -spacing, only specific angles $\theta$ give constructive interference; scanning $\theta$ (equivalently $2\theta$ ) and recording scattered intensity yields a fingerprint pattern of peaks, each corresponding to a distinct family of lattice planes.

Beer–Lambert attenuation for X-ray absorption, directly analogous to optical absorbance:

\ln\left(\frac{I_0}{I}\right) = \mu x = \left(\frac{\mu}{\rho}\right)\rho x

where:

$I_0, I$ are incident and transmitted beam intensities
$\mu$ is the linear attenuation coefficient (cm⁻¹)
$\mu/\rho$ is the mass attenuation coefficient (cm² g⁻¹), tabulated per element and roughly independent of chemical state
$\rho$ is density (g cm⁻³) and $x$ is path length (cm)

Reading: mass attenuation coefficients for a compound are additive by mass fraction, $(\mu/\rho)_{\text{compound}} = \sum_i w_i (\mu/\rho)_i$ , which underlies both quantitative XRF matrix corrections and X-ray absorption edge spectroscopy.

Scherrer equation links diffraction peak broadening to crystallite size:

D = \frac{K\lambda}{\beta \cos\theta}

where:

$D$ is the mean crystallite dimension (nm), valid roughly for $D < 100$ –200 nm
$K$ is the Scherrer shape constant (≈ 0.9 for spherical crystallites)
$\beta$ is the peak full width at half maximum (FWHM, in radians), corrected for instrumental broadening
$\theta$ is the Bragg angle of the peak used

Reading: smaller crystallites give broader diffraction peaks because fewer lattice planes participate in interference, reducing the sharpness of the constructive-interference condition — this is a direct application of wave interference with a finite number of scatterers, the same physics that broadens diffraction-grating lines with fewer illuminated grooves.

Derivation AMoseley's law from the Bohr/hydrogenic model of core transitions

Goal: derive $\sqrt{\nu} \propto (Z-\sigma)$ starting from the hydrogenic energy-level formula applied to a K-shell electron with nuclear charge screened by one remaining K electron.

Step 1. Treat the atom as hydrogen-like for the innermost electrons: an electron in shell $n$ experiences an effective nuclear charge $Z_{\text{eff}} = Z - \sigma$ , where $\sigma$ accounts for screening by other electrons in the same or inner shells. The Bohr energy of a hydrogenic level is

E_n = -\frac{Z_{\text{eff}}^2 \, R_H hc}{n^2}

where $R_H$ is the Rydberg constant ( $1.097 \times 10^7$ m⁻¹) and $hc$ converts it to an energy.

Step 2. For a Kα transition, an electron falls from $n=2$ (L shell) to $n=1$ (K shell), where a single remaining K electron screens one unit of nuclear charge, giving $\sigma \approx 1$ for the initial K-vacancy state seen by the incoming L electron. The emitted photon energy is the difference,

E_{K\alpha} = Z_{\text{eff}}^2 R_H hc \left(\frac{1}{1^2} - \frac{1}{2^2}\right) = \frac{3}{4} R_H hc (Z-1)^2

Step 3. Since $E = h\nu$ , divide through by $h$ :

\nu = \frac{3}{4} R_H c (Z-1)^2

Taking the square root of both sides isolates the linear dependence on $Z$ :

\sqrt{\nu} = \sqrt{\frac{3}{4}R_H c}\;(Z - 1)

Step 4. This has exactly the form of Moseley's empirical law, $\sqrt{\nu} = k(Z-\sigma)$ , with $k = \sqrt{\tfrac{3}{4}R_H c}$ and $\sigma = 1$ for the Kα series. Plugging in constants, $k = \sqrt{0.75 \times 1.097\times10^7 \text{ m}^{-1} \times 2.998\times10^8 \text{ m s}^{-1}} = 4.97\times10^7\ \text{Hz}^{1/2}$ , matching the experimentally fitted Moseley slope to within a few percent (the small residual discrepancy comes from higher-order screening and relativistic corrections neglected in the simple Bohr treatment).

Limiting check: at $Z=1$ (hydrogen itself, with no L-shell electron to eject in a "Kα" sense) the formula degenerates trivially; the physically meaningful range starts around $Z \gtrsim 10$ , where the hydrogenic approximation with a single screening constant becomes reasonably accurate — consistent with Moseley's original data set, which began at aluminum ( $Z=13$ ).

Go deeper · Derivation B · 1 minBragg's law from path-length interference of scattered waves

Derivation BBragg's law from path-length interference of scattered waves

Goal: derive $n\lambda = 2d\sin\theta$ from the requirement that waves reflected from successive lattice planes interfere constructively.

Step 1. Consider two parallel lattice planes separated by spacing $d$ , with an incident X-ray beam of wavelength $\lambda$ striking both planes at glancing angle $\theta$ (measured from the plane, not the normal — the crystallographic convention). Part of the beam reflects off the first plane; part penetrates and reflects off the second plane one spacing $d$ deeper.

Step 2. The ray reflecting off the second, deeper plane travels an extra distance compared to the ray reflecting off the first plane. Drop a perpendicular from the first-plane reflection point to the incident and reflected rays of the second plane; geometry (right triangles with hypotenuse $d$ and angle $\theta$ ) shows the extra path on each side is $d\sin\theta$ , for a total extra path length

\Delta = 2d\sin\theta

Step 3. Constructive interference — a diffracted beam of significant intensity — occurs only when this path difference equals a whole number of wavelengths, since only then do the two reflected wavefronts arrive in phase and reinforce rather than cancel:

\Delta = n\lambda, \quad n = 1, 2, 3, \ldots

Step 4. Equating the two expressions for $\Delta$ gives Bragg's law directly:

n\lambda = 2d\sin\theta

For any lattice plane spacing $d$ and wavelength $\lambda$ , only discrete angles $\theta_n = \arcsin(n\lambda/2d)$ satisfy this condition; between them, waves from the many planes deeper in the crystal interfere destructively (this constructive reinforcement across thousands of planes is also why diffraction peaks are so sharp — a point returned to in the Scherrer analysis).

Limiting check: as $d \to \lambda/2$ (the smallest spacing for which first-order diffraction is geometrically possible), $\sin\theta \to 1$ and $\theta \to 90°$ , meaning the beam must reflect almost straight back — consistent with the physical requirement that diffraction cannot occur for lattice spacings smaller than half the wavelength, exactly analogous to the resolution limit in optical diffraction gratings.

Worked examples

Example 1 (Foundational): identifying an unknown metal from its Kα energy. An XRF spectrum of an unknown metal foil shows a strong emission line at $E_{K\alpha} = 5.415$ keV. Identify the element using Moseley's law with the Kα screening constant $\sigma = 1$ and $k = 4.97\times10^7$ Hz $^{1/2}$ (from the derivation above, expressed here via energy). Convert the derivation's frequency form into an energy form by multiplying by $h$ :

\sqrt{E_{K\alpha}} = h k (Z-1) \equiv k_E (Z-1)

Calibrate $k_E$ using a known reference line — take Fe ( $Z=26$ ), $E_{K\alpha}=6.404$ keV, as the calibration point:

k_E = \frac{\sqrt{6.404\ \text{keV}}}{26-1} = \frac{2.5306}{25} = 0.10122\ \text{keV}^{1/2}

Now solve for the unknown's atomic number:

Z - 1 = \frac{\sqrt{5.415\ \text{keV}}}{0.10122\ \text{keV}^{1/2}} = \frac{2.3271}{0.10122} = 22.99

Z = 23.99 \approx 24

$Z = 24$ is chromium. Checking against the tabulated value, Cr Kα₁ is 5.415 keV exactly — a match confirming the identification. This is precisely the logic of energy-dispersive XRF elemental identification: measure line energy, compare (via a calibration curve built from standards, not just one reference point in practice) to the known Kα energies tabulated for all elements, Z = 11 (Na) through Z = 92 (U) and beyond.

Example 2 (Applied): quantitative XRF of a brass alloy and matrix absorption correction. A brass sample (Cu–Zn alloy) is analyzed by XRF. The Cu Kα (8.048 keV) net peak intensity from the unknown is 4820 counts/s. A pure copper standard under identical conditions gives 15,200 counts/s. A naïve estimate would say

w_{\text{Cu, naive}} = \frac{4820}{15200} = 0.317 \ (31.7\%)

But this ignores matrix absorption: zinc in the sample absorbs some of the Cu Kα photons (both incident-beam and emitted-fluorescence absorption) differently than pure copper does, so a direct intensity ratio is not simply proportional to concentration. Using a simplified fundamental-parameters absorption correction,

w_{\text{Cu}} = w_{\text{Cu,naive}} \times \frac{(\mu/\rho)_{\text{Cu standard, eff}}}{(\mu/\rho)_{\text{brass, eff}}}

Suppose the effective mass absorption coefficient (at the Cu Kα-exciting and Cu Kα-emitted wavelengths combined) is $(\mu/\rho)_{\text{Cu}} = 52.5$ cm² g⁻¹ for pure copper and $(\mu/\rho)_{\text{brass}} = 61.8$ cm² g⁻¹ for the brass matrix (zinc absorbs more strongly in this energy window). Then

w_{\text{Cu}} = 0.317 \times \frac{52.5}{61.8} = 0.317 \times 0.8495 = 0.269

So the corrected copper mass fraction is 26.9%, not the naive 31.7% — a 5-percentage-point matrix correction, illustrating why quantitative XRF requires either matrix-matched calibration standards or a fundamental-parameters/alpha-coefficient correction model, never a bare intensity ratio, whenever the matrix composition of standard and unknown differ substantially. (Real brass is typically 60–70% Cu, 30–40% Zn; this worked value is for illustration of the correction method, using a single simplified absorption term rather than a full multi-line fundamental-parameters calculation.)

Example 3 (Exam/competition level): crystallite size and phase identification from a powder XRD pattern. A powder XRD pattern of a catalyst support is recorded with Cu Kα radiation ( $\lambda = 1.5406$ Å). A peak appears at $2\theta = 36.68°$ with FWHM $\beta = 0.42°$ (after subtracting instrumental broadening of $0.08°$ in quadrature). Determine (a) the $d$ -spacing of the reflecting planes, and (b) the mean crystallite size, then (c) identify whether the phase is more consistent with anatase or rutile TiO₂, given anatase's strongest reflection near $2\theta = 25.3°$ ( $d=3.52$ Å) and rutile's strongest near $2\theta=27.4°$ ( $d=3.25$ Å), plus a known anatase (101)/(004) secondary near $2\theta=37.8°$ .

(a) d-spacing. $\theta = 36.68°/2 = 18.34°$ . Using Bragg's law with $n=1$ :

d = \frac{n\lambda}{2\sin\theta} = \frac{1 \times 1.5406\ \text{Å}}{2\sin(18.34°)}

$\sin(18.34°) = 0.3147$ , so

d = \frac{1.5406}{2(0.3147)} = \frac{1.5406}{0.6294} = 2.448\ \text{Å}

(b) Crystallite size via Scherrer. First correct the observed FWHM for instrumental broadening (Gaussian deconvolution, combined in quadrature):

\beta_{\text{sample}} = \sqrt{\beta_{\text{obs}}^2 - \beta_{\text{inst}}^2} = \sqrt{0.42^2 - 0.08^2} = \sqrt{0.1764 - 0.0064} = \sqrt{0.1700} = 0.4123°

Convert to radians: $\beta = 0.4123° \times (\pi/180) = 7.196\times10^{-3}$ rad. Using $K=0.9$ :

D = \frac{K\lambda}{\beta\cos\theta} = \frac{0.9 \times 1.5406\ \text{Å}}{(7.196\times10^{-3})\cos(18.34°)}

$\cos(18.34°) = 0.9492$ , so the denominator is $7.196\times10^{-3} \times 0.9492 = 6.831\times10^{-3}$ :

D = \frac{1.3865\ \text{Å}}{6.831\times10^{-3}} = 202.9\ \text{Å} = 20.3\ \text{nm}

So the crystallite size is approximately 20 nm — a nanocrystalline material, consistent with a high-surface-area catalyst support.

(c) Phase identification. The peak at $2\theta = 36.68°$ ( $d=2.448$ Å) does not directly match either headline reflection given (25.3° or 27.4°), but anatase TiO₂ has a well-known secondary reflection assigned to the (004) plane near $2\theta \approx 37.8°$ and the (200) near $2\theta\approx 48.0°$ ; more diagnostically, anatase's (105)/(211) cluster appears near 2θ = 53–55°, while a reflection at 2θ close to 36.6–36.9° with d ≈ 2.43–2.49 Å corresponds to anatase's (103)/(004)/(112) unresolved cluster region, whereas rutile's analogous mid-angle reflection, (101), sits at 2θ ≈ 36.1° (d = 2.487 Å) — close, but rutile's dominant (110) at 27.4° would almost always appear as the strongest line if rutile were present in significant quantity. Given that neither the 25.3° (anatase 101) nor 27.4° (rutile 110) principal lines were reported as present in this problem's pattern description, a rigorous conclusion requires the full pattern; the exercise's pedagogical point is the calculation method (Bragg $d$ -spacing and Scherrer size), with phase assignment in practice always resting on matching the entire measured pattern against a reference database (the ICDD Powder Diffraction File), never a single peak.

Limiting cases, scope & common traps

Situation	What happens	Why
Light elements (Z < 11, Na)	XRF sensitivity drops sharply; Kα energies < 1 keV are strongly self-absorbed	Fluorescence yield $\omega_K \propto Z^4/(Z^4+a)$ falls for low Z; Auger decay dominates instead of photon emission
High X-ray energy far above an edge	$\mu/\rho$ decreases smoothly (~ $E^{-3}$ ) except at edges	Photoelectric cross-section falls off with energy once well above threshold
Amorphous or glassy samples	XRD gives only broad halos, no sharp peaks; EXAFS/XANES still work fine	Diffraction requires long-range periodic order; absorption spectroscopy needs only local coordination
Crystallite size > ~200 nm	Scherrer equation underestimates broadening contribution; peaks approach instrumental width	Peak width becomes dominated by instrumental resolution, not size broadening
Trace element in a heavy matrix	XRF detection limit degrades due to strong matrix absorption	Matrix absorbs both exciting and characteristic photons before they escape to the detector
Sample is a mixture of polymorphs	XRD peaks broaden/split even for "pure" one compound geometry	Different unit cells (even of the same chemical formula) or preferred orientation and strain change $d$ -spacings and intensities

Common trapMisconception: "a diffraction peak position tells you the crystallite is 'strained' or 'small' the same way peak width does"

Peak position (via Bragg's law) reports $d$ -spacing — the average lattice constant. Peak width (via Scherrer/Williamson-Hall) reports crystallite size and/or microstrain. These are independent pieces of information from the same peak: a shifted peak means a changed lattice parameter (e.g., alloying, thermal expansion, or applied stress), while a broadened peak means smaller coherently-diffracting domains or nonuniform strain. Conflating the two — for instance, attributing peak broadening to a compositional shift, or a peak shift to crystallite size — is one of the most common errors in interpreting powder patterns, and rigorous work always separates size and strain contributions using a Williamson-Hall plot ( $\beta\cos\theta$ vs. $\sin\theta$ ) before quoting a crystallite size.

Common trapMisconception: "XRF intensity is directly proportional to concentration, so a simple ratio to a standard gives the composition"

This holds only when sample and standard are matrix-matched (same overall absorption and enhancement behavior), because the measured intensity depends not just on how much of the analyte element is present but on how strongly the rest of the matrix absorbs both the exciting beam and the emitted fluorescence (as shown quantitatively in Worked Example 2). Ignoring matrix absorption/enhancement effects is the single most common source of large systematic error in quantitative XRF, and is precisely why standard methods (fundamental parameters, Lucas-Tooth/de Vries influence coefficients, or matrix-matched calibration curves) are mandatory for accurate work, not merely good practice.

The chemist behind it

Henry Moseley was a young British physicist working under Ernest Rutherford at Manchester and then Oxford in 1913, barely two years after Rutherford's nuclear atom was proposed and one year after Bohr's model of hydrogen. Moseley built an X-ray spectrometer using a crystal (following the newly established Bragg diffraction method) to measure the characteristic Kα and Lα X-ray wavelengths emitted by roughly forty different elements, from aluminum to gold, mounted in turn as the anode target of a specially built X-ray tube. He found, with striking precision, that the square root of the frequency of each element's Kα line increased linearly with an integer he called atomic number — not atomic weight. This immediately resolved several standing anomalies in Mendeleev's periodic table (for instance, why tellurium, heavier than iodine, nonetheless belongs before it), and gave the periodic table its correct, physically grounded ordering variable for the first time. Moseley's work was cut short: he enlisted in the British Army in the First World War and was killed at Gallipoli in 1915, at age 27, a loss so widely regretted within the physics community that it directly influenced British policy to keep scientists out of front-line combat roles for the remainder of the war. The complementary discovery — that X-rays diffract from crystals at all, and the geometric law governing where — belongs to William Henry Bragg and his son William Lawrence Bragg, working together at Cambridge and Leeds starting in 1912–1913, building directly on Max von Laue's 1912 demonstration that crystals could diffract X-rays. The younger Bragg, only 22 at the time, derived the simple geometric condition for constructive interference that now bears the family name, and the pair went on to solve the first crystal structures ever determined by X-ray diffraction (NaCl, diamond, and others), founding the discipline of X-ray crystallography that would, decades later, reveal the structures of DNA, hemoglobin, and myriad drug targets.

Origins · 1 minHenry Gwyn Jeffreys Moseley and how this was discovered

Henry Gwyn Jeffreys Moseley

1887–1915 · Manchester and Oxford, working in Rutherford's laboratory during the founding years of nuclear and quantum atomic theory

Moseley's systematic measurement of characteristic X-ray spectra across the periodic table gave atomic number a direct experimental, physical meaning — not merely a place in a list, but the actual positive charge on the nucleus, measurable by an X-ray line. His law remains the operating principle of every energy-dispersive and wavelength-dispersive XRF instrument used today.

History

- 1895 — Wilhelm Röntgen discovers X-rays, observing their ability to penetrate matter and expose photographic plates.

- 1912 — Max von Laue demonstrates X-ray diffraction by crystals, confirming both the wave nature of X-rays and the periodic structure of crystals.

- 1913 — William Henry and William Lawrence Bragg formulate Bragg's law and solve the first crystal structures (NaCl, diamond) by X-ray diffraction.

- 1913 — Henry Moseley measures characteristic X-ray spectra across the periodic table and establishes Moseley's law, fixing atomic number as the correct periodic ordering variable.

- 1948 onward — Development of energy-dispersive and wavelength-dispersive XRF spectrometers transforms elemental analysis in metallurgy, geology, and industry; later, synchrotron sources extend absorption spectroscopy (EXAFS/XANES) and protein crystallography to atomic-resolution structural biology.

Why don't chemical bonds affect X-ray emission energies the way they affect UV-visible absorption?

Characteristic X-ray transitions involve core electrons (K, L, M shells) that lie very close to the nucleus and are strongly bound, typically hundreds to tens of thousands of eV below vacuum, far below the valence molecular orbitals (a few eV) that participate in bonding and in UV-visible transitions. Because core orbitals barely overlap with neighboring atoms' orbitals, their energies are set almost entirely by the nuclear charge and inner-shell screening, changing by at most a few eV even when an atom's oxidation state or coordination changes drastically. This near-invariance is what makes XRF a robust elemental (not molecular) probe: you can identify and quantify "iron" regardless of whether it is metallic, Fe²⁺, or Fe³⁺, without needing to know the compound in advance. The small residual shifts that do occur (chemical shifts of order 1–10 eV) are exactly what XANES exploits for oxidation-state determination, requiring a much higher-resolution, energy-scanning measurement than routine XRF.

What is the difference between energy-dispersive XRF (EDXRF) and wavelength-dispersive XRF (WDXRF)?

EDXRF uses a solid-state detector (commonly silicon drift detector) that directly measures the energy of each incoming X-ray photon via the size of the charge pulse it produces, recording an entire spectrum simultaneously; it is compact, fast, and forms the basis of portable and handheld analyzers, but has modest energy resolution (~130–150 eV FWHM), which can cause line overlaps between neighboring elements. WDXRF instead diffracts the emitted X-rays off an analyzing crystal (using Bragg's law) and scans angle mechanically, achieving far superior energy resolution (a few eV) at the cost of being slower, larger, and mechanically more complex, since only one wavelength is measured at a time for a given crystal-detector angle. WDXRF remains the standard for high-precision, high-throughput elemental analysis in cement, steel, and mining laboratories, while EDXRF dominates portable and multi-element survey applications.

Can XRF or XRD be used on liquids, or only solids and powders?

XRF works well on liquids: they are typically held in a thin-film cell with an X-ray-transparent window (often a thin polymer film like Mylar or polypropylene), and quantitative analysis proceeds with liquid calibration standards, widely used for petroleum products, environmental water samples, and process-control solutions in industry. XRD, by contrast, fundamentally requires long-range periodic order to produce sharp diffraction peaks, so ordinary liquids give only a broad, featureless scattering halo with no useful crystallographic information; liquid crystals or suspensions of crystalline particles can still show diffraction from the ordered component. For structural information on molecules in solution, X-ray absorption spectroscopy (EXAFS) or small-angle X-ray scattering (SAXS, which probes larger-scale structure rather than atomic-resolution lattice spacing) are the appropriate techniques instead.

Why is the Scherrer equation said to break down above roughly 100–200 nm crystallite size?

As crystallites grow larger, the extra peak broadening they contribute becomes smaller and smaller (broadening scales as $1/D$ ), until it becomes comparable to or smaller than the instrument's own inherent peak width, set by things like beam divergence, detector resolution, and sample geometry. Once size broadening is swamped by instrumental broadening, the measurement can no longer distinguish "large crystallites" from "very large crystallites" — the peak width simply stops changing appreciably. Practically, labs report Scherrer sizes as reliable only up to a few hundred nanometers, and for genuinely large or well-ordered crystallites, other methods (electron microscopy, or line-profile analysis with synchrotron-resolution instruments) are needed to assess grain size.

How does an X-ray tube actually generate the X-rays used in these experiments?

A conventional sealed X-ray tube accelerates electrons from a heated filament (cathode) across a high voltage (commonly 20–60 kV) onto a metal target (anode, e.g., copper, molybdenum, or rhodium). Most of the electron kinetic energy converts to heat, but a small fraction produces X-rays in two ways: continuous bremsstrahlung radiation from the deceleration of electrons in the target's Coulomb field, spanning a broad energy range up to the full accelerating voltage, and discrete characteristic lines (like Cu Kα at 8.048 keV) when accelerated electrons themselves eject core electrons from target atoms, exactly analogous to the photon-induced core-hole process described for fluorescence, just driven by electron rather than photon impact. Synchrotron sources instead generate X-rays from relativistic electrons bent by magnetic fields, giving vastly higher intensity and full wavelength tunability, essential for EXAFS/XANES experiments that must scan smoothly across an absorption edge.

What does it mean physically that mass absorption coefficients are "additive"?

For a compound or mixture, the total probability that an X-ray photon is absorbed per unit mass is simply the mass-weighted sum of each constituent element's own mass absorption coefficient at that photon energy, because absorption events (photoelectric ejection of a core electron from a specific atom) are essentially independent, uncorrelated interactions between the photon and individual atoms, unaffected by which molecule the atom happens to be part of. This additivity, $(\mu/\rho)_{\text{compound}} = \sum_i w_i(\mu/\rho)_i$ , is what allows matrix absorption corrections in quantitative XRF to be calculated from tabulated elemental data alone, without needing separate absorption measurements for every possible compound, and is the same principle that lets radiologists estimate tissue attenuation from elemental composition (H, C, N, O, Ca) in medical imaging.

Why do heavier elements generally give stronger, easier-to-detect XRF signals than lighter ones?

Two effects compound in the same direction. First, the fluorescence yield $\omega$ — the probability that a core hole decays by photon emission rather than the competing nonradiative Auger process — rises steeply with atomic number, from a few percent for light elements like sodium up to over 90% for heavy elements like tin or lead, because Auger emission (ejecting another electron instead) becomes energetically and quantum-mechanically less favorable as the core-hole binding energy grows. Second, characteristic X-ray energies for heavy elements are higher, so they suffer less self-absorption while escaping the sample and less attenuation in air and detector windows en route to the detector. Together, these effects mean XRF becomes progressively less sensitive below about atomic number 11–13 (sodium, magnesium, aluminum), and elements lighter than that (H, He, Li, Be, B, C, N, O, F) are essentially invisible to conventional XRF, requiring alternative techniques such as electron microprobe analysis with specialized light-element detectors, or entirely different methods like combustion analysis.

X-Ray Spectrometric Methods

Where this lives today

The physical picture — thorough prose

The core ideas — governing equations

Worked examples

Limiting cases, scope & common traps

The chemist behind it

Before this