Materials characterisation: XRD, SEM and TEM

In short

Materials characterisation is the set of experimental techniques used to find out what a material is made of, how its atoms are arranged, and what its internal structure looks like at scales from millimetres down to atomic spacing. X-ray diffraction (XRD) identifies crystalline phases and measures lattice parameters, residual stress, texture and crystallite size; the scanning electron microscope (SEM) images surfaces and fracture faces with composition by EDS and orientation by EBSD; the transmission electron microscope (TEM) resolves dislocations, precipitates, interfaces and atomic planes in thin foils. Every metallurgy and materials syllabus (AICTE model curriculum, UGC CBCS) teaches these as a core laboratory and theory course, and GATE MT asks Bragg's law, indexing, structure factor, Scherrer broadening and electron–matter interaction almost every year. Indian employers — Tata Steel, JSW, SAIL R&D, BARC, DRDO (DMRL Hyderabad), ISRO/VSSC, HAL, GE, Hindalco, pharma and battery firms — run these instruments in quality control, failure investigation and R&D, and interviewers test them hard.

Explained

Origin of diffraction and Bragg's law

• X-rays (wavelength 0.05–0.25 nm) are comparable to interatomic spacing, so a crystal acts as a three-dimensional diffraction grating; visible light cannot resolve lattice planes.
• Bragg's law: constructive interference occurs when the path difference 2d sinθ between rays scattered from successive parallel planes equals a whole number of wavelengths, nλ = 2d sinθ. θ is the Bragg angle measured from the plane, not the normal; the detector moves to 2θ.
• Since sinθ ≤ 1, diffraction needs λ ≤ 2d — a maximum of n values is observable for a given λ; n > 1 reflections are conventionally re-written as first order from (nh nk nl).
• Laue condition (equivalent, reciprocal-space form): diffraction when the scattering vector equals a reciprocal lattice vector, Δk = G. The Ewald sphere construction shows which reciprocal lattice points can diffract.
• Sources: sealed X-ray tube with Cu (λ Kα₁ ≈ 0.15406 nm, Kα average ≈ 0.1542 nm), Co, Cr or Mo target; Kβ removed by a Ni filter or monochromator. Cu radiation fluoresces iron, so Co or Cr tubes are preferred for steels. Synchrotron sources (Indus-1 and Indus-2 at RRCAT Indore) give high-brilliance tunable X-rays.
• Detection: scintillation counter, position-sensitive or 2D area detectors; modern Bragg–Brentano powder diffractometers scan 2θ from about 10° to 120°.

XRD: structure factor, indexing and extinction

• The intensity of a reflection ∝ |F_hkl|², where the structure factor F_hkl = Σ f_j exp[2πi(h x_j + k y_j + l z_j)] summed over all atoms j in the unit cell with atomic scattering factor f_j and fractional coordinates (x,y,z).
• Selection (extinction) rules follow from F_hkl = 0:
  • Simple cubic — all (hkl) allowed.
  • BCC — h+k+l must be even (110, 200, 211, 220, 310, 222…).
  • FCC — h, k, l all odd or all even (111, 200, 220, 311, 222, 400…).
  • Diamond cubic — FCC rule plus h+k+l ≠ 4n+2, so 200 is absent (111, 220, 311, 400…).
  • HCP — (00·l) absent for odd l; (hk·l) with h+2k = 3n and l odd absent.
• Indexing a cubic pattern: 1/d² = (h²+k²+l²)/a², so sin²θ ∝ (h²+k²+l²). Form the ratios sin²θᵢ/sin²θ₁ and match to the allowed integer sequence — 1:2:3:4:5:6:8 for BCC, 3:4:8:11:12:16 for FCC. This is a standard GATE and viva question.
• Other intensity factors: multiplicity, Lorentz–polarisation factor, temperature (Debye–Waller) factor, absorption factor, and preferred orientation.
• Rietveld refinement: whole-pattern least-squares fitting of a structural model to the observed profile; gives quantitative phase fractions, lattice parameters, site occupancy and microstrain. Goodness judged by R_wp and χ² (GoF); common software: FullProf, GSAS-II, TOPAS.

What XRD measures in practice

• Phase identification: match peak positions and relative intensities against the ICDD PDF database (JCPDS cards). Detects crystalline phases typically above about 1–3 vol% — amorphous content shows only as a broad hump.
• Quantitative phase analysis: retained austenite in quenched/carburised steel by the direct comparison method (ASTM E975 is the common international practice); ferrite–austenite balance in duplex stainless steel.
• Lattice parameter and solid solubility: peak shift gives a; Nelson–Riley or cos²θ/sinθ extrapolation to θ = 90° removes systematic errors; Vegard's law relates a to solute content.
• Crystallite (domain) size: Scherrer equation from peak broadening, valid roughly 2–100 nm. Instrumental broadening must first be removed using a standard (LaB₆, Si).
• Microstrain and dislocation density: Williamson–Hall plot separates size broadening (1/cosθ dependence) from strain broadening (tanθ dependence).
• Residual stress: sin²ψ method — measure the shift of a high-angle peak at several tilt angles ψ; stress from the slope using X-ray elastic constants. Non-destructive, samples only a few tens of micrometres depth.
• Texture: pole figures from a four-circle/Eulerian cradle; quantified as an orientation distribution function (ODF). Critical for deep-drawing steels and electrical steel (Goss texture).
• Thin films: glancing-angle XRD (GIXRD) to limit penetration; X-ray reflectivity (XRR) for thickness, density and roughness.
• Limitations: no information on individual grains or morphology, poor for amorphous and very dilute phases, averages over a large illuminated area (mm²), needs flat powder or polished specimen, light elements scatter weakly.

SEM: electron–specimen interaction, signals and attachments

• A focused electron beam (typically 1–30 kV) is scanned in a raster over the specimen; signals collected point by point build the image. Resolution typically 1–10 nm; magnification about 10× to 300,000×; depth of field is very large, which is why SEM fractographs look three-dimensional.
• Electron guns: thermionic tungsten hairpin (cheap, robust), LaB₆ (brighter), field emission (FEG) — highest brightness and best low-kV resolution.
• Interaction volume is teardrop-shaped, roughly 0.1–5 µm deep; it grows with accelerating voltage and shrinks with atomic number. Signals come from different depths:
  • Secondary electrons (SE): energy < 50 eV, escape only from the top few nanometres — give topographic contrast and the best resolution; collected by an Everhart–Thornley detector.
  • Backscattered electrons (BSE): high energy, elastically scattered, yield rises with atomic number Z — give compositional (Z) contrast; heavy phases appear bright. Also used for channelling/orientation contrast.
  • Characteristic X-rays: from deeper in the interaction volume — give elemental analysis (EDS/EPMA).
  • Also Auger electrons (surface-sensitive), cathodoluminescence, absorbed/specimen current, and transmitted electrons in STEM mode.
• EDS (energy-dispersive spectroscopy): Si(Li) or silicon-drift detector measures X-ray energies; Moseley's law links energy to Z. Fast, all elements at once, typically Be/B upward; energy resolution about 125–140 eV at Mn Kα, so peak overlaps occur. Quantification by ZAF or φ(ρz) correction; accuracy a few per cent for major elements, poor below about 0.1 wt%. Cannot detect hydrogen or helium, and light-element analysis (C, N, O) is semi-quantitative.
• WDS / EPMA (electron probe microanalyser): diffracting crystal spectrometer, energy resolution about 10 eV, detection limits down to tens of ppm — the standard for accurate quantitative microanalysis and light elements.
• EBSD (electron backscatter diffraction): Kikuchi patterns from a 70°-tilted specimen give grain orientation, grain size, grain-boundary character, phase discrimination, texture and local misorientation (KAM for stored strain). Needs excellent, strain-free surface preparation (final electropolish or colloidal silica).
• Specimen preparation: sectioning, mounting, grinding, polishing, etching (nital, picral, Keller's reagent) for microstructure; fracture surfaces examined as-received, unetched and uncleaned chemically. Non-conductors (ceramics, polymers, oxide scale) are sputter-coated with Au/Au-Pd or carbon, or viewed in low-vacuum / variable-pressure mode. ESEM allows hydrated samples.
• Artefacts: charging (bright flashes, image drift), beam damage of polymers, edge effect, astigmatism, contamination.

TEM: optics, imaging modes and diffraction

• Electrons at 100–300 kV pass through a specimen thinner than about 100 nm (ideally < 50 nm). Resolution 0.1–0.3 nm in modern instruments; atomic-resolution imaging is routine with aberration correction.
• de Broglie wavelength λ = h/√(2 m₀ e V) non-relativistically; relativistic correction matters above about 100 kV — at 200 kV λ ≈ 0.00251 nm, far below visible light, which is why resolution is so high. Practical resolution is limited by spherical aberration of the objective lens, not by λ.
• Column: electron gun → condenser lenses (illumination) → specimen → objective lens (forms diffraction pattern in its back focal plane and image in the image plane) → intermediate and projector lenses → fluorescent screen/CCD/direct-electron detector.
• Imaging modes:
  • Bright field (BF): objective aperture admits the direct beam; thicker, heavier or strongly diffracting regions appear dark (mass–thickness and diffraction contrast).
  • Dark field (DF): aperture admits one diffracted beam; the feature producing that reflection lights up — used for precipitate identification and stacking faults.
  • Diffraction contrast: dislocations, stacking faults and strain fields imaged via two-beam conditions; g·b = 0 invisibility criterion determines the Burgers vector.
  • Phase contrast (HRTEM): many beams interfere to show lattice fringes and atomic columns; interpretation needs image simulation.
  • STEM: focused probe scanned; HAADF (high-angle annular dark field) gives Z-contrast (∝ Z^1.7 approximately), ideal for heavy precipitates and interfaces.
• SAD (selected area diffraction): a spot pattern from a small region; camera equation R·d = λ·L, where R is spot distance from the centre and L the camera length. Gives crystal structure, orientation, zone axis, and orientation relationships between phases (e.g. Kurdjumov–Sachs in steels). Rings indicate polycrystalline/nanocrystalline, diffuse halo indicates amorphous, extra spots indicate superlattice ordering, streaks indicate planar faults. Kikuchi lines allow precise orientation determination.
• Analytical TEM: EDS for point/line/map composition; EELS (electron energy loss spectroscopy) for light elements, bonding, valence state and thickness; energy-filtered TEM (EFTEM) elemental maps.
• Specimen preparation — the hardest part: punch 3 mm discs → mechanical thinning/dimpling → twin-jet electropolishing (metals) or ion milling (PIPS) for ceramics; FIB lift-out for site-specific samples (semiconductor cross-sections, coatings); ultramicrotomy for polymers and biological samples; carbon replicas and carbon extraction replicas for precipitate identification in steels; dispersion on a holey carbon grid for powders and nanoparticles.
• Limitations: destructive and tedious preparation, tiny sampled volume (statistically unrepresentative), thin-foil relaxation changes dislocation structure, beam damage and contamination, expensive (a 200 kV FEG-TEM runs into tens of crores of rupees with facility costs); available in India mainly at IITs, IISc, NITs, CSIR labs, BARC, DMRL and SAIFs (Sophisticated Analytical Instrument Facilities).

Choosing a technique, and the standards that govern the work

• Length-scale ladder: eye/stereo microscope (mm) → optical microscope (0.2 µm limit, Abbe criterion) → SEM (nm, surfaces) → TEM/STEM (atomic, thin foils) → atom probe tomography (atomic, 3D composition). XRD sits alongside, averaging over a volume rather than imaging.
• Decision rule: "what phases and how much, averaged?" → XRD. "What does it look like and what is this particle?" → SEM+EDS. "Dislocations, nanoprecipitates, interfaces, lattice" → TEM. "Accurate composition of a 1 µm feature" → EPMA/WDS. "Bulk chemistry" → XRF, OES spark spectrometry or wet chemistry.
• Complementary techniques: XRF (bulk elemental, non-destructive, used routinely in Indian steel and cement plants), XPS/ESCA (surface chemistry and oxidation state, top 10 nm), AFM (surface topography and roughness), DSC/TGA (thermal), optical metallography (ASTM E112-type grain size), micro-CT (3D internal porosity).
• Standards and practice in India: BIS standards cover metallographic practice and grain size determination (quote the current IS number only after checking the BIS catalogue, as several older metallography IS numbers have been revised or withdrawn); ASTM E3 (metallographic specimen preparation), E407 (etchants), E112 (grain size), E975 (retained austenite), E986 (SEM electron beam size characterisation) and E766 (SEM magnification calibration) are the industry references in Indian labs. Testing laboratories are accredited by NABL under ISO/IEC 17025; instruments need traceable calibration (NIST SRM 640/660 or Si/LaB₆ for XRD, magnification standards and Cu/Co for SEM-EDS).
• Radiation safety: X-ray diffractometers and XRF units are radiation-generating equipment regulated by the Atomic Energy Regulatory Board (AERB) under the SHANTI Act, 2025 (which replaced the Atomic Energy Act, 1962) and the Atomic Energy (Radiation Protection) Rules, 2004 (framed under the 1962 Act; check for rules notified under the new Act) — registration/licensing through the e-LORA portal, a designated Radiological Safety Officer for applicable installations, interlocked enclosures and TLD personnel monitoring (BARC badge service). Workplace safety and occupational health in the lab now fall under the Occupational Safety, Health and Working Conditions Code, 2020, one of the four Labour Codes in force from 21 November 2025, which replaced the Factories Act, 1948.

Key points

• Bragg's law — nλ = 2d sinθ; θ is measured from the diffracting plane, and diffraction needs λ ≤ 2d.
• Interplanar spacing (cubic) — d = a/√(h²+k²+l²); peak positions therefore give the lattice parameter.
• Structure factor — decides intensity; F = 0 gives extinction, so BCC shows h+k+l even and FCC shows all-odd or all-even indices.
• First peak — BCC starts with (110), FCC with (111), diamond cubic with (111) but misses (200).
• Scherrer equation — D = Kλ/(β cosθ); broadening increases as crystallite size falls below about 100 nm.
• Williamson–Hall — separates crystallite size from microstrain using the different θ-dependence of the two broadenings.
• Rietveld refinement — whole-pattern fitting for quantitative phase fractions and structural parameters.
• Secondary electrons — low energy, surface-sensitive, topographic contrast, best SEM resolution.
• Backscattered electrons — high energy, yield rises with Z, give atomic-number (compositional) contrast.
• Interaction volume — teardrop region 0.1–5 µm deep; sets the spatial resolution of X-ray microanalysis, not the probe size.
• EDS vs WDS — EDS is fast and simultaneous but ~130 eV resolution; WDS/EPMA is slow, ~10 eV, accurate to ppm.
• EBSD — Kikuchi patterns at 70° tilt give orientation, grain size, texture and boundary character.
• TEM specimen — must be electron-transparent (< 100 nm); prepared by electropolishing, ion milling or FIB.
• Bright field vs dark field — direct beam vs a chosen diffracted beam; dark field highlights the feature that produced the reflection.
• g·b = 0 — the invisibility criterion used to determine a dislocation's Burgers vector.
• Camera equation — R d = λ L; converts SAD spot spacings into interplanar spacings.
• HAADF-STEM — Z-contrast imaging, intensity rising roughly as Z^1.7.
• Technique choice — XRD averages phases over a volume; SEM shows the surface; TEM shows defects and lattice; EPMA quantifies micro-composition.
• Regulation — XRD/XRF installations are registered with AERB (e-LORA); test labs are NABL-accredited to ISO/IEC 17025.

Formulas and examples

Example 1 — Indexing a cubic pattern and finding the lattice parameter. A powder pattern taken with Cu Kα (λ = 0.1542 nm) shows first peaks at 2θ = 44.0°, 64.2° and 81.4°.

• θ = 22.0°, 32.1°, 40.7°. sinθ = 0.3746, 0.5315, 0.6521 → sin²θ = 0.1403, 0.2825, 0.4252.
• Ratios to the first: 1 : 2.013 : 3.030 ≈ 1 : 2 : 3 → h²+k²+l² = 2, 4, 6 → (110), (200), (211) → BCC.
• From the first peak: a = λ√2/(2 sinθ) = 0.1542 × 1.4142 / (2 × 0.3746) = 0.2181/0.7492 = 0.2911 nm — consistent with α-iron (0.2866 nm), i.e. a BCC ferritic phase slightly expanded by solute.
• Density check for BCC Fe with a = 0.2911 nm: ρ = 2 × 55.85 / (6.022×10²³ × (2.911×10⁻⁸)³) = 111.7 / (6.022×10²³ × 2.467×10⁻²³) = 111.7/14.86 = 7.52 g/cm³.

Example 2 — Crystallite size by the Scherrer equation. A nano-TiO₂ powder gives a peak at 2θ = 25.3° with measured FWHM 0.52° and instrumental broadening 0.10° (Cu Kα, K = 0.9).

• Corrected broadening (quadratic subtraction): β = √(0.52² − 0.10²) = √(0.2704 − 0.01) = √0.2604 = 0.510°.
• In radians: β = 0.510 × π/180 = 8.90 × 10⁻³ rad.
• θ = 12.65°, cosθ = 0.9757.
• D = 0.9 × 0.1542 / (8.90×10⁻³ × 0.9757) = 0.13878 / 8.684×10⁻³ = 15.98 nm ≈ 16 nm.
• Note: this is the coherent domain size, not the SEM-visible particle size, which is usually larger because particles are agglomerates of several domains.

Example 3 — Retained austenite in a carburised gear by direct comparison. Integrated intensities: austenite (220)_γ = 180 counts·° with R_γ = 20.1; ferrite/martensite (211)_α = 1,050 counts·° with R_α = 52.5 (R values from standard tables for Cr Kα).

• I_γ/R_γ = 180/20.1 = 8.955; I_α/R_α = 1050/52.5 = 20.00.
• V_γ = 8.955/(8.955 + 20.00) = 8.955/28.955 = 0.309 → about 31 vol% retained austenite — far above the 10–15% usually specified for case-carburised gears, so sub-zero treatment or a tempering change is needed.

Example 4 — TEM electron wavelength and resolution argument. For a 200 kV TEM:

• Non-relativistic λ = 1.226/√V nm·V^½ = 1.226/√200000 = 1.226/447.2 = 2.74 × 10⁻³ nm.
• Relativistic correction factor: 1 + eV/(2m₀c²) = 1 + 200/(2 × 511) = 1.1957; λ = 2.74×10⁻³/√1.1957 = 2.74×10⁻³/1.0935 = 2.51 × 10⁻³ nm (0.0251 Å).
• That is about 220,000 times smaller than green light (550 nm), yet real TEM resolution is only ~0.1 nm — because lens spherical aberration, not wavelength, is the limit.

Example 5 — Reading a selected-area diffraction pattern. A SAD pattern of an aluminium alloy foil (200 kV, λ = 0.00251 nm, camera length L = 800 mm) has its nearest spot at R = 10.3 mm.

• d = λL/R = 0.00251 × 800 / 10.3 = 2.008/10.3 = 0.195 nm.
• For FCC Al (a = 0.4049 nm), d₍₂₀₀₎ = 0.4049/2 = 0.2025 nm and d₍₂₂₀₎ = 0.4049/2.828 = 0.1432 nm → the spot is (200), within about 4% of the expected value (camera length calibration error).

Where it is used

• Integrated steel plants (Tata Steel Jamshedpur, JSW, SAIL, RINL): XRD for sinter/pellet phase analysis, iron-ore and coke mineralogy, retained austenite and texture of auto-grade and electrical steels; SEM-EDS for inclusion rating, scale and coating defects; XRF for routine bulk chemistry.
• Failure investigation and quality labs: SEM fractography to distinguish ductile dimples, brittle cleavage, fatigue striations/beach marks, intergranular SCC and creep cavitation — routine in boiler, railway, aerospace and automotive component failures; often reported alongside NDT findings in a NABL-accredited report.
• Aerospace, defence and nuclear R&D (DMRL Hyderabad, VSSC/ISRO, BARC, IGCAR, HAL, GTRE): TEM for γ′/γ″ precipitates in superalloys, radiation damage and void swelling in cladding, oxide-dispersion-strengthened steels, titanium and shape-memory alloys.
• Electronics, semiconductors and batteries: GIXRD and XRR for thin films, FIB-TEM cross-sections for device failure analysis, operando XRD and TEM for Li-ion cathode degradation (ISRO, C-MET, battery start-ups and IIT labs).
• Cement, ceramics, refractories, pharma and geology: quantitative Rietveld analysis of clinker phases (C₃S, C₂S, C₃A, C₄AF), polymorph/crystallinity checks in drug substances (CDSCO/ICH-driven documentation), mineral identification in exploration cores.
• Job roles: metallurgist/materials engineer in QA or R&D, failure analyst, characterisation facility scientific officer at SAIFs, applications and service engineer with instrument vendors (Bruker, Rigaku, PANalytical/Malvern, Thermo Fisher, JEOL, Zeiss, TESCAN) — roles typically asking for hands-on XRD/SEM experience plus data interpretation.

Interview questions

• Q: Why are X-rays used for crystal structure determination instead of visible light? A: Diffraction needs a wavelength comparable to the spacing being probed; X-ray wavelengths (~0.15 nm) match interatomic distances, while visible light (~550 nm) is thousands of times too long.
• Q: Why does an FCC crystal show no (100) reflection? A: The face-centre atoms scatter exactly out of phase with the corner atoms for mixed-index reflections, so the structure factor is zero; only all-odd or all-even (hkl) survive.
• Q: Two patterns give peak ratios 1:2:3:4 and 3:4:8:11. Which is BCC and which FCC? A: 1:2:3:4 (h²+k²+l² = 2,4,6,8, all even sums) is BCC; 3:4:8:11 is FCC starting with (111) and (200).
• Q: A peak is broad. Is the crystallite small or the material strained? A: Both broaden peaks, but differently with angle — size broadening scales as 1/cosθ and strain as tanθ; a Williamson–Hall plot separates them.
• Q: What limits XRD's ability to detect a minor phase? A: Its scattering power and peak overlap — typically below about 1–3 vol% peaks disappear into the background, and amorphous phases give no peaks at all.
• Q: Why use Co or Cr radiation rather than Cu for steels? A: Cu Kα excites iron fluorescence, which raises the background badly; Co or Cr radiation avoids it and also spreads peaks to higher 2θ for better resolution.
• Q: Difference between secondary and backscattered electron images? A: SE come from the top few nanometres and show topography at the best resolution; BSE are high-energy, come from deeper, and their yield rises with atomic number, giving compositional contrast.
• Q: Why does SEM-EDS have poorer spatial resolution than the SEM image? A: X-rays are generated throughout the micrometre-scale interaction volume, so analysis resolution is set by that volume, not by the electron probe diameter.
• Q: Can EDS detect hydrogen or quantify carbon reliably? A: No — H and He have no characteristic X-rays, and light elements such as C, N and O suffer absorption and overlap problems, so carbon is at best semi-quantitative; use combustion analysis or EPMA/EELS instead.
• Q: Why must a TEM specimen be so thin, and what is the risk of thinning? A: Electrons must transmit with limited multiple scattering, so thickness must be under about 100 nm; thinning can relax dislocation structures, introduce artefacts, cause preferential etching and heat or ion damage.
• Q: What does the g·b = 0 criterion tell you? A: If a dislocation is invisible for two independent g vectors, its Burgers vector is perpendicular to both — that fixes b, the standard TEM method for identifying dislocation type.
• Q: You have a fractured gear from the field. What sequence of characterisation would you run? A: Visual and stereo examination first, then unetched SEM fractography of the as-received surface to find the initiation site and mode, EDS on any deposits or inclusions, then sectioning for metallography, hardness and case-depth, with XRD for retained austenite and residual stress if case quality is suspect.
• Q: Why is EBSD preparation so demanding? A: Kikuchi patterns come from a very shallow, lattice-perfect layer; any mechanical deformation or relief from polishing destroys the pattern, so a final electropolish or colloidal-silica step is needed.

Common mistakes

• Measuring θ from the surface normal, or substituting the 2θ value straight into Bragg's law — always halve 2θ first.
• Using FWHM in degrees in the Scherrer equation, or forgetting to subtract instrumental broadening (both inflate or deflate D badly); also quoting Scherrer size as "particle size".
• Claiming a material is single-phase because XRD shows one phase set, ignoring amorphous content and the detection limit, or that it is amorphous when the peaks are merely very broad.
• Confusing secondary with backscattered electrons, and assuming EDS point resolution equals the probe size.
• Cleaning or etching a fracture surface, or coating it before examining it as-received — the evidence of the failure mode is destroyed.
• Treating a TEM observation of a few square micrometres as representative of a whole heat or component, or over-interpreting HRTEM lattice images without simulation.
• Mixing up the diffraction pattern plane with the image plane in the TEM, and forgetting that camera length must be calibrated before converting R to d.
• Reporting EDS results to two decimal places as if they were wet-chemistry accurate, or quoting normalised totals that hide an unanalysed element.

Related topics

Crystallography and crystal structures · Phase diagrams and phase transformations · Non-destructive testing · Nanomaterials · Mechanical behaviour and fracture · Heat treatment of steels