Past exam of the MPhil in micro- and nanotechnology enterprise course of the University of Cambridge / 2017 / Paper 1 / Section B / 8

In transmission electron microscopy, the electron objective lens forms an intermediate real-space image and intermediate/projector lenses enlarge it onto a screen or camera. The total optical magnification is the product of the relevant stage magnifications. The user changes imaging-lens excitations to select magnification, while separately controlling illumination and focus. Display resizing introduces an additional enlargement, so the recorded pixel calibration matters.

In scanning transmission electron microscopy, a focused probe is rastered across the specimen and a detector signal is assigned to each probe position. The displayed magnification is display width divided by scanned specimen width. The user increases magnification by reducing raster-deflection amplitude, independently of choosing a smaller incident probe with the probe-forming lenses. TEM changes image-forming lens magnification; STEM changes the raster field of view. A smaller field or more pixels does not create finer physical resolution than the probe, specimen scattering and detection permit. A calibrated real-space scale is more transferable than the apparent size on a monitor.^[ref]

Choose cyclic reciprocal units, so the incident and elastically scattered wavevectors have magnitude $1/\lambda$. Place the reciprocal lattice origin $O$ at the end of the incident vector from the sphere centre $C$. A reciprocal vector $\mathbf{g}$ is excited when its endpoint $G$ lies on the Ewald sphere, becauseA specimen tilt rotates the reciprocal lattice relative to this construction. The very short electron de Broglie wavelength makes the sphere nearly flat across several ordinary reciprocal-lattice spacings. In a thin foil, finite thickness $t$ broadens reciprocal points into reciprocal lattice rods approximately along the foil normal, with characteristic longitudinal extent $1/t$. A rod may intersect the sphere even when its ideal point is slightly off it. Consequently several reflections can appear simultaneously near a zone axis, and thinning changes the excitation tolerance. A small particle has finite-size broadening in other directions too. This geometrical excitation condition does not guarantee a particular spot intensity: crystal basis structure factors, specimen thickness and dynamical diffraction still matter.

Indexed electron diffraction ring positions from a polycrystalline deposit identify candidate phases and give lattice parameters after camera-length calibration. Spot patterns from a small selected area establish zone axis directions, epitaxial orientation relationships or crystal twinning. Nanobeam mapping can reveal grain orientation changes, while systematic spot shifts can measure relative strain if lens distortions and scan drift are controlled. Spot or ring broadening can reflect finite coherent size, but also strain and instrumental broadening, so it is not an unqualified particle-diameter measurement. Use a calibration standard and an empty support region, vary dose to check electron beam damage, and combine spacing information with elemental analysis when chemically distinct phases have similar lattice parameters.

The resolved periodic contrast comes principally from electron phase contrast: transmitted and diffracted electron waves interfere after passing through the image-forming optics. The resulting intensity depends on specimen thickness, orientation, defocus and the contrast transfer function; bright dots cannot automatically be labelled gold-atom positions without image simulation. The surrounding nonperiodic background is consistent with an amorphous support or contamination, but its chemistry is not determined by contrast alone.

The particle contains differently oriented fringe regions, and the Fourier transform has several opposing spot pairs with an approximately pentagonal/tenfold composite appearance. Together with the faceted outline, this is consistent with a multiply twinned gold particle, plausibly a decahedral arrangement, rather than one perfect cubic lattice with fivefold rotational symmetry. Opposite Fourier spots occur because the recorded image intensity is real; counting ten spots is not itself proof of a tenfold crystal. Overlapping domains, contrast-transfer effects and a slightly off-axis view can complicate the pattern. A tilt series or spatially windowed transforms across individual domains would test the twinning interpretation. The stability of such particles involves a competition among surface, twin-boundary and elastic strain energies.^[ref]

For cubic lattice planes, $d_{hkl}=a/\sqrt{h^2+k^2+l^2}$. The face-centred cubic selection rule permits indices that are all odd or all even. ThusThe first two spacings exceed the quoted 0.19 nm real-space resolution, whereas $d_{220}$ does not. This restriction is for the resolved image and its transform; a diffraction instrument need not have the same cutoff.

A natural candidate is a beam near a local $⟨110⟩$ zone axis. For example, the $[110]$ zone law $h+k=0$ includes four $\{111\}$ reflections and two $\{200\}$ reflections, all within that real-space cutoff. Their reciprocal-space length ratio is $g_{200}/g_{111}=2/\sqrt{3}=1.1547$. The angle between $(1\bar{1}1)$ and $(1\bar{1}\bar{1})$ reciprocal vectors is ${\cos }^{-1}(1/3)=70.53^{\circ }$. In contrast, a $[001]$ view gives a fourfold arrangement of the resolvable $\{200\}$ family, while the lowest ordinary $[111]$ zone reflections are $\{220\}$ and would not be resolved at the stated cutoff.

In a five-domain decahedral interpretation, neighboring tetrahedral fcc regions meet on $\{111\}$ twin planes around a common near-$⟨110⟩$ axis. Five unstrained wedges supply $5\times 70.53^{\circ }=352.64^{\circ }$; elastic distortion accommodates the remaining $7.36^{\circ }$. Their rotated diffraction contributions need not resemble a single six-spot $[110]$ pattern. Near-$⟨110⟩$ fcc domains in a multiply twinned particle are a defensible orientation proposal, not a unique full indexing. The printed transform has no reciprocal-space scale or indexed spot list, so it cannot establish every domain rotation or exclude an overlapping-particle contribution. Calibrated spacings and transforms from individual regions would settle those distinctions. The five-domain geometry is independently observed in atomic-resolution gold-particle measurements.^[ref]