Past exam of the MPhil in micro- and nanotechnology enterprise course of the University of Cambridge / 2022 / Paper 1 / Section A(i)

Diffraction contrast arises because elastic scattering transfers transmitted electron intensity between the direct beam and Bragg diffraction beams. A local change in crystal orientation, strain or thickness changes that transfer. An objective aperture selecting the direct beam makes strongly diffracting regions relatively dark in bright-field transmission electron microscopy; selecting a diffracted beam reverses the selection in dark-field transmission electron microscopy. This is a change in scattering conditions, not necessarily a change in chemical composition.

Three characteristic effects are bend contours, thickness fringes and dislocation contrast. Bend contours follow positions where a gently curved crystal meets a particular Bragg diffraction condition. Thickness fringes arise from thickness-dependent exchange of intensity between coupled transmitted and diffracted waves. A dislocation distorts nearby planes, changing their diffraction condition and producing localized contrast; changing the diffraction vector can suppress particular dislocations. Moiré fringes from overlapping lattices provide another acceptable example, but their coarse spacing is a beat spacing rather than an atomic spacing.

Use conventional bright-field transmission electron microscopy, with nearly parallel illumination, a suitably small objective aperture and low magnification covering the whole cross-section. More strongly scattering material sends more intensity outside the selected central beam and therefore looks darker. Tilt crystalline layers away from strong Bragg diffraction conditions to reduce diffraction contrast, and use a near-focus image so that electron phase contrast does not dominate.

The nominally uniform $150\mathrm{nm}$ thickness helps separate composition from thickness variations, but does not make the contrast purely compositional: density, atomic number and multiple scattering still matter. Choose an accelerating voltage sufficient to transmit this thickness, commonly $200$–$300\mathrm{kV}$ for a robust inorganic film, subject to its damage limit. High-angle annular dark-field imaging in scanning transmission electron microscopy is a useful alternative for approximately monotonic mass-thickness contrast, but is a different acquisition mode and still requires thickness and detector-angle control.

Use a low-dose, dose-efficient acquisition rather than assuming that a lower accelerating voltage is always gentler. Focus and align on a sacrificial neighbouring region, blank the beam between exposures, illuminate only when recording and use an efficient detector. A broad, low-current-density transmission electron microscopy exposure can avoid a high instantaneous local dose; scanning can also work if probe current, dwell time and repeated visits are controlled.

Radiolysis starts with inelastic scattering and ionization. For an appropriate thin specimen, increasing the electron energy can reduce the inelastic interaction probability and energy deposition per incident electron, so a relatively high voltage can help. However, increasing voltage can also activate knock-on displacement. The optimum therefore depends on the material and the amount of useful information obtained per damaging event; $150\mathrm{nm}$ alone does not determine it. Cooling may reduce diffusion-driven chemical changes but does not eliminate primary ionization. An experimental study of nonbiological layered polymer films explicitly found that the voltage choice must balance damage with image signal.^[ref]

Use high-resolution transmission electron microscopy with a coherent, well-aligned beam and an objective aperture admitting the direct beam and the spatial frequencies required at the interface. Orient a crystalline region near a suitable zone axis so that useful lattice fringes are excited; minimize drift, vibration and astigmatism. Choose defocus and aberration correction to transfer the target frequencies, rather than treating the image as a direct map of atoms at arbitrary defocus.

An aberration-corrected scanning transmission electron microscopy probe with high-angle annular dark-field imaging can instead resolve an interface with more directly interpretable atomic-number contrast. A nominal $150\mathrm{nm}$ thickness may be too great for clean atomic-resolution interpretation: thin the interface further if necessary, finish focused ion beam milling at lower ion energy to reduce damaged surface layers, and check for curtaining and implantation. The best mode depends on crystallinity and damage sensitivity; an amorphous interface calls for resolving its boundary and composition rather than inventing lattice fringes.

In piezoresponse force microscopy, a conducting atomic force microscopy tip contacts a sample with a counter-electrode. Apply $V(t)=V_{\mathrm{dc}}+V_{\mathrm{ac}}\cos \omega t$. The converse piezoelectric effect produces an oscillatory displacement, approximately $u_{\omega }=d_{\mathrm{eff}}{V}_{\mathrm{ac}}$. Contact with the moving surface bends the cantilever, and a lock-in amplifier records the displacement amplitude and phase relative to the drive.

For opposite out-of-plane ferroelectric domains, the effective piezoelectric coefficient changes sign. Their vertical response therefore differs by approximately $180^{\circ }$ in phase. Equivalent antiparallel domains can have similar amplitudes, while cancellation or reduced response near a wall can produce an amplitude minimum. Phase identifies the relative response sign; an uncalibrated detector phase is not an absolute polarization direction.

Use a small drive below the switching threshold while imaging, a stable contact force and a well-characterized contact resonance. Electrostatic forces, contact stiffness and surface charging can also create signals at the drive frequency; observing contrast alone is insufficient evidence for polarization switching.^[ref]

Measure cantilever torsion with the lateral channel of a four-quadrant optical detector. An in-plane surface displacement perpendicular to the cantilever axis twists the cantilever. The reflected laser spot moves sideways, so the left-minus-right detector signal, demodulated at $\omega$ by a lock-in amplifier, gives the amplitude and phase of the lateral piezoresponse. The top-minus-bottom signal primarily records bending.

Displacement along the cantilever axis can instead cause buckling and leak into the nominally vertical channel. Rotate the specimen by $90^{\circ }$ and repeat the measurement to sample the other in-plane component; use a reference specimen and sensitivity calibration before converting a voltage into displacement or force. A torsional spring constant and lever-arm calibration are additionally needed for an actual lateral-force value. Thus lateral piezoresponse is not automatically a calibrated force measurement.

A surface slope converts some vertical displacement into lateral motion of the tip and changes the normal contact geometry. Roughness also changes friction, contact area and local contact stiffness, and can mix cantilever bending, buckling and torsion. A lateral image that follows a height gradient can therefore appear even when the relevant in-plane polarization component is absent.

Record topography and both piezoresponse force microscopy channels simultaneously, and compare the lateral signal with the local slope rather than height alone. Reverse the scan direction to identify frictional contributions, rotate the sample relative to the cantilever, and compare the vector transformation expected for an in-plane response with the geometrical cross-talk. Repeat on a smoother region or a polished specimen, at controlled force and drive amplitude. Track contact resonance or compare sufficiently far from it so that local stiffness does not masquerade as a change in response.

A useful control is a nonpiezoelectric surface of similar roughness. Electrically switching a domain while leaving its topography unchanged provides a further comparison when feasible. No single trace–retrace test proves the absence of cross-talk: genuine in-plane response and slope-induced signals can both transform under rotation, so several controls must agree.