Past exam of the MPhil in micro- and nanotechnology enterprise course of the University of Cambridge / 2021 / Paper 1 / 8

The main characteristic peaks identify nickel and phosphorus: the spectrum has a strong Ni $K\alpha$ peak near $7.48\mathrm{keV}$, the Ni $L$ region near $0.85\mathrm{keV}$ and P $K\alpha$ near $2.01\mathrm{keV}$. Weaker associated lines, including Ni $K\beta$ near $8.26\mathrm{keV}$, can accompany them. The continuous background is bremsstrahlung from decelerated electrons.

Small carbon and oxygen signals at low energy can arise from contamination, surface oxide or preparation materials; a copper contribution can arise from the adjacent substrate within the finite interaction volume. The Cu $L$ region lies close to Ni $L$, so overlap is particularly troublesome at low energy. A weak Cu $K\alpha$ contribution near $8.04\mathrm{keV}$ is better separated, though still close to Ni $K\beta$. At $30\mathrm{kV}$ the beam can excite these K lines.

Peak height is not an atomic fraction: ionization cross-sections, fluorescence yields, absorption, detector response, background and line overlap require correction. The large nickel peak therefore supports the nickel-rich assignment but does not directly give the nickel-to-phosphorus atomic ratio.

Use the supplied polynomial as an effective-atomic-number approximation for the alloys. Interpreting each quoted phosphorus number as an atomic fraction $x_P=N_P/(N_P+N_{\mathrm{Ni}})$,The gold reference yield is $\eta (79)=0.486996$. Substitution gives:

The ordering is gold brightest, then copper, NiP${}_z$, NiP${}_x$ and NiP${}_y$. The phosphorus-rich layer has the lowest effective atomic number.

There is a wording ambiguity in “atomic fraction of P relative to Ni.” If those numbers instead mean $r=N_P/N_{\mathrm{Ni}}$, first convert $x_P=r/(1+r)$, giving $\bar{Z}=(28+15r)/(1+r)$. The three alloy contrasts then become $0.4243$, $0.4361$, $0.4129$, respectively. The qualitative ordering is unchanged. Both calculations explicitly use $\eta (\bar{Z})$; this simple alloy approximation is not a full electron-transport simulation.

Atomic-number contrast can explain why the gold-rich regions are bright and why the phosphorus-rich NiP${}_y$ layer should be somewhat darker than the other nickel–phosphorus layers. However, the calculated alloy yields differ only modestly: the NiP${}_y$ and NiP${}_z$ yields differ by about $5.5\%$ relative to NiP${}_z$. Composition-only backscattering does not explain all of the strong local edge and defect contrast.

An Everhart–Thornley detector normally collects secondary electrons, not a pure backscattered-electron signal. Secondary-electron escape and collection depend strongly on local slope and edge geometry. Differential polishing, relief at interfaces, a crack or cavity, surface oxidation and local charging can therefore alter brightness without the corresponding change in bulk atomic number. backscattered electrons can also generate secondary electrons elsewhere. Crystallographic channeling and orientation can contribute additional contrast. The observed dark and bright areas should therefore be correlated with topography and composition rather than assigned a phase from intensity alone.

In high-angle annular dark-field imaging scanning transmission electron microscopy, electrons scattered to large angles give predominantly incoherent mass–thickness and atomic-number contrast. For comparable thickness, gold is much brighter than nickel–phosphorus because its atomic number is much larger. carbon in the protective cap scatters weakly, whereas its tungsten component scatters strongly. The cap is not a homogeneous low-atomic-number reference.

The gold layer follows a curved, locally distorted interface. The dark central opening in the overview is consistent with a crack, void or locally missing/thinned foil: little material along the beam path gives little high-angle scattering. The granular and columnar variations beneath the gold indicate nanoscale heterogeneity in the affected NiP${}_y$ region. Differences in thickness, density, porosity and composition can all contribute; diffraction/channeling effects are reduced in high-angle annular dark-field imaging but not eliminated in every geometry.

A black region is therefore evidence of low projected scattering mass, not by itself the identity of a chemical phase. In particular, a single high-angle annular dark-field imaging image does not distinguish a pre-existing void from material lost during specimen preparation.

The thin scanning transmission electron microscopy foil restricts scattering to a much smaller projected specimen volume than bulk SEM at $30\mathrm{kV}$. It reveals the nanoscale shape and continuity of the gold/NiP interface, the narrow low-density opening and the fine heterogeneous region underneath the gold, which the larger SEM interaction volume and surface-sensitive collection obscure.

The enlarged view resolves structure on the tens-of-nanometres scale: local interface roughness, small bright and dark domains and the relationship of the gold layer to the damaged or porous-looking NiP${}_y$ region. This helps distinguish an interface-associated defect from merely a change between nominal alloy layers. It still does not identify a crystalline phase or quantify composition without diffraction or spectroscopic analysis. Since the SEM surface and focused ion beam milling lamella need not cut precisely the same volume, correlations should allow for section position.

The conspicuous approximately vertical streaks are focused ion beam milling curtaining: different sputtering rates beneath the rough, compositionally varying cap and interfaces leave stripes of different foil thickness. The uneven lower foil edge, contrast gradient and locally thinned regions are also consistent with nonuniform milling; redeposition can add material to other regions.

A dark gap or pull-out may have been enlarged or created during milling, so its original dimensions cannot be inferred securely from this one image. ion implantation and an amorphized surface layer are additional possible focused ion beam milling effects, but they are not independently demonstrated by these intensity variations. A lower-energy final polish, improved protective cap and examination of another lamella could test the preparation contribution.