Past exam of the MPhil in nuclear energy course of the University of Cambridge / 2020 / Module NE3 (Nuclear materials)

Original paper: Released original PDF, physical pages 32–35; questions on pages 33–34.

The characteristic effect is an irradiation-induced upward shift of the ductile-to-brittle transition in ferritic steel. A component previously used above its transition can then fail by cleavage fracture at the same service temperature, with much less warning from plastic deformation.

Ferrite has a body-centered cubic lattice. At low temperature, dislocation glide becomes difficult enough that brittle crack propagation can precede substantial plastic relaxation. Irradiation hardening raises the yield strength by introducing obstacles to dislocation glide; the competition with cleavage fracture consequently shifts toward brittle failure. Austenite, with its face-centered cubic lattice, does not normally show this same sharp cleavage transition. This does not make austenitic steel immune to loss of ductility, helium embrittlement or other radiation damage in materials.

Examine the fracture surface for cleavage fracture facets and river patterns, and compare Charpy impact test energy-versus-temperature curves from exposed material and an unirradiated reference of the same composition and heat treatment. The reference is important: a low impact energy from one failed specimen cannot establish an irradiation-induced shift. Transition curves can also use lateral expansion or the proportion of ductile fracture appearance.^[ref]

A direct fracture toughness test over the transition region complements the Charpy impact test; the two measurements are not interchangeable. Combine these observations with exposure history, local temperature, microstructure and crack location. Brittle appearance supports the failure mechanism, while a measured transition shift and suitable controls support its attribution to irradiation.

Neutron capture by boron-10 can generate helium, leading to helium embrittlement, especially at elevated temperature. The relevant nuclear reaction isThe emitted alpha particle eventually becomes a helium atom in the metal. The product branch can also involve gamma emission, without changing this helium-producing balance. The production depends on boron isotopic content and the neutron energy spectrum; a thermal component is particularly relevant to this reaction.

Helium has very low solubility in steel and can be trapped at vacancies, interfaces and grain boundaries. Boron already segregated near grain boundaries can therefore create a locally unfavorable distribution of helium. Stabilized helium–vacancy clusters develop into bubbles or cavities. Under tensile stress, boundary cavities can grow and link, reducing ductility, fracture toughness and creep-rupture life through intergranular fracture. Early thin-foil observations and mechanical tests associated high-temperature ductility loss with helium-containing boundary bubbles and their coalescence.^[ref]

Irradiation temperature controls the transport and clustering of both helium and vacancies. At lower temperature, reduced long-range transport can retain small trapped clusters in the matrix. These can contribute to irradiation hardening, but extensive boundary-cavity growth is kinetically less favorable. At higher temperature, migration, cavity growth and material creep become faster, making intergranular fracture and loss of high-temperature ductility more likely. Trap distribution and prior microstructure can redistribute the gas, so there is no composition-independent critical temperature or universal monotonic law.

The mechanical-test temperature also matters. Material irradiated cold can become embrittled during subsequent hot testing or service when trapped gas and defects become mobile. Distinguish gas production during irradiation from the later conditions under which its mechanical consequences are expressed.

The key imbalance is preferential loss of self-interstitials to biased defect sinks, leaving a vacancy supersaturation that can feed void swelling. The initial displacement cascade creates both vacancies and self-interstitials, often in equal net numbers as Frenkel pairs; it need not create more vacancies than interstitials at birth.

Many nearby Frenkel pairs recombine. Others separate, become trapped or form dislocation loops. Surviving self-interstitials are usually much more mobile than vacancies over the relevant range. Dislocations also have an interstitial bias: their elastic interaction with defects makes them more efficient sinks for self-interstitials. Preferential interstitial removal to the dislocation structure can leave the mobile vacancy population above its thermal equilibrium value.

A minimal radiation damage rate theory makes the distinction explicit:Here $c_i,c_v$ are defect number densities, $G$ is the surviving mobile-pair production rate per volume, $K$ the recombination coefficient, $D_i,D_v$ the diffusion coefficients, and $k_i^2,k_v^2$ the respective sink strengths. The vacancy loss term includes thermal emission in an effective form. Equal production terms do not require equal steady populations or equal partitioning among sinks. This simplified model omits an explicit cluster-size distribution.

A stable cavity receiving a greater vacancy influx than interstitial influx and vacancy emission grows. Vacancies therefore condense into voids, producing void swelling. Helium can stabilize small cavity nuclei, but a growing vacancy void need not be a high-pressure gas bubble. Dislocation loops, precipitates and changes in the dislocation density evolve alongside the voids and modify the competing sink strengths.

There is no violation of atom conservation: atoms represented by removed self-interstitials have joined extended defects or surfaces, while the corresponding unfilled lattice volume can accumulate in cavities. The classic electron-microscope observations of cavities in irradiated type316 steel provide direct evidence for this microscopic basis of swelling.^[ref]

Void swelling is generally favored by an intermediate irradiation temperature window. The balance involves defect mobility, stable nucleation, absorption at defect sinks and thermal emission; a larger vacancy population alone does not guarantee a larger void.

At low temperature, vacancy migration is slow. Self-interstitials may migrate and form dislocation loops, and small clusters can cause considerable irradiation hardening, while vacancies cannot readily feed large growing cavities. Defects retained in the matrix can evolve during later heating.

At intermediate temperature, vacancies can migrate far enough to reach stable cavities. Interstitial bias and the competition among defect sinks sustain a net vacancy flux into these cavities. Once the incubation period has produced suitable nuclei and sink populations, substantial void swelling can develop.

At high temperature, the thermal equilibrium vacancy population and vacancy emission from small cavities rise. A small void's curvature further favors emission, so maintaining stable nuclei becomes harder. Recovery changes the dislocation population, and recombination, gas transport and evolving precipitates also change the flux balance. The irradiation-driven vacancy supersaturation can fall, suppressing void nucleation and net swelling. Existing cavities may still coarsen, and gas bubbles, material creep and other damage can remain important.

Thus low-temperature defect retention, intermediate-temperature cavity growth and high-temperature recovery/emission are different regimes. The boundaries depend on alloy composition, initial microstructure, gas production, accumulated displacements per atom and displacement dose rate. No alloy-independent peak temperature can be inferred from the paper's data.

Irradiation softening occurs when irradiation removes pre-existing strengthening obstacles faster than it creates new ones. It is particularly plausible in an initially hardened alloy, rather than an initially soft annealed reference.

In a material strengthened by cold working, the starting dislocation density is high. Irradiation-generated point defects can assist dislocation climb, rearrangement and annihilation, reducing the contribution of work hardening. A schematic estimate iswhere $\mu$ is the shear modulus, $b$ the magnitude of the Burgers vector, ${\rho }_d$ the dislocation density, and the last two terms represent precipitate and irradiation-obstacle strengthening. It is an illustrative decomposition, not an exact additive law for every alloy. A sufficiently large decrease in the pre-existing contributions can exceed the increase in $\Delta {\sigma }_{\mathrm{irr}}$.

In a material strengthened by precipitation hardening, ballistic mixing can dissolve or disorder the original obstacles. Radiation-enhanced diffusion can change their composition or accelerate coarsening; increasing obstacle spacing can reduce strength. In an ordered alloy, irradiation-induced loss of order can also remove a strengthening contribution. None of these outcomes is inevitable: irradiation can instead create new strengthening precipitates.

The required conditions therefore include an appropriate initial microstructure, enough displacement dose to modify it, and a temperature and displacement dose rate at which obstacle removal or recovery can compete with new defect accumulation. Compare with an unirradiated specimen held for the same time at the same temperature to separate ordinary thermal aging from an irradiation effect.

A drop in yield strength or hardness need not restore ductility: cavities, helium embrittlement or localization of dislocation glide can coexist with softening. Conversely, an initially annealed steel may show strong irradiation hardening under the same exposure conditions.

Irradiation creep adds stress-directed, time-dependent strain driven by the irradiation-generated defect population. It can be appreciable where ordinary thermal material creep would be slow, because irradiation maintains mobile point defects far above their equilibrium populations.

Absorption or emission of vacancies and self-interstitials permits dislocation climb. Applied stress can make differently oriented dislocations or dislocation loops absorb defects at different rates: stress-induced preferential absorption then produces a directional strain. Climb can also allow a pinned dislocation to bypass an obstacle and continue dislocation glide, providing a second route to strain.

A useful small-strain empirical form, over a limited range of conditions, isfor constant stress $\sigma$ and approximately constant irradiation-creep compliance $B$. Here $d$ is the accumulated displacements per atom and $\dot{d}$ the displacement dose rate; $B$ is measured per unit stress and per unit dpa. This relation explains why creep can scale approximately with both stress and neutron flux when the damage spectrum is fixed. It is not a constitutive law valid at every dose or temperature.

The total strain can also include thermal material creep, void swelling, irradiation growth and changes due to precipitation or recovery. Stress-free swelling is not itself irradiation creep. Experiments on pressurized type316 tubes separated these contributions and found that correcting stress-enhanced swelling materially affected comparisons between annealed and specimens strengthened by cold working.^[ref]

At sufficiently high temperature, thermal material creep can dominate. Changes in microstructure caused by irradiation hardening, softening, bubbles and cavities can either alter the creep rate or shorten rupture life. Thus enhanced defect-driven creep is the central effect, while the net rate and lifetime remain material- and history-dependent.

Assume a homogeneous Fe–Cr–Ni alloy, normal incidence, independent scattering events and the stated effective scattering cross-sections. Neglect absorption, other constituents and changes in neutron energy within the thin sheet. The atomic fractions are not supplied; take $x_{\mathrm{Fe}}=0.70$, $x_{\mathrm{Cr}}=0.20$, $x_{\mathrm{Ni}}=0.10$ as an explicit representative assumption, not a composition measured or specified in the paper.

The conventional cell of a face-centered cubic lattice containsatoms. The number density is thereforeFor a mixture, the scattering probability per unit path length is ${\Sigma }_s={\sum }_j{n}_j{\sigma }_j=n{\bar{\sigma }}_s$, whereThus ${\Sigma }_s=27.751{\mathrm{m}}^{-1}$. For thickness $t=10^{-3}\mathrm{m}$, the unscattered fraction $U$ satisfies $dU/dz=-{\Sigma }_{s}U$ with $U(0)=1$. The fraction undergoing at least one neutron scattering event isThe thin-sheet approximation gives $P_s\simeq {\Sigma }_{s}t=2.78\%$, consistent to the accuracy of the assumed composition. A simple iron-dominated estimate ${\bar{\sigma }}_s\simeq 3.2$ barns gives essentially the same about $3\%$ scattered conclusion.

For any other assumed composition, use the displayed weighted formula; mass fractions must first be converted to atomic fractions. Since the supplied elemental cross-sections span $2.8$–$4.1$ barns, any mixture of these three elements at the stated number density gives approximately $2.39$–$3.48\%$. This is a bound within the idealized mixture model, not a claim that every such composition is stable austenite.

Use the same atomic fractions and collision assumptions as in part(c). Let $A$ be the irradiated area. The number of atoms in the sheet is $N_t=ntA$, while the number of incident neutrons is $\Phi A$, where $\Phi$ is the neutron fluence.

Counting one primary knock-on atom for each neutron that scatters and using $\nu =430$ displacements per recoil gives the first-collision estimateHence the nominal damage is about $55$ displacements per atom under the assumed composition. In the thin-sheet approximation the density and thickness cancel:The $1.4\%$ difference comes from using first-collision attenuation rather than the linear approximation. Treating repeated collisions after deflection would require angular and energy information not provided; this correction is already smaller than the precision warranted by the composition assumption.

There is also a counting convention in “other atoms”: $430$ counts secondary displacements, while including the displaced primary knock-on atom itself gives $\nu =431$. The corresponding first-collision result is $54.58$ rather than $54.45$ dpa, an increase of only $0.23\%$. Both conventions round to about $55$ dpa. The numerical estimate is explicitly conditional on assigning the stated displacement yield to each scattering event.

A value greater than one is physically possible. Displacements per atom counts displacement events relative to the initial atom inventory: atoms can be displaced repeatedly, and many vacancies and self-interstitials recombine. It does not mean $55$ surviving vacant sites per original lattice site, or that every atom has moved exactly $55$ times. This is a large nominal exposure for which evolving dislocation loops, cavities and irradiation hardening matter; the surviving microstructure additionally depends on irradiation temperature, displacement dose rate, recombination and defect sinks.