Cryogenic response and temperature-dependent failure

Cooling changes mobility and dimensions

Low temperature can change plastic relaxation, phase state, electrical behaviour and thermal contraction. Some materials become more fracture-sensitive under particular conditions; others retain useful ductility. Avoid applying one family’s transition to all materials.

Imagine a cold assembly whose metal support needs more stress to yield, yet whose notched specimen fractures with less plastic deformation. At the same time, its joint may acquire a new stress as the components contract differently. There is no contradiction: yielding, crack resistance and dimensional compatibility answer different questions. The relevant temperature is the material’s actual state during loading, rather than only the surrounding coolant or a later room-temperature measurement.

Yield stress marks the onset of the defined plastic response in a specified test. Fracture resistance concerns crack advance under a stated loading and constraint; tensile elongation and impact energy are other measures. Increasing one does not algebraically require an increase in the others. Use the actual grade, processing state, orientation, temperature and loading rate when comparing them.

The competition is between accommodation and crack advance

Plastic deformation in many metals depends on dislocations crossing obstacles. Thermal fluctuations assist some of those events. Cooling can reduce that assistance, so a greater applied stress is needed to sustain the same plastic strain rate. Lattice resistance is particularly consequential in many ferritic steels with a body-centred-cubic structure. The effect depends on the obstacles and available slip processes; it is not a rule that every metal’s strength changes by the same amount.

Plastic flow near a crack can blunt its sharp geometry and redistribute stress. If that accommodation becomes more difficult, high tensile stresses may develop before sufficient plastic deformation occurs to resist cleavage, in which a crack separates crystallographic planes. This provides a mechanism for a ductile-to-brittle transition in susceptible materials. It does not say that all cracking stops above a particular temperature or that yield strength alone predicts the transition.

The observed transition depends on how it is defined: a specified impact energy, fracture appearance or crack-resistance criterion need not give the same temperature. Notches, specimen thickness and multiaxial constraint change the local state; loading rate changes the time available for accommodation. Grain structure, impurities and processing also matter. A transition quoted without its material state and test definition is incomplete evidence.

Many face-centred-cubic metals can sustain plastic accommodation at low temperature without the pronounced cleavage transition characteristic of some ferritic steels. That is a useful comparison, not a certification from a crystal label. Certain austenitic alloys can transform to martensite during cooling or deformation, adding another source of strengthening and changed fracture response. Phase stability and processing must therefore accompany the structural description.

Use the mechanism appropriate to the material

In a polymer, cooperative chain-segment motion supplies modes of deformation and relaxation. When cooling makes their relaxation time τ large compared with the loading time t_load, the material responds more like a glass on that timescale. The ratio τ/t_load helps organise the comparison: a process with τ much smaller than the event time can relax during the event, whereas one with τ much larger cannot. A quicker event and a lower temperature can both suppress that accommodation.

The glass transition T_g describes a mobility change in the amorphous material, not melting and not one temperature-independent fracture criterion. Crystallinity, crosslinks, additives, secondary relaxations and the loading mode affect the result. A seal may become stiffer yet lose the conformability needed to keep contact. A polymer’s increased tensile strength would not, by itself, establish that sealing function.

Many structural ceramics and glasses already have limited plastic accommodation in ordinary tensile loading. Their cold response need not involve the same new ductile-to-brittle transition as a susceptible steel. Flaws, tensile stress, interfaces and thermal gradients can still control fracture. The precise ceramic composition and microstructure matter; neither the word ceramic nor its high-temperature stability establishes cryogenic crack resistance.

In a fibre composite, fibre direction, matrix response and interfaces introduce several coupled changes. Differently oriented plies and constituents need not have the same free thermal strain. Their constraint can generate local residual stress and matrix microcracking even under spatially uniform cooling. A NASA study of two carbon-fibre/epoxy tank laminates examined this mismatch and its relation to microcracking. Its lesson is that retaining fibre strength does not settle laminate integrity or permeability; it is not a universal failure prediction for all composites.

A measured counterexample to universal embrittlement

Gludovatz and colleagues tested a particular equiatomic CrCoNi alloy at 293, 198 and 77 K. For that processed material, the authors reported increasing strength, ductility and crack resistance as temperature fell. Microscopy found deformation-induced nanotwinning alongside dislocation plasticity. Twins are reoriented crystal regions; their formation can supply additional deformation and help sustain strain hardening, delaying localisation. The observations support a mechanism interpretation for this alloy and state.

The reported crack-initiation comparison uses K_JIc, a stress-intensity-equivalent value calculated from a J-based fracture assessment. J is an energy-based measure used to describe crack driving force under suitable elastic–plastic conditions. K_JIc has units MPa√m, but that unit does not make it an independently validated linear-elastic K_IC measurement. The paper also followed resistance with crack extension; initiation and later crack-growth resistance are different comparisons.

The rounded main-text initiation values are 208 MPa√m at 293 K and 273 MPa√m at 77 K. Treat these as the paper’s reported equivalents: its printed J, elastic-modulus and Poisson-ratio values do not reproduce these K_JIc values using its stated conversion. The percentage comparison below is arithmetic on the reported central equivalents, not an independently revalidated J-to-K conversion. Scatter, test conditions and the accompanying J-based and microstructural evidence still matter; another grade or component is a separate question.

The assembly also contracts

Different components contract by different amounts. Connections, seals and residual stresses can govern performance even if bulk strength improves. Use low-temperature data for the actual grade, orientation and relevant fracture or deformation mode.

Define the linear expansion coefficient α(T) = (1/L)(dL/dT) in the direction of interest. For small thermal strains, free strain over a temperature interval is ε_th ≈ ∫ from T_0 to T_1 α(T) dT. With a stated interval-mean coefficient ᾱ, use ε_th ≈ ᾱ(T_1 − T_0), and the free length change is ΔL ≈ L_0 ε_th. Positive ᾱ and cooling give negative strain. Actual cryogenic data can vary strongly across the interval; a room-temperature coefficient is not automatically the interval mean.

Free contraction is a kinematic preference, not a stress. An isolated uniformly cooled bar can shorten without mechanical stress in this simple model. If two parts cannot undergo their different free strains because a joint forces compatible displacement, elastic or inelastic strain must make up the difference. Joint stiffness, slip, geometry and external supports then determine the stress. A mismatch number alone is insufficient.

Uniform temperature does not remove mismatch between materials. Conversely, even a single material can develop thermal stress when its surface and interior have different temperatures during a transient. Thermal shock and expansion mismatch develops that spatial-gradient problem; the model below deliberately uses a uniform final temperature to isolate compatibility.

An explicit model turns mismatch into stress

Consider two initially stress-free straight bars A and B of equal length L_0 and equal cross-sectional area. They run in parallel between ideal rigid end blocks. Ideal guides keep the blocks parallel and prevent rotation while permitting axial translation. The guides can supply the required moment reactions but no external axial force. Both bars therefore have the same axial strain ε. Ignore bending, buckling, joint deformation and initial residual stress. Assume small-strain, uniaxial linear elasticity at the final uniform temperature, with moduli E_A and E_B.

Each bar’s total strain is its free thermal strain plus mechanical strain: ε = ε_th, A + σ_A/E_A = ε_th, B + σ_B/E_B. Take tension as positive. Force balance with equal areas and no external axial force gives σ_A + σ_B = 0. Substitution gives σ_A = [E_A E_B/(E_A + E_B)](ε_th, B − ε_th, A), with σ_B = −σ_A. The common strain is ε = (E_A ε_th, A + E_B ε_th, B)/(E_A + E_B).

These equations show both the stress sign and the role of stiffness. If A wants to contract more, ε_th, A is more negative. Compatibility holds A longer than its free cooled length, putting it in tension, while B is compressed relative to its own free cooled length. The common strain lies between the two free strains and is weighted towards the stiffer bar’s preference.

The figure uses the exact illustrative pair in the worked case: free shortenings of 0.36 and 0.16 mm over a 100 mm initial length. With E_A = 100 GPa and E_B = 200 GPa, the common shortening is about 0.227 mm. Its horizontal axis measures shortening only, rather than drawing the full bars to a misleading exaggerated scale. The arrows compare a constrained bar with its own free state; the plotted displacement is not a stress contour.

If the bars can slide independently, their equal-displacement equation is removed and this unloaded model gives zero mechanical stress in each. If the end blocks are fixed in space, force balance must instead include the external reactions. Yielding, cracking or joint slip would also change the stated elastic model. Those are different physical boundary conditions, rather than alternative answers to the same problem.

Separate material response from the measurement

A cryogenic comparison must distinguish actual specimen temperature, mechanical deformation and the response of the measuring system. For example, a bonded resistance strain gauge can change its electrical signal because of temperature as well as substrate strain; the gauge and specimen also have their own thermal response. An uncorrected signal cannot automatically be interpreted as stress-induced strain.

The same care applies to fracture and assembly evidence. A smooth tensile specimen, a notched impact specimen and a precracked fracture specimen sample different conditions and report different quantities. A warm test after exposure can reveal retained damage, but cooling-induced phase changes or the cold joint stresses may not persist unchanged after warming. Measure the response relevant to the claim and distinguish during-exposure behaviour from what remains afterwards.

For an unfamiliar cold component, first identify the possible failure process, then choose evidence that separates its causes: plastic accommodation, phase change, crack advance, local contraction constraint or loss of function. The alloy case supplies a real counterexample to a universal trend; the joined-bar case supplies a conditional stress calculation. Neither substitutes for evidence about the component’s actual material, joints and loading history.

Worked examples

Try the reasoning

1. A particular steel shows a higher cold yield stress but a lower notched impact energy than at room temperature. Explain a compatible mechanism and why these two results cannot by themselves give a linear-elastic fracture toughness.

   Show the answer and reasoning

   Cooling can make plastic accommodation more difficult while a notch develops high local tensile stresses, permitting less plastic energy absorption before fracture. Yield stress and impact energy are different quantities, and neither directly supplies K_IC.

   This is a plausible competition, not proof of the exact failure mechanism in that steel. Fracture appearance, material state, test definition and suitable crack-resistance evidence are needed. An impact-energy value in joules cannot be relabelled MPa√m.

2. A 200 mm free bar cools from 293 to 93 K. Its interval-mean α is 16 × 10^−6 K^−1 from 293 to 193 K and 8 × 10^−6 K^−1 from 193 to 93 K. Under the small-strain approximation, calculate its length change. Compare using 16 × 10^−6 K^−1 for the whole interval.

   Show the answer and reasoning

   The length change is approximately −0.48 mm. Using 16 × 10^−6 K^−1 throughout would give −0.64 mm, overestimating the shortening magnitude by 0.16 mm.

   Add the strain contributions: 16 × 10^−6(−100) + 8 × 10^−6(−100) = −0.0024. Multiply by 200 mm. Using the same initial length for both increments is the stated small-strain approximation; the point is to integrate temperature dependence rather than treat a room-temperature coefficient as constant.

3. The illustrative bars have the same thermal coefficients and cooling interval as the worked case, but the ends now slide independently with no axial load. A report assigns ±133 MPa because the free mismatch is still 0.20 mm. Explain the error and state the stresses in this model.

   Show the answer and reasoning

   The report retained a compatibility constraint that no longer exists. Each free bar can adopt its own thermal strain, so the model gives zero mechanical stress in each.

   Mismatch measures different preferred displacements; stress requires resistance to those displacements. Material coefficients alone do not determine the boundary conditions. Nonuniform cooling or friction would be additional physical inputs, excluded here.

4. Keep the joined-bar geometry, free thermal strains and all elastic-model assumptions, but double both moduli. Predict the common shortening and the stress changes before calculating them.

   Show the answer and reasoning

   The common shortening stays about 0.227 mm; the stress magnitudes double to about 267 MPa, with A tensile and B compressive.

   A common multiplier cancels in the stiffness-weighted strain. It multiplies E_A E_B/(E_A + E_B) by two, so the strain difference produces twice the stress. This remains a conditional elastic calculation, not evidence that the higher-stress state is physically sustainable.

5. A synthetic polymer has a relevant relaxation time τ = 0.01 s at one temperature and τ = 100 s at a lower temperature. Compare its ability to accommodate a one-second contact change. Why could its cold seal performance worsen even if a tensile test reports increased strength?

   Show the answer and reasoning

   The ratios τ/t_load are 0.01 and 100. The first process can substantially relax during the event; the cold process is too slow on that timescale. Loss of conformability can impair contact even with higher tensile strength.

   The numbers describe an illustrative relaxation mode, not a particular commercial seal or a complete fracture law. Real response also depends on other modes, geometry, contact and material state. The functional requirement must match the property evidence.

6. A composite joint has unchanged room-temperature strength after one cold exposure. A reviewer concludes that it had no cold microcracks, no contraction stress and no loss of sealing during exposure. Identify which parts of the conclusion are unsupported and what kinds of evidence would distinguish them.

   Show the answer and reasoning

   All three conclusions exceed that one warm strength measurement. Cold contraction/loading evidence addresses stress; suitable damage observations address microcracks; a functional response under the relevant cold condition addresses sealing.

   Some reversible stresses disappear on warming, while local damage can exist without a large change in the selected bulk strength test. Structural, mechanical and functional claims need evidence for their own response and time, rather than one result used as a universal proxy.

What to carry forward

• Cooling changes competing deformation and fracture processes; higher yield strength does not establish greater crack resistance.
• A material class or crystal structure helps identify mechanisms but does not determine every grade’s cryogenic response.
• Integrate thermal expansion over the temperature interval and distinguish free strain from constraint-induced stress.
• Compatibility and force balance determine the joined-bar result; changing supports changes the problem.
• Compare the actual fracture quantity and specimen state, and separate cold behaviour from what remains after warming.