A structure that carried its load through summer, autumn and most of the winter breaks at dawn on the coldest morning of the year. Nothing about the load changed, nothing was modified, and the material certificate is in order. What changed is the material's capacity to absorb energy before it separates. In body-centred-cubic metals, including most carbon and low-alloy structural steels, that capacity does not decline gently with temperature. It collapses across a narrow band, and a structure can pass through it overnight.
Toughness is a function of temperature
Strength and toughness are different quantities and they behave differently in the cold. Yield and tensile strength typically rise slightly as temperature falls, so a steel tested to ASTM E8 at low temperature looks, on paper, no worse than it did at room temperature.
Toughness, meaning the energy absorbed in propagating a crack, moves the other way and moves sharply. A member can be simultaneously stronger and far less forgiving of a flaw than it was the week before, which is the part that surprises people reading the certificate alone.
Which materials have a transition at all
The behaviour is characteristic of body-centred-cubic crystal structures. Ferritic and carbon steels have it. Austenitic stainless steels, aluminium alloys and most face-centred-cubic metals do not exhibit the same sharp transition and stay comparatively tough down to very low temperatures.
This matters when a substitution is at issue. Two materials can meet the same strength requirement and behave entirely differently in cold service, and a substitution justified on strength alone can quietly remove the toughness margin the original selection carried.
What a Charpy curve actually is
ASTM E23 governs the notched-bar impact test. Specimens of standard geometry are broken at a series of temperatures and the absorbed energy is plotted, producing a curve with a high upper shelf, a low lower shelf, and a transition region between them.
Percent shear on the broken specimen and lateral expansion are recorded alongside energy, because the fracture appearance changes across the same range from fibrous to flat and crystalline. The curve is a picture of the same behavioural shift a fracture surface displays.
The transition temperature is a defined number, not a physical one
There is no single temperature at which a steel becomes brittle. The transition is a region, and a transition temperature is the point where the curve crosses an agreed criterion: a specified absorbed energy, a specified percent shear, or a specified lateral expansion, as the governing code requires.
Different criteria applied to the same curve give different temperatures. When an opinion turns on whether service was above or below the transition, the criterion used is a question worth asking early, because it is where two competent analyses most often diverge.
What the test does not establish
A Charpy result is a comparative and qualification measure. Its notch geometry and impact rate are fixed by the standard and are not those of the component, so absorbed energy is not a design toughness value and cannot be used directly to calculate a critical flaw size.
Where a quantitative toughness is needed, ASTM E399 for plane-strain fracture toughness and ASTM E1820 for J-integral and CTOD supply it. Charpy is often the only data available because it is what the material specification required, which is a reason to use it carefully rather than a reason to overreach.
Section thickness moves the transition
A standard Charpy specimen is small. The constraint at a crack tip in a thick structural member is greater, which shifts the effective transition to a higher temperature than the small specimen suggests. Thicker material behaves as though it is colder than it is.
So a Charpy curve produced on coupon material is not automatically transferable to a heavy section, and testing that samples the actual thickness, orientation and location within the product form is worth insisting on.
Establishing what the temperature actually was
The metallurgical question meets a records question. Nearby weather station data, process temperature logs, control system historians and wind chill on an exposed member all bear on the metal temperature at the moment of fracture, which is rarely the same as the published daily low.
The fracture surface itself carries an indication. A predominantly cleavage surface is consistent with fracture at or below the transition; a fibrous, dimpled surface is consistent with the ductile regime. Read together with exemplar testing, that is a meaningful cross-check on the records.
Conditions that shift toughness over time
The transition is not fixed for the life of a part. Grain size, heat treatment and deoxidation practice set it initially, with finer grain structures generally giving lower transition temperatures. Cold working followed by ageing, certain thermal exposures, and grain-boundary embrittlement can all raise it in service.
Where an intergranular fracture appears rather than cleavage, that is a signal to examine grain-boundary condition and thermal history rather than to treat the failure as a straightforward temperature effect.
What the material record should contain
The useful documents are the design specification's service temperature, the material specification and whether it required impact testing at a stated temperature, the mill certificate showing any impact results, and any procurement substitution. Absence of an impact requirement in a cold-service specification is itself a finding.
Preserving the broken pieces separated and unhandled keeps the fracture-appearance evidence available, and retaining unfailed material from the same member gives the laboratory exemplar stock to test at temperature.
This article is general technical orientation, not a failure analysis, an engineering opinion, or advice on any specific matter. Determining the cause of a particular incident requires hands-on examination by a credentialed expert.