How to Cook an Egg That Cannot Be Cooked at One Temperature

A Boundary Condition Problem With Breakfast On It

Say "boundary condition" to most people and you get a blank look. Say it to the seven researchers who wrote this paper — a materials-and-chemistry group in Naples working on heat transfer, polymer processing and food chemistry — and you apparently get an egg. Emilia Di Lorenzo, Ernesto Di Maio and colleagues published "Periodic cooking of eggs" in Communications Engineering in February 2025, and the framing they chose is the interesting part: not a recipe, not a kitchen hack, but an engineering paper that happens to be about Sunday brunch [1].

Here is the problem as they state it. Egg white and egg yolk are different materials with different proteins, and they set solid at different temperatures. Cook the egg in one bath and you get to choose exactly one outcome: overcook one part, undercook the other, or land on some compromise that does neither well [1]. Every method anyone has used for centuries — boiling, poaching, coddling, sous-vide — is a single-temperature method, because a pot of water only has one temperature to give. The paper's actual contribution is not a new appliance. It is the observation that the constraint "one bath, one temperature" was never load-bearing, and that the fix is to stop holding still.

Two Proteins, Two Set Points

The numbers are almost insultingly simple once you have them. The paper states that albumen wants roughly 85°C to cook properly and yolk wants roughly 65°C [1]. Twenty degrees apart, inside one shell, cooked at the same time, by the same pot of water, for the entire history of egg cookery.

The reason sits in composition, not custom. Albumen is almost entirely a mixture of different proteins — ovalbumin, ovotransferrin, ovomucoid and others — each of which denatures and forms a solid network at its own characteristic temperature, and the paper is specific about the coldest of them: at low bath temperatures, "only one albumen protein is capable of doing so," and that protein is ovotransferrin [1]. Below whatever temperature that protein needs, the albumen never becomes anything but syrup, no matter how long you wait. Yolk is a different kind of material entirely, mostly lipoproteins and granules suspended in water, and it sets earlier and more gradually, moving from liquid through custardy to crumbly as its own temperature climbs. That an egg's various proteins each have their own characteristic set point, rather than one shared "doneness" temperature, is itself established enough to have its own dedicated food-biophysics literature [3].

Put those two set points on the same number line and there is no bath temperature that sits on both of them, because they are not the same number. That is the whole problem. It sounds almost too plain to be a real scientific puzzle, and that is exactly why nobody had solved it: it was never a puzzle about ignorance of the chemistry, since cooks have known "yolk wants cooler, white wants hotter" forever. It was a puzzle about what a boundary condition is allowed to do.

Why the Standard Fixes Fail

Two existing methods each solve half the problem, which in a two-variable problem is the same as solving none of it.

Boil it. A rolling boil holds the outside of the egg at 100°C, comfortably above the albumen's 85°C requirement, so the white sets hard and fast. Good news for the white, bad news for the yolk: heat keeps conducting inward for as long as the egg sits in the pot, and a yolk that spends twelve minutes anywhere near 100°C sails straight past 65°C into the chalky, crumbling texture of a classic hard-boiled yolk. The paper's own cooking-time comparison lists conventional hard-boiling at 12 minutes [1].

Sous-vide it. Hold the whole egg in a water bath at exactly 65°C for as long as you like, and the yolk comes out precisely as intended, custardy and never overshooting, because it can never see a hotter environment than the one number you chose. The albumen sits in the same bath and never reaches the temperature its stiffer proteins need, so it never fully sets. It thickens, but it stays translucent and runny at 65°C even after the paper's comparison time of 60 minutes [1] — a full hour, and still not there.

Both failures are the same failure from opposite directions: whichever set point you pick as your one bath temperature, the other ingredient inherits it whether it wants it or not. The literature on sous-vide cooking exists specifically because holding a bath dead level at one chosen temperature is, for a single-protein target, a genuinely good idea [2]; the egg is just the case where you have two targets and one bath.

The Trick: Make the Boundary Condition Move

The paper's actual move is to stop treating "bath temperature" as a constant and start treating it as a function of time. In heat-transfer language, the temperature imposed at the outer surface of the object is the boundary condition, and every classical cooking method uses a boundary condition that is fixed: T(surface) = constant. Di Lorenzo and colleagues instead impose a periodic boundary condition, a temperature that cycles between two values on a fixed schedule, and they solve the resulting heat-conduction problem to see what that does to the inside of the egg [1].

Their protocol, stated plainly: submerge the egg in a 100°C bath for 2 minutes, transfer it to a 30°C bath for 2 minutes, and repeat that pair eight times, for a total cooking time of 32 minutes [1]. Nothing about the egg itself changes. No new equipment, no chemistry added to the water. The only thing that moved is the one input every previous method had nailed down: the temperature the outside of the egg is allowed to sit at.

Why would alternating between two wrong temperatures produce a result that is right for both parts of the egg? The short version, which the rest of this piece works through with the paper's own numbers: the two halves of the egg do not experience the same boundary condition equally. One of them is thin and sits right against the water. The other is buried, and being buried changes everything.

Building an Egg Out of Equations

Before cooking a single real egg, the group built a mathematical one. They modelled the egg's shape as an ellipsoid with axes of 7 cm and 5 cm, with a sphere of radius 1.5 cm sitting at the geometric centre standing in for the yolk, everything starting at a uniform 20°C [1]. Heat moves through both regions by conduction alone, governed by the standard heat-diffusion equation, with a thermal diffusivity for both yolk and albumen on the order of 10^-5 m² s^-1 [1] — unsurprising, since both are mostly water, and that number is close to water's own diffusivity.

Conduction alone tells you the temperature at every point in the egg over time. It does not tell you whether the egg is cooked, because cooking is a chemical event, not a thermal one: proteins denature and cross-link into a gel once enough of them have individually unfolded, and that unfolding is a temperature-activated reaction like any other. The paper couples the heat equation to a second equation, a cooking-degree variable X that runs from 0 (raw) to 1 (fully set), governed by an Arrhenius rate law:

for each region i (yolk or albumen), with its own activation energy E_a,i and pre-exponential factor A_i. The whole coupled system, temperature field plus two cooking-degree fields, was solved numerically by finite elements, using a direct UMFPACK solver, stepping forward in time every 0.5 minutes [1]. That is the machine the paper actually built. Everything downstream, the 67°C yolk and the swinging albumen, is what came out of running it.

What the Model Says Happens Inside the Egg

Run the periodic protocol through that model and the paper reports a genuinely elegant split. The yolk, protected by roughly a centimetre or two of albumen in every direction, "sees a constant temperature of 67°C." The albumen, sitting right at the boundary, does not get that luxury: during each hot phase it ranges over roughly 100 to 87°C, and during each cold phase it ranges over roughly 30 to 55°C [1].

Read those two results side by side and the mechanism becomes visible without any further mathematics. The albumen's whole existence, for 32 minutes, is a wide swing that spends real time above its own 85°C threshold on every single hot pulse. The yolk, meanwhile, never really leaves a narrow band centred a couple of degrees above its own 65°C target. Same water, same clock, two completely different thermal experiences, because one material is thin and exposed and the other is thick and buried.

Club Arithmetic, Part One: Why 67 and Not 65

Club Arithmetic, Part Two: How Deep the Thermal Wave Reaches

Club Arithmetic, Part Three: The Clock, Compared

Putting Real Eggs on the Bench

None of the arithmetic above would matter if real eggs disagreed with the model, so the paper's second half is entirely experimental: real eggs, four methods, real instruments. They cooked eggs hard-boiled, soft-boiled, sous-vide, and periodic, then ran texture profile analysis (TPA) on both albumen and yolk from each [1].

The hardness values, reported in the paper's Table 2 as the mean of four replicates, tell a clean story:

Table 1. Instrumental texture measurements, paper's own units [1]. On the yolk side, periodic (4.34) sits closer to sous-vide's soft 1.59 than to hard-boiled's crumbling 22.13 — about 5.1 times softer than hard-boiled and 2.7 times firmer than sous-vide. On the albumen side, periodic (4.73) is firmer than sous-vide's barely-set 1.33 by a factor of about 3.6, while staying about 3.0 times softer than hard-boiled's rubbery 14.37.

The sensory panel's own language matches the numbers: periodic cooking produced "a fully set albumen similar to a soft boiled egg but with a creamy yolk similar to that of an egg cooked sous vide" [1]. It is worth being honest about what "fully set" means numerically here — the periodic albumen's hardness score (4.73) is still noticeably below soft-boiled's (9.46), so "similar to soft-boiled" is a sensory judgement about the albumen having crossed from liquid to solid, not a claim that the two are texturally identical.

What Changed in the Proteins

Texture is what you feel; FT-IR spectroscopy is a way of asking the proteins directly what shape they are in. The paper tracks the amide-I absorption band, and specifically the component at 1636 cm^-1, which corresponds to protein aggregation into β-sheet structure — the molecular signature of denaturation setting into a permanent network rather than a reversible unfolding [1]. Across the four methods, periodic cooking produced an intermediate signal, more aggregation than the barely-touched sous-vide proteins and less than the heavily denatured hard-boiled proteins [1].

That is a second, independent line of evidence pointing at the same conclusion as the texture data: periodic cooking is not secretly identical to one of the conventional methods with a different name. It produces a genuinely intermediate protein state, consistent with a genuinely intermediate thermal history — exactly what you would expect from an albumen that got real time above 85°C, repeatedly, but never for as long in one stretch as a 12-minute hard-boil delivers.

A Yolk That Keeps More of Its Polyphenols

The most unexpected result in the paper has nothing to do with texture. The authors screened the yolk for polyphenols and related antioxidant compounds using UHPLC coupled to a high-resolution Orbitrap mass spectrometer, comparing the periodic yolk against yolks from the other three methods [1]. The finding, in their own words: "the periodic yolk group is characterized by higher content of all the investigated polyphenols" relative to the other cooking methods, with named compounds including ferulic acid, chlorogenic acid and daidzein [1].

Two honesty notes belong here, because it would be easy to overstate this result. First, the method the paper actually used is UHPLC-Orbitrap high-resolution mass spectrometry, not classic HPLC with a simpler detector; it is a liquid-chromatography technique in the same family, but a more sensitive and more specific one, and it is worth naming correctly rather than rounding it down to "HPLC." Second, the main text of the paper reports this result qualitatively — "higher content of all the investigated polyphenols" — without publishing the underlying concentration numbers in the body of the article; those details are deferred to the paper's Supplementary Information, which was outside what we could verify directly. We can report that the direction of the effect is real and stated plainly by the authors. We cannot, honestly, hand you a milligram figure for it.

Why periodic cooking would retain more of these compounds than hard-boiling, soft-boiling or sous-vide is left open by the paper, but a plausible physical reading fits with everything above: polyphenols and similar antioxidant compounds can degrade with cumulative thermal exposure, and the periodic yolk's own thermal history — sitting almost the entire time within a few degrees of 67°C rather than briefly approaching 100°C the way a hard-boiled yolk's outer layers do — is the gentlest of the four histories on that specific measure, even though its total elapsed time (32 minutes) is longer than two of the three comparison methods.

What This Is Actually Good For

It would be easy to read this paper as a party trick for a science-adjacent kitchen and stop there. The authors themselves aim higher, closing on the note that the method "holds promise for innovative culinary applications and materials treatment" [1] — and that second phrase is the one worth sitting with. An egg is, structurally, nothing special: it is a composite object made of two materials with different thermal set points, wrapped in a shell that limits how you can access either one individually. That description covers a great many manufactured composites, not just breakfast.

The generalizable idea is not "cycle temperature," which cooks and metallurgists have both done for other reasons for a long time. It is the specific insight this paper demonstrates with numbers: a periodic boundary condition can be tuned, using nothing more exotic than the target object's own geometry and diffusivity, to deliver two different effective thermal histories to two different depths inside the same object, from a single external process. The egg is the proof of concept because its two answers happen to be checkable by taste. The interesting question the paper leaves open is which other twenty-degree conflicts, in materials nobody eats, might yield to the same trick.