The first time I grilled a whole chicken for friends, I spatchcocked it, set the gas grill to maximum, went inside to make a salad, and returned to find that the outside had literally become charcoal while the inside was still raw. Not metaphorical charcoal. Actual charcoal: the organic matter in the skin had pyrolyzed in a low-oxygen contact zone against the grate, converting proteins and fats into elemental carbon. I had inadvertently replicated the ancient process of carbonization, which, in retrospect, suggested a fuel change.

It turned out that the difference between juicy and dry is six degrees Celsius, that salt diffusion obeys Fick’s second law, and that most grilling advice treats these facts as optional. None of the sources I found agreed with each other, none computed anything for the cut actually on my counter, and every one of them assumed without saying so that I could intuit the thermal diffusivity of a bone-in pork chop.

I could not. So I worked out the thermodynamics, and then I built a calculator to do it for me.

What Actually Matters

What matters in grilled meat is, in order: that the interior is cooked to the correct temperature and not one degree beyond it, that the exterior has a deep Maillard crust, that the seasoning has penetrated the full thickness, and that the process is repeatable. Everything else is lore.

Interior temperature. The proteins in meat denature in a specific sequence as temperature rises: myosin first (50°C), then collagen (55–76°C), then actin (66–73°C). The window between myosin denaturation (pleasant firmness, developed flavor) and actin denaturation (tough, dry, juice squeezed out) is where all the quality lives. Sous vide lets you park the interior at a precise point in this window and hold it there.

Crust. The Maillard reaction requires surface temperatures above 140°C (280°F) and proceeds optimally between 150°C and 180°C. Above approximately 180°C, thermal decomposition (charring) competes with Maillard browning, producing bitter compounds (Mottram, 1998). The goal is to flash the surface through the Maillard zone as fast as possible, spending minimal time in the zone where heat conducts inward and overcooks the interior. This requires extreme surface heat, applied briefly.

Seasoning depth. Salt diffusion into meat follows Fick’s second law. Penetration depth is proportional to the square root of time, not linear: doubling the brining time does not double the penetration depth. At refrigerator temperature, salt moves approximately 5–7 mm in 12 hours. A 25 mm steak (1 inch) salted overnight is seasoned most of the way through. The same steak salted 10 minutes before cooking has a salty surface and a bland center.

Repeatability. Traditional grilling is an open-loop system: you guess at the fire temperature, guess at doneness by feel, and pull the meat when it seems right. Sous vide converts this to a closed-loop system. The water bath temperature IS the final interior temperature. The only variable left to control is the sear, and the calculator below tells you how long that should take.

Background

I developed and kitchen-tested the recipe above. AI helped with the research below.

Grilling: older than language

Control of fire by Homo erectus dates to at least 1 million years ago, based on evidence from Wonderwerk Cave in South Africa (Berna et al., 2012). The earliest unambiguous evidence of deliberate cooking (as opposed to scavenging burned carcasses) comes from Qesem Cave, Israel, dated to 300,000–400,000 years ago, where cut marks on animal bones are consistent with butchering cooked rather than raw meat (Stiner et al., 2009). Richard Wrangham’s “cooking hypothesis” argues that the caloric surplus from cooked meat drove the brain expansion in Homo, making cooking the defining technology of human evolution (Wrangham, 2009).

For the vast majority of this history, the technique was direct radiant heat from an open fire; the metal grate came much later. The word “grill” descends from the Latin craticula (small griddle), through the French griller (to broil). Meat over coals on a grate is ancient and global, and it has stubbornly survived as a living tradition: the Argentine parrilla, the gaucho live-fire asado of the South American pampas, is documented to the early 19th century (Gaucho, Wikipedia), and the Japanese shichirin, a portable charcoal grill, has been in everyday use since the Edo period, 1603–1868 (Shichirin, Wikipedia).

What George Stephen actually contributed in 1952 was narrower and more useful than the lore suggests. He cut a marine buoy in half at the Weber Brothers Metal Works in Mount Prospect, Illinois, and bolted on a domed lid with dampers top and bottom, creating the Weber Kettle (Weber, 2022). The lid is the invention. People had been grilling over coals for a hundred thousand years; what Stephen added was the ability to put a roof on the fire and throttle it, turning an open flame into an oven you can argue with.

The sous vide revolution

Sous vide cooking was developed independently by two French researchers in the early 1970s. Georges Pralus, working for the Restaurant Troisgros in Roanne, discovered in 1974 that foie gras cooked in sealed plastic at low temperature retained its original weight and improved in texture (Baldwin, 2012). Bruno Goussault, a food scientist at the Centre de Recherche et d’Études pour l’Alimentation, had been studying the effects of temperature on food since the late 1960s and published the first systematic research on time-temperature relationships in sealed-bag cooking.

For decades, sous vide remained a professional technique because immersion circulators cost thousands of dollars. The home sous vide revolution began around 2009 when Kenji López-Alt published the “beer cooler method” (a polystyrene cooler filled with hot water as a passive thermal bath) and accelerated in 2013 when the Anova Precision Cooker brought accurate immersion circulators below $200 (López-Alt, 2015). For more on the physics of immersion circulators, see our investigation of what your sous vide is actually doing.

Sous vide means cooking in a plastic bag in hot water for hours, and the bag is not as inert as its intact look suggests; the full account of what leaches, what the studies actually measured, and which bag to use is in the sous vide investigation.

Methodology

I. The protein denaturation cascade

Meat is approximately 75% water, 20% protein, and 5% fat (varying by cut; lean beef cuts run closer to 65–70% water and 21–23% protein) (Mortensen et al., 2024). The proteins that determine texture and juiciness denature in a specific sequence as internal temperature rises. Everything else in this article is downstream of that sequence.

Myosin: 40–50°C (104–122°F). Myosin is the dominant contractile protein in muscle fibers. When it denatures, sarcomeres shrink in diameter, the meat firms pleasantly, and the texture transitions from raw to cooked. Myoglobin (the pigment protein) denatures in the same range, driving the color change from red to pink. By 54°C (130°F), myosin denaturation is substantially complete (Tornberg, 2005). This is the “good” denaturation: it produces the texture people associate with a properly cooked steak.

Collagen: 55–76°C (131–170°F). Collagen is the connective tissue protein that forms sheaths around muscle fibers and bundles. Its behavior is time-dependent: at low temperatures (55°C), it begins to shrink and very slowly converts to gelatin. At high temperatures (above 70°C), the conversion is rapid. Endogenous proteases (cathepsins B and L), which actively break down collagen, remain active below 60°C (Aktari et al., 2020). This is why sous vide can tenderize tough cuts at low temperatures given enough time: 24–72 hours at 55°C achieves collagen conversion that would normally require 80°C+ for hours.

Actin: 66–73°C (150–163°F). Actin is the protein to avoid denaturing. When actin denatures, muscle fibers become very firm, shorten in length, and the liquid expelled from the contracting fibers increases dramatically; calorimetry ties the sharpest drop in juiciness to this transition (Martens et al., 1982). Moisture loss roughly doubles as the interior climbs from 49°C to 66°C, the point at which actin begins to denature (ThermoWorks, 2024). This is the cliff: meat cooked above 66°C is measurably tougher and drier than meat cooked below it.

The optimal window for tender cuts (steak, lamb chop, pork chop, chicken breast) is between full myosin denaturation and the onset of actin denaturation: 54–65°C. In this range, the meat has developed flavor and pleasant firmness (myosin is done) while retaining maximum moisture (actin is intact). Food scientists call this the “optimal texture zone” (Tornberg, 2005).

II. Dark vs. light: why one temperature does not fit all chicken

The USDA recommends cooking all chicken to 165°F (74°C). This is a temperature at which Salmonella is killed instantly, requiring zero hold time. It is the safest possible recommendation for a consumer who owns no thermometer and cannot be trusted to hold a temperature for a specified duration. It is also a temperature that renders chicken breast dry and chalky, because 74°C is well past actin denaturation.

Breast and thigh are two different meats with opposite relationships to temperature, and cooking them as if they were the same is the original sin of roast chicken.

Chicken breast is white muscle (Type IIb fast-twitch fibers), minimally exercised, with very little connective tissue. It contains approximately 15.9 mg/g collagen on a dry weight basis, roughly 1.6 g per 100g of raw meat (Kim et al., 2023). There is minimal collagen to render, so there is no benefit to temperatures above 65°C. Above that point, actin denatures and the lean meat (only 1.9% fat, per USDA) dries out rapidly. The optimal temperature is 65°C (149°F), held long enough for pasteurization (about 5 minutes at core temperature per USDA Appendix A).

Chicken thigh is dark muscle (Type I slow-twitch fibers), heavily exercised, rich in connective tissue. It contains approximately 30.4 mg/g collagen on a dry weight basis, roughly 3.0 g per 100g of raw meat, about twice that of breast (Kim et al., 2023). This collagen must convert to gelatin for the thigh to become tender. Conversion begins at 65°C but proceeds much faster at 74°C+. The gelatin that forms compensates for moisture loss from actin denaturation, which is why thighs actually become more tender as temperature increases, up to about 90°C. America’s Test Kitchen found the optimal serving temperature for thighs to be 79–90°C (175–195°F) (America’s Test Kitchen, 2023).

Cooking breast and thigh to the same temperature is a compromise that produces optimal results for neither.

III. Fick’s law and salt diffusion

Dry brining (salting meat in advance) follows a three-phase process. In the first minutes, salt crystals dissolve in surface moisture. Within 3–10 minutes, the concentrated surface brine draws additional water from the muscle via osmosis, and the surface becomes visibly wet. After 30–40 minutes, the dissolved salt begins diffusing back into the meat along with the exuded liquid, carrying sodium and chloride ions deep into the muscle (López-Alt, 2015).

The diffusion of salt into meat is governed by Fick’s second law:

\frac{\partial C}{\partial t} = D_{\text{eff}} \frac{\partial^2 C}{\partial x^2}

where C is NaCl concentration at depth x and time t, and D_{\text{eff}} is the effective diffusion coefficient. Published values for D_{\text{eff}} in refrigerated meat range from 2.2 × 10⁻¹⁰ to 1.3 × 10⁻⁹ m²/s depending on muscle type and temperature (PMC 10094276, 2023).

The characteristic penetration depth scales with the square root of time:

\delta \sim \sqrt{D_{\text{eff}} \cdot t}

Using D_{\text{eff}} \approx 5 \times 10^{-10} m²/s at 4°C:

Time	Penetration depth	Implications
1 hour	~1.3 mm	Surface only
4 hours	~2.7 mm	Outer ring seasoned
12 hours	~4.6 mm	Nearly halfway through a 25 mm steak
24 hours	~6.6 mm	More than halfway; optimal for most cuts

The Fickian model is a conservative lower bound. Real penetration may be somewhat faster due to capillary transport along cut muscle fibers and microstructural channels in the meat. Using the conservative model for brining recommendations is the safer approach: slightly over-brining produces a uniformly seasoned result, while under-brining leaves a bland center.

The calculator uses the Fickian model (penetration depth = 2\sqrt{D_{\text{eff}} \cdot t}) to compute brine times. For a 25 mm steak, this gives approximately 22 hours for salt to reach the center, which is longer than some empirical rules of thumb suggest. The discrepancy reflects the difference between a thermodynamic lower bound and real meat with its network of connective tissue channels.

The practical consequence: salting 10–30 minutes before cooking is the worst possible timing. At that point, salt has drawn moisture to the surface (creating a wet surface that inhibits browning) but has not had time to be reabsorbed. The surface is wet, the outer fibers are dehydrated, and the interior is unseasoned. Either salt immediately before cooking (no time for osmosis to start) or salt at least 40 minutes ahead, preferably 12–24 hours (López-Alt, 2015); the larger and denser the cut, the earlier it should be salted (Nosrat, 2017).

III.b. Why salting frozen meat works: thermal vs. mass diffusion

Heat and salt both diffuse into meat, but at dramatically different speeds. Thermal diffusivity in unfrozen meat is approximately 1.3 × 10⁻⁷ m²/s (Baldwin, 2012); the calculator derives this value for each cut from its composition using the standard component model (Choi and Okos, 1986). Salt diffusivity is approximately 5 × 10⁻¹⁰ m²/s (PMC 10094276, 2023). Heat moves through meat roughly 260 times faster than salt.

In frozen meat, the gap is even wider. Ice has a thermal diffusivity of approximately 1.1 × 10⁻⁶ m²/s, about 8 times faster than unfrozen meat and roughly 2,000 times faster than salt diffusion. This means a thawing front moves through frozen meat far faster than a salt front moves through thawed meat.

So the trick works like this: when you salt a frozen steak and place it in the refrigerator, the outer surface thaws within the first hour or two. Salt immediately begins diffusing into this thawed layer. Meanwhile, the thawing front continues advancing toward the center, progressively exposing more tissue to the salt that is already on the surface. By the time the center thaws (which takes longer due to the latent heat of fusion absorbing energy at 0°C), the outer several millimeters have been brining for most of the defrost duration.

For a 32 mm steak, defrost takes approximately 10 hours and salt penetration to the center approximately 36 hours (the conservative 2√Dt brine model above). Defrost first and then salt, and the two run back to back, 46 hours; salt while frozen and the defrost hides inside the brine time, 36 hours total, saving 10. The savings shrink with the cut, since brine time falls with the square of thickness: a 13 mm steak defrosts in about 4 hours and takes the salt in about 6, so salting it frozen finishes in 6 hours instead of 10.

Salt also restructures the meat’s protein matrix. Chloride ions bind to muscle filaments, increasing electrostatic repulsion between myosin and actin filaments, causing the lattice to expand and absorb additional water. Harold McGee notes that a 3% salt solution dissolves parts of the protein structure that supports the contracting filaments, preventing the dense aggregation that squeezes water out during cooking (McGee, 2004). Cook’s Illustrated found that brined chicken lost only 7% of its weight during cooking, compared to 18% for unbrined chicken.

That same dissolution is the leading edge of curing. The firm, springy bite of ham or sausage is the next stage of the same chemistry: once the salt inside the meat reaches roughly 2–3% (an ionic strength near 0.5 M), the salt-soluble myofibrillar proteins go into solution in quantity and then set into a dense gel on cooking. Protein extractability and the swelling that comes with it both climb with salt concentration up to that point (Richardson & Jones, 1987).

The effect depends only on ionic strength, not on whether the salt arrived dry or in a brine; the two delivery methods differ in their ceiling. Dry salt sits on the surface and is never rinsed away, so the outer layers have no upper bound and drift into the curing range given enough days, which is why I time a dry brine in hours to a day or two, not weeks. An immersion brine instead drives the meat toward equilibrium with the bath, and the bath concentration is a hard ceiling: a dilute brine that keeps the meat below the 2–3% threshold cannot cure it however long it soaks, while a brine strong enough to push the meat into that range cures the outer layers in time.

Raise the concentration further and the proteins stop dissolving and aggregate instead. That is the over-salted, mushy failure mode; in cod, heavy salt uptake drops the myosin heavy chains out of the salt-soluble extract through that aggregation (Martínez-Alvarez & Gómez-Guillén, 2006).

One difference favors the brine. Immersion adds water rather than drawing it off, with uptake scaling to brine strength (Aliño et al., 2010), so a wet-brined cut sitting at curing-range salt reads juicier and more swollen than true cured meat, which is comparatively dry and firm. There is no nitrite anywhere in this, so it is a texture change only, not preservation and not the pink of cured ham.

Introducing the cling brine

So much for dry brine versus wet. The method I actually use splits off from the dry brine, and I spent years salting straight past the gap that makes room for it. A dry brine is doing two unrelated jobs at once. One is seasoning: the salt diffuses inward and swells the lattice so the meat holds its juice, which is Fick’s law and nothing else, indifferent to whether the surface above it is wet or bone dry. The other is the thing people actually mean by “leave it uncovered in the fridge”: the exposed face air-dries, its water activity falls, and only a dry face will take a Maillard crust (Section V). The first job does not need air. The second is the only thing the open fridge buys, and on a steak it buys it a full day too early. So I split the two apart, and the split needed a label, so it got one: the cling brine.

Salt the cut on every face, wrap it tight in cling film, and leave it. Sealed against the meat, the salt walks in at exactly the rate it would in open air, because the diffusion front is buried millimetres deep and could not care less what the surface is doing; the cut comes up seasoned to the bone with not one outer layer leathered. Then, for the last hour before it cooks, I unwrap it and pat it dry, handing the surface-drying to a step that is actually sized for it. Beside the two brines above, the cling brine is no exotic third thing: it is a dry brine, with the deeper seasoning, the absence of any added surface water, and no dilution ceiling, that simply postpones the drying to where it earns its keep.

On a steak or a chop that place is the sear, and the timing is the whole argument. A roast bird air-dries for a day because its crust builds slowly over a long cook; a steak gets its crust from a forty-five-second blast on a screaming surface, so the only water that matters is the water sitting there at the instant the heat lands. A wet face fights that sear head-on, pinned at 100°C boiling itself off while the Maillard reaction stands idle (Section V). Leaving a thin cut uncovered for hours, meanwhile, leathers an uneven rim and gives back nothing the final unwrapped hour and a hard pat-dry would not.

So the cling brine puts the dryness exactly where it pays: a face that flashes to deep brown in seconds, edge-to-edge pink underneath because the heat never lingered, and the whole slab seasoned through instead of salty on top and bland in the middle. I had been air-drying steaks overnight on borrowed roast-chicken habit, parching the one layer I most wanted juicy a full day before it ever met the fire.

There is a small bonus to the wrap I would not have guessed. A sealed, near-anaerobic surface keeps better in the fridge than an exposed one: aerobic Pseudomonas, the usual slime-former on chilled beef, needs oxygen, so wrapping nudges the surface flora toward the gentler lactic bacteria and slows spoilage (Hernández-Macedo et al., 2011). A day in the wrap leaves the steak fresher than a day in the open air, which is a strange thing to be able to say about cling film.

IV. Time-temperature pasteurization

Pasteurization is not a single temperature. It is a function of both time and temperature, governed by the D-value (time to kill 90% of a pathogen at a given temperature) and the z-value (temperature increase needed to reduce the D-value tenfold).

For Salmonella in beef and pork, the z-value is approximately 10.8°F (6.0°C). The USDA FSIS Appendix A provides the following hold times for a 6.5-log reduction (required at the coldest point in the meat, after it has reached the target temperature):

Temperature		Hold time
130°F	54°C	112 minutes
135°F	57°C	36 minutes
140°F	60°C	12 minutes
145°F	63°C	4 minutes
150°F	66°C	67 seconds
155°F	68°C	15 seconds
160°F	71°C	Instant

Source: USDA FSIS Compliance Guideline, Appendix A (USDA, 2021).

The relationship is exponential, but a single log-linear approximation can underestimate hold times by 10% or more at the low end of the range. The calculator uses the USDA Appendix A tabulated values directly, with linear interpolation between entries, rather than a curve fit. This avoids underestimating pasteurization times where accuracy matters most (low temperatures, long holds).

This means a medium-rare steak at 130°F is perfectly safe if the center is held at that temperature for 112 minutes. In a sous vide bath, the center reaches 130°F and then stays there indefinitely. A 1-inch (25 mm) steak at 130°F needs approximately 75 minutes to heat through (from Baldwin’s slab model) plus the pasteurization hold: roughly 3 hours total. Most home cooks do not bother with full pasteurization for intact-muscle steak (the interior of an intact cut is sterile; pathogens live on the surface, which is killed during searing). The minimum cook time for a properly heated interior is the heating time alone.

For poultry, the USDA requires a 7-log Salmonella reduction (stricter than the 6.5-log for beef). At 149°F (65°C), the published hold time is roughly 5 minutes at core temperature (USDA FSIS, 2021); at 165°F it is instant. Held to that standard, chicken at 149°F is equally safe and dramatically more juicy than chicken flash-cooked to 165°F (Baldwin, 2012).

V. Searing thermodynamics: why hotter means less overcooking

The central paradox of grilling: a hotter fire produces a less overcooked interior. This is counterintuitive but follows directly from the physics of heat transfer.

A sear achieves two things simultaneously: it deposits Maillard-reaction products on the surface (desirable), and it conducts heat into the interior (undesirable). Both processes operate on the same timescale. The question is how to maximize the first while minimizing the second.

The Maillard reaction requires a surface temperature of 140°C+ and an adequate crust forms in a roughly fixed amount of energy input per unit area. The time to deliver that energy depends on the heat flux at the surface. On a charcoal grill, the dominant heat transfer mechanism is thermal radiation from the coals, governed by the Stefan-Boltzmann law:

q = \varepsilon \sigma T^4

where \varepsilon is emissivity (~0.95 for charcoal, nearly a perfect blackbody), \sigma is the Stefan-Boltzmann constant (5.67 × 10⁻⁸ W/m²·K⁴), and T is the absolute temperature of the coal surface. Radiant power scales with T^4. Going from 600°F (589 K) coal surface to 1,200°F (922 K) increases radiant output by a factor of (922/589)^4 \approx 6\times.

The radiant intensity at the food surface also obeys the inverse square law: I \propto 1/r^2. Halving the distance between coals and grate quadruples the intensity. The combination of hotter coals and closer proximity can increase effective searing speed by an order of magnitude.

At 500°F grate temperature, developing a proper crust takes 90–120 seconds per side. During those 90 seconds, heat conducts steadily into the interior, overcooking a layer 3–5 mm thick (the “grey band”). At 800°F+ (chimney searing), the same crust forms in 30–45 seconds, and the grey band is under 1 mm (Stefan Gourmet, 2020).

A dry surface, and why not a starch coating. Browning is rate-limited by surface temperature, and a wet surface cannot climb past 100°C until the water has boiled off; every second spent evaporating is a second the Maillard reaction, which needs 140°C+, does not run. Patting dry and the cold-water pre-chill, which draws moisture out of the outer few millimetres, are what make a fast crust possible. A dusting of cornstarch is sometimes recommended to accelerate this, and on a moderate skillet it works: the starch absorbs surface moisture and browns into a thin crisp shell. But starch browns (dextrinizes) only between roughly 150 and 200°C (ScienceDirect, 2024); above that window it pyrolyzes to bitter char. The sears here run far hotter, from 280°C for cast iron to 500°C for a torch, so a starch film blows straight through its browning window and scorches black before the meat’s own crust forms. With a properly dried surface it is unnecessary, and at full sear heat it is counterproductive, so this recipe omits it.

V-A. The pre-sear chill: water, not the fridge, and only for sous vide

The pre-chill earns its place because the sear’s damage is set by how much thermal headroom the surface has when the heat lands. A sous vide steak is the worst case: the whole slab sits at the target temperature, so every degree the sear adds to the outer layer is a degree of overcooking. Cooling the outer few millimetres before searing buys that layer room to absorb the sear without crossing into grey. The only question is how to cool it, and the answer is not the refrigerator.

Cooling is convection-limited at the surface, and water is a far better convector than air. The model above, the same slab conduction the cook times come from, puts the cold-tap-water heat-transfer coefficient near 55 W/m²·K against roughly 8 for still refrigerator air, about a sevenfold difference. The consequence is stark: a cold-water dunk pulls the surface from 54°C to about 32°C in eight minutes, while five minutes in the fridge removes only about 6°C, which is to say almost nothing. To match the dunk in the refrigerator you would wait roughly twenty minutes, and the bag dunk is also the version that draws moisture out for a faster crust. This is why the recipe dunks in cold water rather than telling you to rest the bag in the fridge.

Two limits bound the trick. Chilling harder is not strictly better: push it far enough (around forty-five minutes of refrigeration in the model) and the cold reaches the center, so the steak finishes below its target and you have traded a thin grey band for an under-temperature middle. The calculator makes that limit its rule: it gives a cut the chill only when the cold can cool the surface without reaching the center. A thick cut, whose center the cold front never reaches in the time available, gets the chill; a thin cut, whose center it reaches almost at once, gets none, because a sous vide center is cooked to its final temperature and a chill that drops it is pure loss. The thin cut instead has its sear held to a brief flash: with no chilled surface layer to absorb the heat, a full sear would drive its locked-in center past the target, and a thin cut simply cannot take both a hard crust and a precise center (for a hard crust on a thin cut, the reverse-sear path, which pulls early, is the way). And the chill is not the largest lever in the first place; a hotter, faster sear is, because it deposits a thinner pulse of heat (Section V). The pre-chill is the cheap second move after a screaming fire, not a substitute for one.

Why this is a sous vide step and not a grilling step. It is tempting to bag and dunk a reverse-seared steak too, but the model says it backfires. The reverse-sear path pulls the meat off the heat below its target on purpose and relies on carryover to finish it; a cold-water dunk drains the very stored heat that carryover needs. In the simulation, dunking a steak pulled at 48°C shaves only about a millimetre off the grey band while dropping the finished center from 56°C to 51°C, undercooked. Sous vide benefits from the dunk precisely because it has no carryover to spend: the meat is held at target, so cooling the surface is pure gain. So the cold-water chill belongs to the sous vide path alone, which is where the recipe puts it.

VI. The charcoal fire

Lump vs. briquettes. Lump charcoal is pure hardwood carbonized in a low-oxygen environment. Briquettes contain roughly 70% charcoal plus starch binders, limestone (for the white ash consumers expect), borax, and sodium nitrate accelerant (The Charcoal Factory, 2024). Lump burns hotter (surface temperatures up to 760°C / 1,400°F vs. 430–540°C / 800–1,000°F for briquettes) because it is purer fuel with no temperature-moderating fillers. For searing, lump is the correct choice.

Coal color and temperature. Glowing coals are approximate blackbody radiators. Their color follows Wien’s displacement law: faint red at ~630°C, cherry red at ~730°C, orange at ~900°C, yellow-orange at ~1,100°C. The common advice to “wait until coals are covered in white ash” means you have passed peak temperature. White ash is calcium carbonate residue (from limestone in briquettes) or mineral ash from burned wood. It acts as an insulating blanket over the glowing coal surface, reducing radiant heat output. The hottest searing window is when coals glow bright orange with minimal ash formation, typically 15–20 minutes after lighting (AmazingRibs, 2024).

The chimney afterburner. The highest-heat searing method on a charcoal grill is to place a wire grate directly on top of a lit chimney starter. The chimney acts as a natural draft accelerator, forcing air up through the coal column. Measured grate temperatures exceed 430°C (800°F). In the Adam Savage and Kenji López-Alt searing comparison (testing blowtorch, Searzall, charcoal grill, chimney, and an aluminum forge at 650°C), the chimney produced the best combination of searing speed and crust quality (López-Alt, 2015).

Oxygen is the limiting reagent. In a blacksmith’s forge with forced air, charcoal reaches 1,650–2,200°C. On a backyard grill with natural draft, you get 760°C maximum. The difference is oxygen supply. Fully opening the bottom vents, removing the lid during searing, and pointing a hair dryer at the intake vent are all legitimate techniques for increasing oxygen flow and therefore combustion temperature (Blonder, AmazingRibs).

The slow indirect fire. The reverse sear inverts the searing problem. Instead of maximum flux for seconds, you need minimum flux for most of an hour, holding the chamber near 225°F (107°C) while the center slowly drifts up. A Weber kettle dumped full of lit coals runs far too hot for this; the whole difficulty of low cooking on a kettle is keeping it cool enough. The solution is to light only a fraction of the fuel and let it propagate. In the Minion method (named for Jim Minion, who developed it for competition barbecue), a small charge of lit coals is placed at one end of a bank of unlit briquettes. The lit coals set their neighbors alight one after another, so only a handful burn at any instant and the fire creeps along the bank for hours rather than flashing all at once.

The number that matters is not the starter charge but the steady burning count, and it is small. A 22-inch kettle sheds roughly 800 to 900 watts at 225°F (107°C), the sum of convection and radiation from its shell and the enthalpy carried up the open vent, while a slowly burning Kingsford briquette, holding about 525 kJ at the roughly 21 MJ/kg of a filled briquette (Blonder, 2024) and giving it up over the better part of an hour, supplies on the order of 175 watts. Only four or five need be alight at a time.

This is why scattering the lit coals is a mistake. Spread a dozen across the whole bed instead of banking them at one end and all of them catch at once, three or four times the needed power; because radiative losses climb as the fourth power of temperature the chamber does not rise threefold but settles well into the 400s°F (above 200°C), which is plenty to ruin a steak. The model behind these numbers, and the way the briquette count scales with the target temperature, is shared code (charcoalFire in the recipe engine), so every charcoal recipe on the site computes its own fire rather than quoting a rule of thumb.

The snake (or fuse) variant arranges briquettes two wide and two high in a C around the kettle perimeter and lights one end, turning the fuel into a slow-burning fuse good for six to eight hours. Combustion rate, and therefore temperature, is set by oxygen at the bottom intake vent (Blonder, 2024); the top exhaust vent stays open to maintain draft and sits over the meat so heat is pulled across it. Because the snake’s burn rate is governed by its geometry rather than by airflow, it holds a low temperature even on a kettle whose vents do not close, where the lit-charge size and snake length become the only controls; on a kettle with adjustable vents, the bottom intake gives finer trim.

Briquettes are the correct fuel here, which is the exact inverse of the searing case: the limestone and starch fillers that cap their peak temperature and make them useless for a sear are precisely what makes them burn low and steady through a long cook (Meathead Goldwyn, 2016). When the center reaches its pull temperature, rake the coals into a tight mound, add a fresh charge of lit lump, and open every vent to convert the gentle fire into a searing one.

VI-A. Smoke, the stall, and the wrap

The indirect path above is a short reverse sear: an hour in the chamber, then a sear. Stretch it to a real low-and-slow cook – a brisket, a rack of ribs, anything that needs the better part of a day – and a different chemistry takes over, one I spent an embarrassing number of hours not understanding while I sprayed apple juice at meat like a man watering a houseplant. Smoke flavor is not something the meat soaks up like a marinade. It is a surface deposit, laid down from the vapour phase of the smoke onto the film of moisture sitting on the cool, damp exterior. A wet surface takes up roughly twenty times the phenols of a dry one over the same smoking time (Tóth & Potthast, 1984).

The transport behind that number tells you how to help it. Smoke rides the chamber’s air currents, but a stagnant boundary layer clings to the meat and the lightest particles mostly glide past it. What drags them through is a cold, damp surface, working three ways at once: a tacky film that holds whatever impacts it, the outward rush of evaporating water that pulls smoke in behind it, and thermophoresis, the drift of suspended particles from hot gas toward a colder surface (Blonder, 2011). All three are strongest when the meat is cold and wet. That is the unglamorous reason to start a smoke cook with the cut straight from the fridge and keep its surface moist: Blonder traces between half and three-quarters of the deposited particles to that opening cold, and the smoke ring deepens with it.

The mahogany color and most of the smoke taste, then, are written onto the meat early, in the first hours, while the surface is still cold and moist and the bark has not yet set. Once it dries, uptake falls off a cliff.

The smoke ring runs on the same early clock. The pink halo just under the surface is not smoke at all but chemistry: nitrogen dioxide from the fire dissolves in the surface moisture and drives the formation of the same pink nitrosyl pigment that makes cured ham pink (Cornforth et al., 1998). It is a reaction that needs a wet surface and live myoglobin, and it stops growing once the myoglobin just under the skin denatures and can no longer bind the gas, which pitmasters peg at around 170°F (Goldwyn), a practical number from the pit rather than a lab measurement. So the ring and the bulk of the smoke are both essentially done in the first few hours, set into a surface that is by then drying toward bark.

Here is the part that took me too long, and it is the same failure that opens this article in its loud form: the charred-outside, raw-inside chicken was only the obvious version of a trap thick cuts set more discreetly. Once the smoke has done its early work, more time in the chamber buys almost no extra flavor and mostly just dries the meat out, and a low fire dries it in a particular, ruinous way.

The surface decides its own fate by its own temperature and moisture, paying no attention to the air. While it is still wet, the water boiling off cools it and pins it near the chamber’s wet-bulb temperature, the same evaporative plateau that stalls a barbecue for hours (Blonder, 2011); it loses water steadily but stays soft. To brown, it has to be both dry and properly hot, because the Maillard reaction does not begin until 140°C (Mottram, 1998). A 107°C oven, or a pit held at the same 225°F, splits those two conditions cruelly: hot enough to dry the surface, nowhere near hot enough to brown it.

Run it long enough and the surface dries past its critical moisture content and slips into the falling-rate drying period, where evaporation can no longer keep up and the outer layer case-hardens into a leathery rind (Mediani et al., 2022). You can drive water out of that rind but not back in, so a terminal sear merely browns leather; the come-up has spent an hour tanning the hide while your back was turned. A wrapped brisket, by contrast, climbs straight through to the set point while its naked twin stalls and dries (Modernist Cuisine, 2012), which is the whole case for the Texas crutch: once the smoke is banked, wrap the meat so it stops shedding water.

How you wrap depends on what you are cooking in. The oven is the easy case: it makes no smoke worth keeping, so there is nothing to lose by wrapping the cut in foil for the whole come-up and unwrapping it only for the sear. Sealed in its own vapour it cannot dry, and the crust is built in the final blast regardless, so the wrap costs nothing.

A covered grill asks for a lighter touch, because the first hour is the one you do not want to give up: while the surface is still wet it drinks smoke and is in no danger of leathering, so the cook runs open through that hour and only then disappears under foil for the drying remainder, until the sear. Under all of it the grill carries a water pan the whole time, whose humidity raises the wet-bulb temperature and slows the drying, but only slows it; on a long cook the pan is no substitute for the wrap.

Foil, not butcher paper, and the choice goes against instinct, because paper is what a brisket cook reaches for. Paper breathes, so it keeps the bark crisp and lets a little smoke through; but you wrap only after the smoke window has closed, so there is almost no smoke left to keep, and a rib or a steak is rescued at the sear, where the crust is rebuilt from nothing. That leaves moisture as the only thing the wrap is really for, and on moisture foil wins outright: a tight foil seal drives the humidity against the meat to nearly 100% and drops evaporation to near zero, where permeable paper only slows it (Blonder, 2013).

So foil, wrapped tight with no slack air gap to leave the seal incomplete. Pitmasters wrap a brisket in foil this way and call it the Texas crutch; the physics is indifferent to the fact that we have borrowed it for a steak. The one-hour split is reasoned, not measured, the rough point where the surface leaves its wet, constant-rate phase, but the true moment turns on the grill’s own humidity and airflow that the model does not see, so a cook that wraps up inside the hour simply never wraps at all.

Wrapping is the fix for cuts thick enough to need it, but on a covered grill there is a cruder way to dry a cut out, and it is the cook. Every lift of the lid vents the warm, humid pocket that keeps the surface damp and out of the falling-rate drying, and the water pan you so thoughtfully set inside cannot save you, because its humidity walks out the moment you do. Check the meat every fifteen minutes for five hours, as I once patiently did, and you have built an open fire and forgotten to baste it. The cure is to stop opening the lid at all: a leave-in probe in the meat and a second clipped to the grate read both the meat and the chamber from a base on the counter, so the one reason to lift the lid evaporates. And if you must open it, mist the surface afterward with salmuera, the asador’s light salt-water spray, to put back the moisture you let out.

That salmuera looks like the spritzing I mistook for technique, and it is not the same thing. Misting the surface with juice or vinegar does not drive smoke deeper or build bark; what it mainly does is cool the surface by evaporation, holding it back near the wet-bulb temperature and delaying the very browning the cook is working toward. And re-wetting a surface that has already taken its smoke buys little, because uptake needs the surface damp, not glistening. So as a way to add smoke or hurry bark, spritzing is a myth, and a heavy, frequent douse is actively counterproductive.

But a thin coat on a still-moist surface, repeated, is a real anti-leather tool: it holds the surface in its wet, constant-rate phase so it never crosses into the rind, which is exactly what an asador’s salmuera does over a six-hour open fire, where there is no lid to trap humidity and no water pan worth setting, so the spray is the entire defence. The distinction is timing and dose. Mist a still-damp surface lightly and often and you are preventing leather; spray a surface that has already hardened and you are wetting a barrier the water cannot cross. None of this is a measured result, I should say: no one has run a controlled spritzing trial, and this is a deduction from the same wet-bulb physics that explains the stall (Blonder, 2011), not a number from a lab.

If you want a damp chamber without touching the meat, a water pan is the better tool, and nearly free upside, but only under a lid. Sealed in, the steam it gives off holds the chamber’s humidity so the surface stays tacky and takes up smoke longer (Tóth & Potthast, 1984), and wetter is not simply better: deposition peaks near 60% relative humidity (Chan and Toledo, reported in Sérot et al., 2004), the moderate, steady damp a pan holds rather than the drowned surface a spritz leaves. Its other job is thermal: a pan of water is a slab of heat capacity that soaks up the pit’s swings and gives them back, smoothing the temperature through a long cook (Goldwyn).

In the open, both jobs vanish, the humidity blows away and there is no chamber air to buffer, which is why asado relies on salmuera on the surface and not a pan beneath it. The same caveat bites even under a lid, with the vents wide open or in a high-airflow cooker: if the chamber swaps its whole volume of air faster than the pan can boil water off, the humidity never climbs and the pan is decoration (Blonder, 2013). The honest gauge is not whether a pan is present but how fast it loses water; set it near the fire so it runs hot, and in a reasonably sealed, low-airflow cooker the meat’s own escaping juices already do much of the humidifying. With that caveat, a water pan belongs in any long indirect cook in a closed cooker by default.

VII. Resting

Resting meat after cooking allows two processes to occur. First, the steep thermal gradient created by high-heat searing equilibrates: heat flows from the scorching exterior toward the cooler center, evening out the temperature. Second, and more importantly, as muscle fibers in the outer layers cool below 60°C, they relax slightly and reabsorb some of the liquid that was expelled during contraction. Cutting into unrested meat allows this liquid to escape as visible juice loss on the cutting board.

For meat cooked in a hot chamber (the oven or indirect-grill paths), resting is critical because the thermal gradient is steep: the surface may be at 200°C while the center is at 55°C. Carryover cooking is the continued rise of the center after removal from the heat, driven not by any new energy during the rest but by the heat already stored in the hotter outer layers, laid down by the oven and then the sear, redistributing inward. Its magnitude is the difference between medium-rare and medium, so the calculator pulls the meat early by exactly the amount it predicts the center will coast.

That prediction is not a fixed rule of thumb, because carryover is not a fixed number. It scales with the driving gradient. A controlled experiment that roasted identical cuts at two oven temperatures found carryover roughly doubling or tripling with the hotter oven: a beef chunk rose 8.4°F at 300°F but 13.2°F at 425°F, and a pork loin rose 6.5°F versus 15.5°F (ThermoWorks, 2024). At the other extreme, a brisket held in a 225°F smoker shows almost no carryover at all, because the long cook and evaporative cooling at the stall flatten the surface-to-center gradient before the meat ever comes off (Blonder, 2024). A model that depended only on thickness would predict the same coast in every case and be wrong in both directions.

The calculator starts from the one-term conduction solution it already uses for cook times. At the pull moment the slab carries the profile T(x) = T_\text{amb} + (T_c - T_\text{amb})\cos(\zeta_1 x / L); letting that profile redistribute adiabatically raises the center to the slab’s mass average, a rise of (T_\text{amb} - T_\text{pull})(1 - \sin\zeta_1 / \zeta_1), where the eigenvalue \zeta_1 is fixed by the Biot number and so grows with thickness and with the surface heat-transfer coefficient. The bracketed driving term is what makes the same steak carry over more on a 220°C grill than in a 110°C oven. That term is the oven’s share of the coast, and a single rest-loss coefficient is calibrated so it reproduces the oven measurements above (ThermoWorks, 2024).

But the oven gradient is only half the story for anything that gets seared. A sear drives a brief, intense pulse of heat into the outer few millimeters, and during the rest that pulse keeps flowing inward, finishing the center alongside the oven gradient; pull only for the oven’s coast and a hard sear will carry the center clean past the target. The one-term series cannot represent that pulse, because it assumes a fully developed gradient, so for the reverse-sear and grill paths the calculator runs the actual transient on a finite-difference slab: the oven gradient at the pull, then the sear as radiative flux from the coals (Stefan-Boltzmann, surface emissivity 0.9), then a convective rest, solving for the pull temperature whose center peaks exactly at the target. Because the sear delivers a roughly fixed dose of energy into the surface but the mass it spreads through is not fixed, a thin cut coasts much more than a thick one: a 0.75-inch chop given a one-to-two-minute briquette sear climbs about 25°F (14°C) after the pull, while a thick steak off a fast flash climbs only 11 to 13°F, the upper end of what is reported for fast high-heat cooking (Meathead Goldwyn, 2016). That is the reverse of the thickness-only intuition, in which a bigger cut always coasts more, and it is why the recommended pull for a thin seared cut sits so far below its finish temperature.

For sous vide meat, the traditional rest is unnecessary and counterproductive. The interior is already at a uniform temperature (the whole point of sous vide), so there is no thermal gradient to equilibrate and no carryover cooking to account for. A long rest after searing only cools the crust and softens it. Both Meathead Goldwyn and Kenji López-Alt recommend serving sous vide steaks immediately after searing. A brief pause of 1–2 minutes on a wire rack (not a plate, left uncovered) lets surface steam dissipate so the crust stays crisp when it contacts the cutting board.

Conclusion

The entire process reduces to five steps: salt, cook, chill, sear, rest. A thick slab earns a sixth, wrapping the cook, which the calculator slips in only when the come-up runs long enough to leather the surface. The calculator above computes the parameters for each step from your cut’s thickness, weight, and protein type. The underlying physics (Fick’s diffusion, Baldwin’s slab heating model, USDA pasteurization kinetics, Stefan-Boltzmann searing) are all in the science notes.

If you change one thing, change the chill: a few minutes with the sous vide bag in cold water before searing (the calculator sizes it to your cut) drops the surface temperature and buys the outer layer the thermal headroom a sear needs. After that, build a hotter fire. Chimney searing with lump charcoal beats briquettes on a standard grill with the lid on, and it is not close.

References