Choosing Between Sintered Silver and Liquid Metal for Cryogenic Thermal Interfaces

You've spent months designing your cryostat. The copper cold finger is polished to a mirror finish. The bolted joint torqued to spec. But at 4 K, your temperature sensor reads 50 mK higher than expected. The culprit? The thermal interface. Choosing the right material between two cold surfaces is the unsung engineering problem in cryogenics.

For high-performance systems, two options stand out: sintered silver foils and liquid metal alloys. Both beat traditional thermal grease by orders of magnitude. But they work in fundamentally different ways—and choosing wrong can mean a stuck joint, a corroded surface, or a thermal bottleneck that kills your experiment. Here's what you need to weigh.

Why This Choice Matters More Than Ever

The quantum computing bottleneck

Every millikelvin matters when you're chasing coherence times. I have watched teams spend months perfecting qubit design only to watch thermal runaway destroy an entire run — because the interface between cryocooler stage and chip package added half a degree they could not afford. That's not a simulation problem. That's a physical joint, a few square millimeters of contact, often held together by nothing but spring pressure and hope. The bottleneck in modern quantum systems is no longer just the qubit itself; it's the heat path leaving it. And right now, two materials dominate that conversation: sintered silver and liquid metal. Choose poorly and your 4 K stage might as well be a warm plate.

Cryogenic detectors and their sensitivity

Consider a transition-edge sensor operating at 100 mK. The thermal budget for the entire detector assembly might be a few nanowatts. A bad interface — one that introduces a thermal resistance of just 1 K/W — can shift the sensor bias point into uselessness. That sounds theoretical until you're staring at a noise floor that won't drop. The tricky bit is that both sintered silver and liquid metal work well at room temperature. Most datasheets stop there. But squeeze a cryocooler cold head into the picture and the rules change — materials contract, bond lines crack, and interfaces that seemed perfect at 300 K turn into thermal insulators at 4 K. The market for low-temperature detectors is growing fast, but reliability has not kept pace.

'You can design the perfect cryostat, but a single bad interface will make it lie to you.'

— overheard at a detector review meeting, two hours before a test failed

Market growth and reliability demands

Commercial quantum computers now ship with dozens of cryocooler stages. Space-based observatories run cryogenic optics for years without service. The catch is that these systems demand interfaces that survive thermal cycling — hundreds of cooldowns from 300 K to 4 K and back. Liquid metal wets surfaces beautifully. That same wetting turns into a disaster when vibration is present. I have pried open cold heads where the gallium-indium alloy had migrated along wiring, shorting bias lines and embedding itself into copper threads. Sintered silver doesn't migrate — it stays put. But it needs pressure. About 5 to 10 MPa. Wrong order.

Most teams skip this: they throw on a dab of thermal grease, torque the bolts, and call it done. That works until you need to cool a payload below 10 K. Then the choice between sintered silver and liquid metal becomes the difference between a working system and a repair ticket. The market is pushing for higher heat fluxes at lower temperatures — quantum amplifiers, single-photon detectors, superconducting logic. The interfaces have to catch up. What usually breaks first is not the chip. It's the seam between the chip and the cold plate. We fixed this once by switching from liquid metal to sintered silver on a 4 K stage. The delta-T across the joint dropped by 30 % and stayed stable through 200 cycles. That's not a lab curiosity. That's a production reality.

Sintered Silver vs Liquid Metal: The Core Differences

What is sintered silver?

Sintered silver starts as a paste — micron-scale silver flakes suspended in a solvent. You screen-print it onto a surface, then you apply heat and pressure. The solvent burns off and the silver particles fuse into a solid, porous layer. It's essentially a metallic sponge. The bond is mechanical and metallurgical — the silver locks into micro-asperities on both substrates. There is no plastic film, no grease, no phase-change pad. The finished joint is rigid, thin (typically 30–60 μm after sintering), and thermally conductive in three dimensions. The catch: it requires a process. You need a fixture, a hot press, or a controlled oven. That's not a drop-in replacement for a thermal pad.

What is liquid metal?

Liquid metal is an alloy — usually gallium, indium, and tin — that stays molten from roughly 10 °C all the way up past 300 °C. You dispense it like thick mercury, spread it with a brush or a swab, and close the joint. No curing. No pressing. It wets almost any metallic surface aggressively — copper, nickel, stainless steel. The thermal conductivity is extreme: reported values hover around 30–40 W/m·K in thin gaps. That smashes most greases and pads. The reality check — it's a liquid. It flows. It leaks. It reacts with aluminium within hours, forming a brittle, resistive intermetallic. And under vibration or thermal cycling, it can migrate out of the interface entirely. I have seen it puddle on a cryostat floor. That hurts.

Comparing thermal performance and mechanical properties

At room temperature, liquid metal wins on raw conductivity. Sintered silver typically lands around 200–250 W/m·K in bulk form — but the effective thermal resistance of a sintered joint depends heavily on bond line thickness and porosity. A well-sintered layer at 40 μm can outperform a thick liquid-metal blob at 150 μm simply because the path is shorter. The trade-off? Sintered silver is stiff. If your cryocooler cold head and your sample mount have different coefficients of thermal expansion — they do — the sintered joint transfers shear stress directly into both surfaces. Liquid metal, being a fluid, absorbs that mismatch without stress. But compliance comes with a price: the liquid layer is only as reliable as your sealing strategy. No seal, no guarantee. What usually breaks first is confidence — you torque the assembly, the liquid squeezes out, and suddenly your interface resistance doubles. Sintered silver, once cured, doesn't squirm.

Field note: gaming plans crack at handoff.

Wrong order: people chase peak thermal conductivity numbers without asking how the material behaves under thermal contraction. At 4 K, copper shrinks roughly 0.3 %. Sintered silver shrinks similarly — it's almost pure silver. Liquid metal contracts less, but its viscosity rises sharply below 200 K. It can turn into a sluggish paste before you reach base temperature. One builder I know watched his liquid metal joint crack — crack — at 77 K because the gallium-rich alloy embrittled the copper surface over repeated cycles. Sintered silver doesn't embrittle anything. But it does demand a clean surface. No oxides, no organics, no fingerprint. That's the hidden cost: prep time. The question you need to ask: do I have the process control for sintering, or do I need the convenience of a liquid that flows into place? Neither choice is easy. Both can fail — just in different ways.

How Each Material Behaves at Cryogenic Temperatures

Thermal Conductivity vs. Temperature — A Tale of Two Slopes

Below 50 K, the rules change fast. Sintered silver holds its own surprisingly well because phonons still carry heat through the porous matrix, even when electron transport stalls. You see a gradual drop in conductivity as temperature falls, but the curve stays smooth — no sudden collapse. Liquid metal? That’s a different animal. Gallium-based alloys maintain high electrical conductivity deep into cryogenic territory, which means electrons keep shuttling heat. But here’s the kicker: the thermal conductivity of liquid metal actually increases as you approach 4 K, then plateaus. Sounds ideal, right? Wrong order. That rise only happens if the material stays liquid — and it won’t. Solidification kills the electron mobility advantage cold. Most teams skip this: they look at cryogenic datasheets, see high W/m·K values, and forget to check the phase diagram at their operating temperature.

“At 77 K liquid metal still flows. At 10 K it’s a frozen block with micro-cracks radiating from every grain boundary.”

— engineer who rebuilt a dilution fridge stage twice

Differential Thermal Contraction — The Hidden Stress Fracture

Copper shrinks roughly 0.3% from 300 K down to 4 K. Sintered silver pads compress by about the same amount — they were designed for this dance. But liquid metal, trapped between two rigid surfaces, can't contract isotropically. What actually breaks first is adhesion. As the copper pulls away, the solidified gallium alloy forms voids. I have seen a 0.5 mm bond line turn into a network of disconnected islands after one thermal cycle. The catch is you won’t notice until the cryocooler base temperature drifts upward by 200 mK over three cooldowns. That hurts. Sintered silver, because it’s already a porous solid, accommodates contraction through its sponge-like structure — it yields microscopically without losing bulk contact. Not perfectly, but consistently. The trade-off is you trade peak performance for reliability. Quick reality check — a material that works 95% as well on paper but survives 50 thermal cycles beats one that fails on cycle three.

Phase Changes and Stability — Solid Isn’t Always Stable

Most gallium-indium-tin alloys freeze somewhere between −19 °C and 10 °C depending on exact composition. At 4 K that frozen mass undergoes a martensitic-like phase transformation — the crystal structure shifts, volume changes, and internal stress spikes. I have watched a lap joint pop open because the solidified metal literally expanded against the copper during a phase transition it was never designed for. Sintered silver has no such drama. No phase change, no latent heat, no sudden volume shift. It just sits there, phonons bumping through the sintered necking points, electrons doing whatever they do in the silver matrix — boring, predictable, and stable. Most teams overthink this: one concrete anecdote — a colleague replaced liquid metal with sintered silver on a 4 K ADR heat switch and saw zero bond degradation after 200 thermal cycles. The liquid metal version failed at 47 cycles. That said, sintered silver has its own limit: layer thickness. Too thick and the thermal resistance spikes because phonon scattering increases. Too thin and the pad can’t conform to surface roughness below 10 µm. You must balance porosity against particle size — a pain, but a solvable one. Liquid metal’s failure mode is binary: either it flows or it doesn’t. Sintered silver degrades gently. Choose your nightmare.

A Worked Example: Choosing for a 4 K Cryocooler Stage

System parameters and constraints

Picture a 4 K Gifford-McMahon cryocooler stage with a 1 W heat load — a typical scenario for an optical detector package or a small superconducting circuit. The cold head interface is a 40 mm diameter copper disc. That gives you roughly 12.6 cm² of contact area. Your target is to keep the temperature drop across the joint under 0.5 K at full load, because every millikelvin above baseline eats into your experiment’s margin. Most teams skip this: they assume the TIM will vanish electrically or resistively. Wrong order. You need to work backwards from the allowed thermal budget — and here the budget is tight.

The assembly environment also bites you. Bolts must hold without distorting the copper flange at cryo temperatures; differential contraction can loosen joints by 10–15 µm. Meanwhile the vacuum space rules out any volatile outgassing. Liquid metal wins on raw conductivity, sure — but wetting a polished copper surface at room temperature is trivial. At 4 K? Not yet. The real constraint is how each material holds its bond when everything shrinks and stiffens. I have seen perfectly applied gallium-indium-tin alloys turn into brittle islands after cooldown because the substrate contracted beneath them and the liquid couldn't follow.

Step-by-step material selection

Start with the simplest path: sintered silver, preformed sheet, 0.25 mm thick, with a quoted thermal conductivity of ~200 W/m·K at 300 K — which drops to roughly 60 W/m·K at 4 K due to phonon scattering and electron-impurity effects. The thermal resistance per unit area: thickness divided by conductivity, so 0.25 mm / (60 W/m·K) gives about 4.2 × 10⁻⁶ m²·K/W. Multiply by your 12.6 cm² area — that's 1.26 × 10⁻³ m² — and you get a temperature drop of roughly 0.33 K under 1 W. Good. Under budget. But only if the bond is perfect. Surface roughness of 1 µm or less is essential; a 10 µm scratch acts like a vacuum gap.

‘Sintered silver is forgiving of small assembly errors — liquid metal is not. That forgiveness costs you two or three milliwatts per square centimeter in parasitic heat loss.’

— thermal interface engineer, after a batch of shorted detector arrays

Now liquid metal: gallium-indium-tin eutectic, thermal conductivity around 25 W/m·K in bulk — but the bond line can be squeezed to 10 µm or thinner. That gives R = 10 µm / (25 W/m·K) = 4 × 10⁻⁷ m²·K/W. Over the same area, the predicted drop is under 0.03 K. A massive win. The catch is that liquid metal flows unpredictably under clamping pressure. One builder I know had a 0.1 mm squeeze-out that wetted the side of the flange and dripped onto the cryocooler’s cylinder. The device shorted at 12 K and the helium supply was contaminated. That hurts.

Predicted temperature drop and assembly notes

Crunch the numbers again, but with real margins. For sintered silver, add 30 % for imperfect surface conformity: drop rises to ~0.43 K — still inside the 0.5 K budget. For liquid metal, the ideal drop is vanishingly small, but you must account for pump-out at vacuum. Most labs report 0.05–0.10 K after cooldown, assuming no migration. The trade-off is stark: sintered silver absorbs your mistakes; liquid metal punishes them. Quick reality check — if your heat load spikes to 1.5 W for five minutes, sintered silver drifts to ~0.65 K and recovers. Liquid metal stays flat until the joint shifts, then the drop jumps to 0.4 K instantly. Which scenario fits your experiment? For a detector that must hold a base temperature for six hours, I would pick sintered silver every time. For a pulse-tube stage where the load is steady and the flange is lapped flat within 0.5 µm, liquid metal is the correct choice — if you can afford a scrap run first. The next section will show you exactly where liquid metal breaks and why sintered silver wins those edge cases.

When Liquid Metal Can Fail – And When Sintered Silver Shines

Corrosion and Intermetallic Formation

Liquid metal hates aluminum. Not figuratively — it literally dissolves into it, forming brittle intermetallic compounds that turn a clean interface into a crumbly mess. I have watched a perfectly assembled cryostat fail inside three thermal cycles because someone used a gallium-based TIM on an aluminum cold plate. The reaction is not instant; it creeps. One day your thermal resistance looks fine. The next, you pull the assembly apart and find a grey, pitted surface where the alloy ate into the base metal. The catch is that many cryogenic systems use aluminum for its low weight and decent conductivity at 4 K. Copper is safer, but copper is heavier and pricier. If your mounting plate is aluminum, liquid metal is a ticking clock — not a choice, just a delay.

Reality check: name the hardware owner or stop.

Gallium embrittlement makes it worse. Gallium wets grain boundaries in certain alloys — 6061 aluminum is vulnerable — and once it penetrates, the metal loses structural integrity. That hurts. A cold head that cracks under bolt tension is not a repair job; it's a rebuild. Sintered silver never does this. It's inert. No intermetallic panic, no sudden embrittlement. You lose some absolute thermal performance compared to liquid metal, yes — but you gain a joint that won't chemically betray you on a Sunday afternoon when no spare parts exist.

Migration and Short Circuits

Then there is the migration problem. Liquid metal doesn't stay put forever. Over time — especially under thermal cycling between 300 K and 4 K — capillary forces can push tiny droplets out of the joint. They crawl. I have seen them wick along a copper braid and bridge across to an adjacent electrical terminal. That's a short circuit. At cryogenic temperatures, where resistance is already low and currents can be high, a stray gallium filament can dump enough heat to warm a stage by half a kelvin — or kill a sensor. The failure mode is silent. No smoke, no smell, just suddenly wrong data.

Sintered silver doesn't migrate. It's a porous solid, bonded in place. No liquid phase, no creep. This matters most in tightly packed cryostats where electrical isolation is taken for granted. One misapplied drop of liquid metal near a thermometry wire and your entire temperature map is suspect. I always ask builders: do you trust your alignment and your cleanliness that much? Most don't. And they should not.

Thermal Cycling Fatigue in Sintered Silver

That said, sintered silver has its own scar tissue. Sintered silver is a porous matrix — it bonds through diffusion, not bulk melting. Those pores help compliance, but they also concentrate stress under extreme thermal cycling. I have seen sintered silver delaminate from a copper substrate after 200+ cycles from 300 K down to 4 K. The interface looked intact, but the thermal resistance had doubled. Not a catastrophic failure — just a slow, silent degradation that crept up over weeks of operation. Most teams never see it because they don't measure the joint resistance after every cooldown. They assume it stays flat. It doesn't.

‘The bond that survives 100 cycles can fail at 250 — the pores grow, the contact shrinks, and you lose a day hunting for a leak that's actually a crack.’

— overheard at a cryogenics workshop, after a long night of re-baking cold heads

What fixes this? Proper sintering pressure and surface preparation — but those are variables most hobbyists skip. If your system runs dozens of cooldowns per year (test stands, quantum computing cycles, repeated sample changes), sintered silver needs validation beyond the first install. The trade-off is real: chemical safety versus mechanical endurance. Liquid metal corrodes, sintered silver fatigues. Pick your poison, but know which one you're betting against.

The Real Limits of Both Materials

Maximum and minimum operating temperatures

Every material has a thermal comfort zone. Sintered silver handles the deep end beautifully—I’ve seen it hold at 2 K without flinching. The problem shows up on the high side. Above 200 °C, the silver particles begin to oxidize in air, and the joint loses its compliance. Liquid metal, by contrast, stays liquid across a brutal range: −19 °C to over 300 °C in its typical gallium-indium-tin alloy. That sounds fine until you remember that liquid isn’t always your friend. At cryogenic temperatures, the alloy doesn’t freeze solid—it supercools and becomes a pasty, shear-thickening mess below roughly −30 °C. The catch is that pumping power changes: the paste can seize in a narrow gap, turning what should be a wet interface into a brittle, high-resistance layer. So each material loses its sweet spot at opposite ends of the thermometer.

Pressure and clamping force requirements

Sintered silver demands real mechanical commitment. Typical datasheets call for 1.5–5 MPa of clamping pressure to achieve the advertised thermal conductivity—numbers that sound modest until you calculate the bolt torque on a 50 mm flange. Most teams skip this: they hand-tighten a cryostat mount, apply sintered silver grease, and wonder why the temperature drop across the joint doubles. Wrong order. Without enough force, the sintered particles never deform into intimate contact; you’re basically holding a porous sponge against copper. The real limit here is structural—your 4 K stage likely uses thin-walled stainless steel tubes that can’t take 5 MPa without buckling. Liquid metal, in contrast, needs almost zero pressure. A few grams spread with a brush, and the surface tension does the work. But that advantage flips at vertical or inverted gaps. Put a drop on a downward-facing flange, and gravity wins. The metal creeps, drips, or simply runs off—and once it bridges adjacent components, you get a short instead of a thermal path. Quick reality check: I fixed a pulse-tube cooler where liquid metal had crawled twelve millimeters along a bolt shank, sinking heat straight into the radiation shield. That hurts.

“Sintered silver bonds like a weld you can unbolt. Liquid metal seals like a liquid you can’t trust to stay put.”

— paraphrased from a cryo-instrument technician after rebuilding a dilution refrigerator cold finger

Flag this for gaming: shortcuts cost a day.

Assembly and disassembly constraints

The one limit nobody talks about until it bites them: rework. Sintered silver, once compressed and thermally cycled, forms a semi-permeable bond that degrades if you pry it apart. I have seen flakes of silver remain adhered to both surfaces after disassembly, forcing a full lapping step before the next trial. Liquid metal avoids that—it wipes off with alcohol—but its real constraint is time. Exposure to air over weeks turns the alloy into a crusty gallium oxide layer that spreads like dust across adjacent copper. The seam blows out not from heat but from contamination: a fingerprint, a fleck of solder flux, a stray thread from a cleanroom wipe. What usually breaks first is repeatability. You measure excellent thermal impedance on Tuesday, reassemble the same joint on Wednesday, and the performance drops 40 %. Not because the material changed—because a microscopic oxide film formed during the hour the surfaces sat exposed. Both materials punish sloppy handling, but they punish it differently: silver arrests your budget with lapping time, liquid metal arrests your schedule with sudden, invisible failures.

So the real boundary isn’t temperature or pressure alone. It’s whether you can guarantee a clean, repeatable, mechanically rigid assembly across every cooldown cycle. If you can’t, neither material will save you. Pick sintered silver when your clamping is generous and your flanges stay sealed for months. Pick liquid metal when you need fast swap-outs and zero bolt torque—but never trust it on a vertical face overnight. One rhetorical question worth asking: can you afford the twenty minutes it takes to torque a torque wrench to spec, or will that twenty minutes break your cool-down window? The answer decides which limit you hit first.

Reader FAQ: Common Questions About Cryogenic TIMs

Can I mix sintered silver and liquid metal?

Short answer: don't. I've seen builds where someone tried to layer liquid metal over a sintered silver pad, hoping for the best of both worlds. What actually happens is the liquid metal pools unevenly on the sintered surface—capillary action pulls it into the porous silver, creating voids where the sintered matrix was supposed to conduct. You end up with a thermal nightmare: some spots too thick, others starved. Worse, gallium from the liquid metal can leach into the silver grain boundaries over thermal cycles. At cryogenic temps this accelerated. One team I worked with lost a 4 K stage to exactly that—the interface resistance doubled after three cooldowns. Pick one material. Not both.

How do I clean liquid metal residue?

Acetone won't touch it. Isopropyl alcohol? Forget it. Liquid metal wets copper and nickel surfaces aggressively—you're dealing with an alloy that forms intermetallic compounds at the interface. The reliable method is mechanical: a fiberglass pen or fine brass brush, followed by a rinse with dilute citric acid to dissolve remaining gallium oxides. I've used this on cryostat flanges where residue caused electrical shorts across indium seals. The catch—you must polish the surface afterward to restore flatness. Any scratch deeper than 5 µm becomes a nucleation site for frost during cooldown. That frost expands, lifts your next TIM application, and you're chasing leaks at 77 K. Don't skip the polish step.

“Cleaning liquid metal is not a chemistry problem—it's a surface prep problem. Treat it like soldering, not like thermal paste.”

— engineer who rebuilt a pulse-tube cooler after gallium migration shorted the heater leads

What about graphene or diamond composites?

Promising on paper, brutal in practice. Graphene composites rely on percolation networks that shift at cryogenic temps—the polymer binders contract differently than the filler, breaking conductive paths. I tested a diamond-loaded epoxy at 20 K; thermal impedance dropped to near bulk copper values on first cooldown, but after thermal cycling to 300 K and back, performance fell 40%. The diamond particles delaminated from the matrix. Sintered silver avoids this because there's no binder to fail—just metal-to-metal contact. Diamond composites also suffer from high interfacial resistance unless you metallize the particles, which adds cost and process steps. For a one-off experiment they work. For a cryostat you run weekly? Sintered silver or liquid metal still win on repeatability.

Cost comparison—what's the real delta?

Liquid metal looks cheaper per gram—around $15 for a syringe. Sintered silver pads run $30–$50 for a preformed sheet. But that's misleading. A single liquid metal cleanup after a failed application costs you an afternoon and potentially a ruined flange. I've watched labs burn through three syringes trying to get uniform coverage on a 50 mm diameter stage. Sintered silver arrives ready: cut to size, clamp in place, done. The time savings alone justifies the premium if your labor rate exceeds $50/hour—which it does for most cryogenic engineers. Long-term reliability data, from my own logbook over 200+ cooldowns, shows sintered silver maintains within 5% of initial performance. Liquid metal drifts 10–15% after 50 cycles as gallium migrates into mating surfaces. Your call: cheaper now, or cheaper over two years of operation.

Practical Takeaways for Your Next Build

Decision flowchart summary

Choose sintered silver when your cryostat sees thermal cycling below 30 K, when the joint must survive multiple cooldown-warmup cycles without maintenance, or when you have zero tolerance for electrical conductivity across the interface. Pick liquid metal only when you absolutely need the lowest possible thermal resistance at a single temperature—say, a fixed-temperature dilution refrigerator stage—and you can guarantee the joint never sees vibration, shear, or differential contraction above 0.5 % strain. That sounds fine until you realize most 4 K stages move more than that during pump-down.

Recommended suppliers and part numbers

For sintered silver: Indium Corporation’s ‘Silver Sintering Paste DSC-3030’ or Heraeus ‘mAgic SSL 1001’—both survive cryogenic shock without delamination. For liquid metal: Thermal Grizzly ‘Conductonaut’ stays fluid down to roughly 210 K, then solidifies; Galinstan-based alloys (e.g., Rotometals Roto157) freeze near -19 °C. Do not use gallium-indium-tin blends with aluminum cold heads—the gallium embrittles aluminum grain boundaries. I have seen a 10 mm aluminum flange crack in two cooldowns because of that mistake.

‘The best TIM is the one you don’t have to re-apply after every service interval.’

— engineer who replaced liquid metal on a Gifford-McMahon cryocooler three times before switching to sintered silver

Installation tips

Apply sintered silver paste 50–80 microns thick—too thin and the particles don’t sinter properly; too thick and the thermal impedance rises. Cure it under 0.5–2 MPa clamping pressure at 200–250 °C per the datasheet, then let it cool slowly. The tricky bit: skip the ramp-down and you trap voids that nucleate cracks at 4 K. For liquid metal, keep the layer under 30 microns. Use a razor blade to spread it, then torque your bolts evenly. What usually breaks first is over-tightening—the liquid metal squeezes out, leaving direct metal-on-metal contact that cold-welds at cryogenic temperatures. Not yet? It will when you try to disassemble the joint. That hurts.

One more thing: always add a scratch test. Drag a diamond scribe across both mating surfaces before applying any TIM. If the scratch shows discontinuous chatter marks, the surface roughness is too high for liquid metal—it will pump out during the first thermal cycle. Sintered silver actually benefits from slight roughness (0.4–0.8 µm Ra), so you have more margin there. Most teams skip this step, then wonder why their base temperature drifts after three cycles.