Cryogenic Chemistry at Minus 78 C: Cost and Why

Cryogenic chemistry at minus 78 °C shows up on a quote as one short phrase — “cryogenic conditions required” — and a number that is often two to four times what the same step would cost at ambient temperature. Buyers see it most often on organolithium steps, kinetic enolate alkylations, and stereoselective additions, and the usual reaction is to ask whether the CRO is padding the number. Almost always they are not. What changed between the discovery route and the pilot campaign is not the chemistry; it is the physics of getting heat out of a large vessel.

This post is written for the person holding that quote. It explains why −78 °C became the standard condition in the first place, which reactions genuinely need it and which merely inherited it from a discovery notebook, what the cost is actually built from at scale, and the four legitimate ways to remove or reduce it. It sits inside our broader CRO buyer’s guide, alongside the parallel capability questions on high-pressure hydrogenation and flow versus batch processing.

The single most useful idea in this post: at scale, cooling capacity determines your maximum batch size, not vessel volume. A 1,000 L reactor with 8 kW of cooling at −70 °C is, for a strongly exothermic lithiation, a smaller reactor than a 250 L vessel with the same 8 kW.

Why Minus 78 Degrees Celsius Specifically

−78 °C is the standard laboratory cryogenic temperature because it is what a dry ice and acetone bath gives you for free. Solid carbon dioxide sublimes at −78.5 °C at atmospheric pressure, so a slurry of dry ice in acetone (or isopropanol, or ethyl acetate) parks itself at that temperature and holds there without a controller, a thermocouple loop, or any capital equipment at all. Add more dry ice when it stops fizzing. That is the whole technology. The sublimation point is a fixed physical constant, tabulated in the NIST Chemistry WebBook, which is why the bath is so reproducible across laboratories.

This matters because it means −78 °C is a convenience temperature, not a chemistry temperature. Generations of methodology papers specify it because it was the cold bath on the bench, not because the authors screened the window and found a cliff at −75 °C. When a discovery chemist writes “−78 °C, 2 h” in a notebook and that procedure gets handed to a process group, the temperature is frequently untested. Part of what a good process chemistry engagement does is find out how much of that cold is real.

The Practical Temperature Ladder

Cost does not scale smoothly with temperature. It steps, and the steps follow the equipment classes available.

Target temperature	How it is reached at scale	Relative cost impact	Typical use
0 to −20 °C	Standard glycol chiller, common on any jacketed reactor	Baseline, essentially free	Most amide couplings, many quenches
−20 to −40 °C	Single-stage chiller with silicone or HFE heat transfer fluid	Modest, roughly 1.2 to 1.5x a room temperature step	Many Grignard additions, some lithiations
−40 to −70 °C	Cascade or two-stage chiller, low-viscosity fluid required	Substantial, often 2 to 3x	Directed metalation, kinetic enolates
−70 to −90 °C	Cascade chiller at the edge of range, or liquid nitrogen assist	High, 3x and up, batch size limited	t-BuLi chemistry, tight selectivity cases
Below −90 °C	Direct liquid nitrogen, specialised equipment	Rarely offered at scale by any CRO	Research scale only in practice

The important line in that table is −40 °C. It is reachable with widely available single-stage equipment and with heat transfer fluids that are still fluid enough to pump efficiently. Below roughly −45 °C you move into a different equipment class and a different fluid class at the same time, and that is where the cost step is sharpest.

Liquid Nitrogen Is Not the Easy Answer

Buyers sometimes suggest liquid nitrogen as a cheap route to arbitrary cold. At −196 °C it is far colder than needed, and that is exactly the problem. Direct injection of liquid nitrogen into a jacket gives you enormous cooling power with terrible control — you overshoot, freeze material on the vessel wall, and create a fouled heat transfer surface that then cannot remove heat at all. Frozen product on a wall is a common cryogenic failure mode and it degrades both yield and impurity profile. Most pilot plants use liquid nitrogen only as a trim on a cascade system, or for cold traps and condensers, not as the primary jacket coolant.

The Chemistry That Genuinely Requires the Cold

Cryogenic conditions are justified when the cold is doing one of two jobs: keeping a reactive species alive long enough to react where you want it to, or widening the energy gap between the pathway you want and the one you do not. Everything below is one of those two cases.

Organolithium Chemistry

This is the largest category by far. Alkyllithium reagents — n-butyllithium, sec-butyllithium, tert-butyllithium — and lithium amide bases such as LDA (lithium diisopropylamide) and LiHMDS (lithium bis(trimethylsilyl)amide) are extraordinarily strong bases and nucleophiles, and above their stability window they do not sit and wait. They attack solvent. In tetrahydrofuran, n-BuLi deprotonates the alpha position and the resulting anion undergoes retro-[3+2] fragmentation to ethylene and the lithium enolate of acetaldehyde. The half-life of n-BuLi in THF is hours at −20 °C and minutes at 0 °C; t-BuLi in THF is measured in minutes even at −78 °C. Cold is not a nicety here — it is what stops your base from eating the reaction medium.

Directed ortho metalation (DoM) is the workhorse application. A directing group — an amide, a carbamate, an oxazoline, sometimes a halogen — coordinates the lithium and delivers deprotonation to the adjacent ring position with regiochemistry that electrophilic substitution cannot reach. The aryllithium formed is then quenched with an electrophile. Run it warm and you get competing benzylic deprotonation, addition to the directing group itself, and metalation at the wrong position. Substrates like 2-Bromobenzonitrile (CAS 2042-37-7) illustrate the constraint sharply: a nitrile is a strong director but also a perfectly good electrophile, so the aryllithium must be formed and trapped faster than it adds to a neighboring nitrile. That race is won by temperature.

Halogen-metal exchange is the other major use. Treating an aryl iodide or bromide with n-BuLi or t-BuLi swaps the halogen for lithium, generating an aryllithium at a defined position without needing a directing group. It is fast — often complete in minutes at −78 °C — and it is the reason iodinated heterocycles are such valuable building blocks. 3-Iodopyridine (CAS 1120-90-7) is a canonical example: exchange at C3 is clean and rapid at −78 °C, but at −40 °C and above the resulting 3-pyridyllithium begins attacking unreacted starting material and adding to the pyridine ring of its neighbors. Similar behavior governs 4-Bromoindole (CAS 52488-36-5), where the indole N-H must be protected or deprotonated first and the C4 aryllithium has a limited lifetime. Both sit in our halogenated compounds category, which is where most exchange substrates live.

Kinetic Versus Thermodynamic Enolates

This is the textbook case where temperature literally determines which product you get, and it is worth stating plainly because it is the clearest justification a buyer will ever see for a cryogenic line item.

An unsymmetrical ketone has two enolizable positions. Deprotonate with a bulky, strong, non-nucleophilic base — LDA is the standard — at low temperature, and you remove the most accessible proton fastest. That gives the kinetic enolate, usually the less substituted one. The deprotonation is effectively irreversible under those conditions, so the ratio you get reflects the relative rates of the two deprotonations, not their relative stabilities.

Warm it up, or use a weaker base with a proton source present, and the enolates equilibrate. You then get the thermodynamic enolate, the more substituted and more stable one. Same substrate, same base, different temperature, different product — with a regiochemical outcome that is often 95:5 one way and 20:80 the other.

There is no reagent substitution that gets you out of this. If your route depends on alkylating the kinetic enolate, you need the cold, and you need the addition of the electrophile to happen before the enolate has time to equilibrate. Aldehyde electrophiles such as (S)-N-BOC-Prolinal compound the requirement, because the aldol addition itself sets a new stereocenter whose ratio is also temperature-dependent, and because BOC-protected aminoaldehydes racemize at the alpha center on warming.

Selectivity Improves as Temperature Falls

The general principle behind almost every stereoselective cryogenic step is worth stating once, cleanly, because it explains why chemists reach for the cold reflexively.

When two competing pathways lead to two products, the product ratio is governed by the difference in activation free energy between the two transition states — ΔΔG‡. The relationship is exponential and temperature appears in the denominator: the ratio goes as exp(ΔΔG‡ / RT). Lower the temperature and the same energy difference buys you a larger ratio.

The practical numbers matter more than the equation:

ΔΔG‡	Ratio at 25 °C	Ratio at −40 °C	Ratio at −78 °C
1.0 kcal/mol	roughly 84:16	roughly 88:12	roughly 91:9
1.5 kcal/mol	roughly 93:7	roughly 95:5	roughly 97:3
2.0 kcal/mol	roughly 97:3	roughly 98.5:1.5	roughly 99:1

Two things follow. First, the improvement from −40 to −78 °C is real but modest — typically one to three points of selectivity for a step-change in cost. Second, if a reaction is already at 90:10 at −40 °C, going colder will not rescue it; you need a better catalyst or a different auxiliary, not a colder bath. Buyers who understand this table can push back intelligently on a cryogenic specification instead of accepting or rejecting it wholesale. Our post on chiral resolution versus asymmetric synthesis covers the downstream question of what to do when selectivity is not good enough at any temperature.

Grignard Additions and Unstable Intermediates

Grignard reagents are more forgiving than organolithiums and much of their chemistry runs at 0 °C or ambient. The cases that need cold are the selective ones: 1,2 versus 1,4 addition to enones, mono- versus double-addition to esters and nitriles, and additions where an existing stereocenter directs the approach. Chelation-controlled additions in particular degrade quickly with temperature, because the chelate that enforces the facial preference is a weak, reversible interaction that thermal energy disrupts.

The other cold-requiring category is trapping unstable intermediates. Some species exist only long enough to be intercepted. Benzynes generated by base-mediated elimination, certain nitrogen ylides, unstabilized carbanions on electron-poor heterocycles, and the metalated forms of azoles like 3-Bromo-1H-indazole (CAS 40598-94-5) all fall here — the indazole ring system is prone to ring-opening and decomposition once metalated unless the temperature keeps the species in a narrow window. The heterocyclic compounds category is full of scaffolds with this behavior, and it is the single most common reason a heterocyclic step arrives with a cryogenic tag attached.

Ozonolysis

Ozonolysis sits slightly apart. Ozone cleaves carbon-carbon double bonds through a molozonide that rearranges to an ozonide, and both intermediates are energetic. The cold serves three purposes at once: it moderates a strongly exothermic addition, it prevents accumulation of peroxidic species that can decompose violently, and it improves the solubility of ozone in the reaction solvent so that the gas is consumed rather than swept through. Ozonolysis at −78 °C in dichloromethane or methanol is standard practice for a reason, and the safety argument for the cold is as strong as the selectivity argument. At scale, ozonolysis also carries a peroxide accumulation risk that requires in-process monitoring — a point worth raising explicitly with any CRO quoting the step.

The Real Cost Driver: Heat Removal at Scale

Here is the core of the matter. Cryogenic chemistry is cheap in a flask and expensive in a reactor, and the reason is geometry.

A 250 mL round-bottom flask sitting in a dry ice bath has an enormous surface area relative to its contents, and the bath is effectively an infinite heat sink that will absorb whatever the reaction throws at it. You can add n-BuLi in a couple of minutes and the internal temperature barely moves. A 500 L jacketed reactor holds 2,000 times the volume but only about 160 times the heat transfer area. Every doubling of scale makes the heat removal problem worse in a way that nothing on the bench prepares you for.

Where the Money Actually Goes

Break a cryogenic quote into its components and it looks roughly like this:

1. Cooling capacity, not vessel volume, sets batch size. A cascade chiller might deliver 15 kW at −20 °C and only 5 kW at −70 °C — cooling capacity falls sharply as you approach the bottom of the equipment’s range. If your lithiation releases 60 kJ per mole and you are running 40 moles, that is 2,400 kJ to remove. At 5 kW you need at least eight minutes of pure removal time under ideal conditions, and real addition profiles run several times longer than the thermodynamic minimum to keep the internal temperature inside a ±3 °C band.

2. Heat transfer fluid becomes the bottleneck. Silicone oils and hydrofluoroether fluids that work fine at −20 °C become viscous at −70 °C. Viscosity kills the jacket-side heat transfer coefficient, so the jacket removes less heat per degree of driving force exactly when you need it most. Low-temperature-rated fluids cost several times more per liter than standard glycol, and a large jacket loop holds hundreds of liters.

3. The addition exotherm dominates the profile. Most cryogenic steps are not slow reactions run cold; they are fast, strongly exothermic reactions held cold. The heat comes in during the reagent addition, so the addition rate is set by the cooling capacity. This is the origin of the “slow controlled addition” specification you see on process descriptions — a two-hour addition in the lab becomes a six- to twelve-hour addition at 200 L, sometimes longer.

4. Cycle time is usually the dominant cost. Add the cooldown (bringing 300 L of solvent from ambient to −75 °C is itself a multi-hour heat removal job), the extended addition, the hold, the controlled warm-up, and the quench, and a step that runs in a working day at bench scale becomes a 24 to 36 hour operation. You are paying for reactor occupancy, an operator on shift through the addition, and often a second operator for the reagent transfer. This is why the cost gap widens rather than narrows as you scale.

5. Solvent volume goes up, not down. Cryogenic steps often need higher dilution to manage the exotherm and to keep everything in solution at low temperature — lithium salts and substrates that are perfectly soluble at ambient can precipitate at −78 °C and foul the wall. More solvent means more mass to cool, more mass to warm, and more waste to dispose of. Dilution at −78 °C is a genuine second-order cost that rarely appears in bench estimates.

A Rough Order-of-Magnitude Comparison

Actual numbers vary widely by facility, campaign size and molecule, and any CRO quoting from a general table rather than from your specific step should be questioned. But the shape of the cost is consistent:

Factor	Ambient batch step	−40 °C step	−78 °C step
Typical cycle time at 100 to 200 L	8 to 12 h	16 to 24 h	24 to 48 h
Equipment class	Any jacketed vessel	Single-stage chiller	Cascade chiller, limited units
Batch size limit set by	Vessel volume	Usually vessel volume	Almost always cooling capacity
Operator coverage	Single shift	Often extended shift	Frequently two shifts
Relative cost per kilogram	1x	roughly 1.5 to 2x	roughly 2.5 to 4x

The ranges are wide on purpose. Confirm the specifics with your supplier for your step and your volume — the ratio between the last two columns is the number worth negotiating over, and it is highly facility-dependent.

Four Ways to Avoid Paying for Cryogenic Conditions

Before accepting the cost, work through these in order. The first is cheap to test and succeeds surprisingly often.

1. Screen the Temperature Window

Run the reaction at −78, −60, −40 and −20 °C in parallel and measure selectivity and impurity profile at each. This is a one-day experiment on milligram to gram scale and it is the highest-return diagnostic in the whole exercise. Two outcomes are common. Either selectivity falls off a cliff somewhere between −60 and −40 °C, which justifies the cryogenic cost with data rather than tradition, or it degrades gently — 97:3 at −78 °C becoming 94:6 at −40 °C — at which point the question becomes whether three points of selectivity are worth a doubling of cost per kilogram. Often the downstream purification absorbs the difference for far less money.

Ask explicitly whether the CRO has screened the window or inherited the number. Our post on process chemistry optimization from lab to pilot describes where this screen fits in a development sequence.

2. Flow Chemistry

A continuous flow reactor changes the heat transfer arithmetic completely. Channel dimensions on the order of a millimeter give surface-area-to-volume ratios hundreds of times higher than a stirred tank, so the heat generated by a fast exotherm is removed essentially as fast as it appears. Residence times of seconds mean an unstable intermediate is generated and consumed before it has time to decompose, which addresses the stability justification for cold independently of the selectivity one.

The practical consequence is that some chemistry needing −78 °C in batch runs acceptably at −20 to 0 °C in flow, and organolithium and halogen-metal exchange chemistry is one of the best-documented application areas. The catch is real: development cost is higher, pumping cold slurries is the standard failure mode, and lithium halide byproducts precipitating in a 1 mm channel will stop a campaign. We cover the decision framework in flow chemistry versus batch processing.

3. Change the Reagent or the Base

Sometimes the temperature requirement is really a reagent requirement in disguise.

• Turbo-Grignards and knochel-type magnesiates (isopropylmagnesium chloride with lithium chloride) perform halogen-metal exchange on many aryl halides at −20 °C to 0 °C, temperatures reachable with standard equipment. Functional group tolerance is broader than alkyllithium, which is often a second benefit.
• Zincates and magnesium amide bases (TMPMgCl·LiCl and relatives) metalate many heterocycles at or near 0 °C where the corresponding lithium amide needs −78 °C.
• Swapping LDA for LiHMDS or KHMDS changes both the basicity and the aggregation state, and sometimes shifts the workable temperature window by 20 to 30 °C at a modest selectivity cost.
• Solvent choice matters more than most bench chemists assume. 2-Methyltetrahydrofuran extends alkyllithium lifetime relative to THF, and toluene-THF mixtures often allow a warmer operating window than neat THF.

None of these is a guaranteed substitution. Each needs screening. But a reagent change that moves a step from −78 to −20 °C is worth two weeks of development on any campaign above a few kilograms.

4. Redesign the Route

The most expensive option and the most permanent. If a cryogenic step is early in a long sequence, its cost multiplies through everything downstream, and a route that reaches the same intermediate by a warm sequence — a cross-coupling instead of a lithiation, a different disconnection, a purchased building block instead of a made one — may be cheaper overall even at a lower step count penalty. This last point is frequently missed: buying a substituted building block from a catalog often beats making it through a cryogenic step, and a scan of the halogenated compounds and heterocyclic compounds catalogs before committing to a lithiation is a cheap hour. Route redesign belongs on a development budget, and scale-up challenges in custom synthesis covers when it pays for itself.

Handling and Safety at Cryogenic Scale

The cost of a cryogenic step is not only cooling. Most of the reagents that need the cold are also hazardous in ways that add operational overhead, and the safety engineering is a real fraction of the price.

Pyrophoric Reagent Transfer

t-BuLi ignites spontaneously in air. n-BuLi and s-BuLi solutions ignite readily depending on concentration and solvent. Trimethylaluminum and many organozincs behave similarly. At bench scale this is managed with cannula transfer or a gas-tight syringe under nitrogen or argon. At scale it becomes a closed-transfer engineering problem: pressure transfer from a sealed cylinder or tote through hard-piped lines, with inerted headspace, dip tubes, no glass in the transfer path, and a defined procedure for line clearing at the end. Facilities that do this routinely have dedicated pyrophoric transfer stations. Facilities that do not will either decline the work or improvise, and improvisation here is exactly what you do not want.

Moisture and Oxygen Exclusion

Organolithiums are destroyed by water and degraded by oxygen and carbon dioxide. At −78 °C the moisture problem gets worse rather than better, because a cold vessel condenses atmospheric water on any surface exposed during charging, and ice on a reactor wall is both a heat transfer barrier and a slug of quench waiting to happen. Solvent must be dried to specification and verified — Karl Fischer titration on the actual charge, not on the drum certificate. Reagent titration before use matters too: a bottle of n-BuLi that has been opened repeatedly may be well below its nominal molarity, and under-charging a lithiation gives incomplete conversion that looks like a chemistry failure.

Quench Design and Quench Exotherms

The quench is where cryogenic steps most often go wrong. You have accumulated unreacted strong base in a large volume of cold solvent, and quenching it releases a great deal of heat quickly. Add water to a cold lithiation and you can get a thermal runaway, a gas evolution event, or a freeze-out that traps unquenched material under a solid crust — which then melts and quenches all at once. Good practice is a reverse quench into a large, well-stirred, cold aqueous or acidic sink with adequate cooling capacity, an addition rate limited by measured heat release, and a defined maximum accumulation. Quench calorimetry is worth running for any new cryogenic step above a few kilograms, and it is a reasonable thing to ask a CRO whether they have done. The ACS Chemical Health and Safety resources are a useful reference point for the underlying practice.

Cold Burns and Operator Exposure

Contact with a −78 °C surface, a cryogenic fluid line, or spilled liquid nitrogen causes tissue damage quickly, and cold-numbed skin does not warn the operator. Insulated gloves rated for the temperature, face protection during dry ice charging, and adequate room ventilation matter. That last item is not optional: sublimating dry ice and boiling liquid nitrogen both displace oxygen, and a cryogenic operation in an enclosed room is an asphyxiation hazard as much as a cold one. Oxygen monitoring in the room is standard at facilities that do this work regularly. The OSHA guidance on cryogenic and compressed gas hazards and the NIOSH chemical safety resources set out the baseline expectations.

What to Ask a CRO About Cryogenic Capability

Most capability claims about cryogenic chemistry are true and useless, because they omit the variable that matters. “We can run to −78 °C” is meaningless without a vessel size attached. Ask these six questions in writing.

1. What is your minimum achievable temperature, and at what vessel size? A −90 °C capability in a 2 L jacketed reactor tells you nothing about a 200 L campaign. Ask for the pairing, and ask for it as measured internal temperature, not jacket setpoint.

2. What is your cooling capacity in kilowatts at that temperature? Chiller capacity falls off steeply near the bottom of the range. A unit rated 20 kW at 0 °C may deliver 5 kW at −70 °C. This number, not the vessel volume, sets your maximum batch size for an exothermic step.

3. What have you actually run below −60 °C at over 5 L in the last twelve months? This is the question that separates capability on a website from capability in a plant. No names or details are needed — chemistry type, scale, and roughly when is enough. A supplier who cannot answer it has probably not run cryogenic chemistry at scale recently.

4. What is your experience with pyrophoric reagents at scale? Ask specifically about closed transfer from totes or cylinders, dedicated transfer equipment, and the written procedure. Ask whether they have handled t-BuLi, which is the strictest case.

5. How do you design and test the quench? Look for calorimetry, a defined maximum accumulation, and a reverse-quench procedure with adequate cooling on the receiving vessel.

6. What in-process analytics do you run at temperature? Sampling a −78 °C reaction without warming it or letting moisture in is non-trivial. In-process HPLC or GC on a properly quenched aliquot, or in-situ IR, tells you whether the metalation is complete before you commit the electrophile. Without it you are running blind and discovering the problem at workup.

Two supporting questions are worth adding if the campaign is large. Ask about their heat transfer fluid and its rated low-temperature viscosity, and ask what their realistic cycle time is for one batch at your target scale — the second number is what you are actually buying. Our US-based custom synthesis capability map gives a broader picture of how domestic facilities distribute this and other specialized capabilities.

Putting It Together

A cryogenic line item on a quote is neither a red flag nor a given. It is a claim that a specific reaction needs a specific amount of heat removed at a specific temperature, and that claim is testable. The sequence that serves buyers best is short: ask whether the temperature was screened or inherited; if it was inherited, pay for a one-day screen before paying for a cryogenic campaign; if the cold is real, ask for cooling capacity in kilowatts at the vessel you would run in and treat that as the batch size constraint; and check whether flow or a reagent change moves the step into a warmer equipment class before committing to a multi-batch campaign at −78 °C.

The physics does not negotiate. Below about −45 °C you are in a different equipment class, with a different heat transfer fluid, a longer cycle time, and a batch size set by kilowatts rather than liters. What is negotiable is whether you actually need to be down there, and that question is answered with a screen, not with an opinion.

ChemContract Research operates US-based custom synthesis with cryogenic capability to −78 °C, contract R&D for temperature screening and route redesign, and analytical services including chiral HPLC for the selectivity measurements that make these decisions with data rather than assumption. We have run low-temperature organolithium, enolate and ozonolysis chemistry since 2000 from our Huntington Beach facility. If you are holding a quote with a cryogenic line on it, send us the step — the substrate, the electrophile, and the selectivity you need — and we will tell you within 24 hours whether we think the cold is required and what a screen would cost.

Frequently Asked Questions

Why is minus 78 C the standard cryogenic temperature in chemistry?

Because it is the temperature of a dry ice and acetone bath, which holds itself there by sublimation without any control equipment. Solid carbon dioxide sublimes at −78.5 °C at atmospheric pressure, so the bath self-regulates. It became the default laboratory cryogenic condition for reasons of convenience, not because most reactions need exactly that temperature.

Which reactions actually require cryogenic conditions?

Organolithium chemistry such as directed ortho metalation and halogen-metal exchange, kinetic enolate formation with LDA or LiHMDS, additions to unstable intermediates that must be trapped before they decompose, many diastereoselective and enantioselective additions, and ozonolysis. In each case the cold either keeps a reagent alive or widens the selectivity margin between two competing pathways.

Why does cryogenic chemistry cost so much more at scale?

Heat removal. A 250 mL flask has a huge surface-area-to-volume ratio and sits in an infinite heat sink. A 500 L reactor removes heat only through its jacket, so the exotherm of the addition must be matched by the chiller’s cooling capacity in kilowatts. That forces a slow controlled addition, which extends cycle time, and cycle time is usually the dominant cost.

Can minus 40 C replace minus 78 C?

Often yes, and it is worth screening every time. Minus 40 °C is reachable with standard commercial single-stage chillers and common heat transfer fluids, so the cost per kilogram at pilot scale can be a third of the cryogenic figure. Many organolithium additions and some enolate alkylations lose only one or two points of selectivity. Some, particularly t-BuLi work and tight kinetic enolate cases, do not survive the change.

Does flow chemistry eliminate the need for cryogenic cooling?

Not always, but it changes the arithmetic. A flow reactor has a very high surface-area-to-volume ratio and residence times measured in seconds, so heat generated by the reaction is removed almost as fast as it appears and unstable intermediates are consumed before they decompose. Some chemistry that needs −78 °C in batch runs acceptably at −20 to 0 °C in flow. Development cost is higher and pumping cold slurries is the usual failure mode.

What should I ask a CRO about cryogenic capability?

Ask for minimum achievable temperature paired with vessel size, because a −90 °C capability in a 2 L jacketed vessel says nothing about a 200 L campaign. Ask for cooling capacity in kilowatts at that temperature, experience handling pyrophoric reagents at scale, quench design, and what they have actually run below −60 °C at over 5 L in the last twelve months.