If you have spent any time in the HVACR industry, you know the template.
High GWP. Montreal Protocol. Kigali Amendment. Production quotas. Phase-down. Alternatives emerge. Prices rise. The molecule becomes a footnote.
This template has applied to R12, R22, R502, and—more recently—R404A and R410A. It is a clean, linear narrative. Regulation drives substitution. Substitution drives obsolescence. Obsolescence drives replacement.
Then there is **R23**.
CAS 75-46-7. Trifluoromethane. Fluoroform. Molecular weight 70.01. Boiling point -82.1°C . GWP 14,600—actually 14,800, depending on whose AR4 or AR5 calculation you trust . One kilogram of R23 is the atmospheric equivalent of burning 700 gallons of gasoline.
By every environmental metric, this molecule should be extinct by now. It has zero ODP, so the original Montreal Protocol left it untouched. But the Kigali Amendment and the EU F-Gas Regulation put it squarely in the crosshairs. A refrigerant with GWP exceeding 14,000 has no business surviving the 2020s.
Yet here we are, in early 2026. R23 is still manufactured. Still specified. Still shipped in bright green cylinders with UN1984 markings . Still the baseline against which new ultra-low temperature refrigerants are measured .
Why?
Because R23 does not cool your house. It does not cool your supermarket freezer case. It does something far more specialized and far less replaceable: it cools the machines that make the modern world possible.
Part I: The Anatomy of CAS 75-46-7
Before we discuss the regulatory siege and the alternative refrigerants attempting to break it, we need to understand what R23 actually is.
Chemically, it is simple. One carbon. Three fluorines. One hydrogen. CHF₃. The molecule is nearly spherical, highly symmetric, and extraordinarily stable . This stability is what makes it both useful and problematic.
R23 is not a high-pressure refrigerant in the conventional sense. Its critical temperature is **25.9°C** . That means above room temperature—above about 78°F—R23 cannot condense to liquid no matter how much you compress it. This is not a bug; it is the defining feature.
In a standard vapor-compression cycle, a refrigerant condenses at ambient temperature, sheds its latent heat, expands, and evaporates at low temperature to absorb heat. For a refrigerant to reach **-70°C** or **-80°C** evaporation temperatures, its critical temperature must be low enough that it does not "pinch" the heat exchanger approach temperatures. R23’s critical point is deliberately low. It is engineered to operate in the **low-temperature stage of cascade systems**, where its job is not to reject heat to ambient air, but to reject heat to a higher-temperature refrigerant circuit—typically R404A, R448A, or now R449A .
This is the cascade architecture: two refrigeration cycles stacked on top of each other. The top stage runs a conventional refrigerant (R448A, R134a, R404A) and operates at typical commercial temperatures. The bottom stage runs R23 and pulls the temperature down to -70°C or below.
Without this two-stage cascade, there is no reliable, non-cryogenic method to maintain **-80°C** in a mechanical system.
Now, look at the physical constants :
- Boiling point at atmospheric pressure: -82.1°C. That is colder than the Antarctic winter.
- Vapor pressure at 25°C: 35,300 mm Hg—about 46 atmospheres. The cylinder pressure is not an accident; it is the vapor pressure of the liquid at room temperature.
- Specific volume: 0.35 m³/kg. Dense vapor. You store 8 kg in a 10-liter cylinder .
- Relative density: 2.43 times heavier than air . Released R23 does not rise; it pools in pits, trenches, and low spots, displacing oxygen.
The safety data sheets note two primary hazards: **frostbite** from rapid expansion, and **suffocation** from oxygen displacement . There is no acute toxicity—the molecule itself is not poisonous—but it is an asphyxiant. And under fire conditions, it decomposes into hydrogen fluoride, which is absolutely poisonous.
This is the molecule we are discussing. It is not friendly. It is not forgiving. It is a precision tool for a specific thermal problem.
Part II: The Number That Cannot Be Ignored—14,600
Let me state the environmental arithmetic plainly.
**R23 has a GWP of 14,600** .
The European F-Gas Regulation, as revised in 2024, does not immediately ban R23. It cannot—there are no commercially viable A1 alternatives at scale, and the medical and semiconductor supply chains depend on installed ULT equipment . But the regulation creates an unbearable economic trajectory.
Under the F-Gas quota system, high-GWP refrigerants become progressively more expensive as the annual CO₂-equivalent allowance declines. R23 consumes 14.6 tons of CO₂ allowance per kilogram placed on the market. When the quota shrinks, the price of R23 does not rise linearly; it rises asymptotically.
The equipment owners who depend on -80°C freezers—biobanks, vaccine distributors, semiconductor fabs—are not price-sensitive in the conventional sense. If a freezer fails and the only refrigerant available costs ten times what it did last year, they pay it. But the cumulative pressure is unsustainable.
This is why the refrigerant manufacturers and test chamber builders have been racing, for the last eight years, to develop **low-GWP A1 replacements** for R23 .
Part III: The Alternatives—What Works, What Doesn‘t, and What the 2025 Research Reveals
In August 2025, the *Energies* journal published what is currently the most rigorous thermodynamic evaluation of R23 alternatives . The research team at Universitat Jaume I in Spain, part of the ISTENER group, performed a comprehensive modeling study of four CO₂-based mixtures:
- **R469A** (GWP 1,357)
- **R472B** (GWP 526)
- **R472A** (GWP 353)
- **R473A** (GWP 1,830)
All are classified **A1**—non-flammable, non-toxic under normal handling. This is critical. Previous proposed alternatives to R23, such as R170 (ethane) and R290 (propane), have excellent thermodynamic performance but carry **A3 flammability** classifications . In a laboratory environment, where freezers are often located in occupied spaces, near electronic instrumentation, and subject to continuous operation, A3 refrigerants are non-starters for many users.
The researchers modeled these four mixtures in a two-stage cascade configuration with R448A in the high-temperature stage. They examined three evaporation temperatures: -50°C, -60°C, and -70°C. They optimized the cascade temperature for each condition. And they compared two internal heat exchanger configurations—suction-discharge (SDHX) and suction-liquid (SLHX)—against the basic cycle.
The results are sobering.
**No alternative matches R23’s coefficient of performance.**
R473A came closest, with COP **0.74% to 1.26% lower** than R23 . R469A, which is already commercialized by Weiss Technik under the trade name WT69, showed a COP penalty in the range of 2-5% depending on operating conditions .
R472A and R472B, despite their dramatically lower GWP (353 and 526, respectively), exhibited larger efficiency penalties. And all alternatives required **optimization of the cascade temperature**. You cannot simply dump R469A into an R23 system and expect identical performance.
The study also confirmed something that experienced service technicians already suspected: **internal heat exchangers reduce COP** in ULT cascade systems . They are necessary for oil return—compressor oil becomes treacly below -40°C—but they impose an efficiency penalty. R472B, with its high temperature glide, was the least negatively affected. But the penalty was still present.
What does this mean for the equipment owner?
It means the transition away from R23 is not a simple drop-in. It is a system-level re-engineering challenge. The new refrigerants work. They reduce CO₂-equivalent emissions by **up to 95%** . But they require optimized heat exchanger sizing, adjusted expansion devices, and—in many cases—re-commissioning of the entire refrigeration system.
Part IV: The Commercial Reality—WT69/R469A and the Retrofit Path
The most mature R23 alternative on the market, as of early 2026, is **WT69**, also designated **R469A** .
Weiss Technik, a German manufacturer of environmental simulation chambers, developed this refrigerant in partnership with chemical suppliers. They have been testing it since approximately 2020—over 200,000 operating hours across more than 50 system configurations .
The specifications are publicly documented:
- **GWP: 1,357** (91% reduction from R23)
- **ASHRAE classification: A1** (non-flammable, non-toxic)
- **Temperature capability: -70°C**
- **Compatibility:** Retrofittable to existing chambers via WT77 (a related blend)
Weiss Technik has received multiple innovation awards for this development, including the German Innovation Award gold prize in the chemical industry category . This is not marketing fluff; the German Design Council does not hand out gold awards for incremental tweaks. R469A represents a genuine breakthrough.
But there are constraints.
First, **R469A cannot reach -80°C**. Its lower practical limit is approximately -70°C. For applications requiring -80°C storage—certain biological sample archives, some aerospace testing protocols—R23 remains the only A1 option.
Second, **retrofit is possible but not trivial**. Weiss Technik offers retrofit kits for their own chambers using WT77, a related blend . For third-party equipment, the suitability depends on compressor displacement, heat exchanger surface area, and control logic. There is no universal conversion chart.
Third, **adoption is uneven**. The pharmaceutical industry, under intense regulatory and investor pressure to reduce carbon footprints, is moving aggressively to R469A and similar blends. The semiconductor industry, which uses R23 not primarily for refrigeration but for **plasma etching**, is a separate case entirely.
Part V: The Other R23—Semiconductors and the Etch Process
If you search CAS 75-46-7 in semiconductor procurement databases, you will not find it listed as a refrigerant.
You will find it listed as **electronic grade trifluoromethane** .
Here is a fact that surprises most HVAC professionals: a substantial fraction of the global R23 supply never enters a refrigeration compressor. It is consumed in **dielectric barrier discharge** and **capacitively coupled plasma** reactors, where it dissociates into CF₂⁺, F⁻, and other radical species that etch silicon dioxide with extraordinary precision .
The semiconductor industry values R23 for three properties:
- **Selectivity**: CHF₃-based plasmas etch SiO₂ much faster than silicon, enabling self-aligned contact etches.
- **Stability**: The gas can be stored, purified, and delivered with extreme consistency.
- **Fluorine supply**: It delivers fluorine radicals without the extreme global warming potential of NF₃ or SF₆.
This application has no direct substitute. The industry has experimented with C₄F₈, C₄F₆, and CH₂F₂, but each requires different power settings, pressure regimes, and hardware configurations. A chip fabrication line running 50,000 wafers per month cannot switch etch chemistry overnight.
Moreover, the environmental calculus is different. In refrigeration, R23 is a working fluid that eventually leaks or is vented. In semiconductor manufacturing, R23 is **process-consumed**. The vast majority of the gas fed into an etch chamber is dissociated in the plasma and pumped out as non-GHG fragments or captured in abatement systems. The CO₂-equivalent emissions per wafer are not zero, but they are orders of magnitude lower than a leaking R23 freezer.
This dual-use profile complicates every regulatory discussion. A flat ban on R23 production would crinot just the ultra-low temperature refrigeration sector, but also a critical node of the semiconductor supply chain.
Part VI: The 2026 Quota Landscape—What the Chinese Numbers Actually Mean
On December 9, 2025, China’s Ministry of Ecology and Environment published the **2026 HFC production and consumption quotas** .
Most of the commentary on this document has focused on R32, R134a, and R143a—the high-volume refrigerants. R23 is mentioned only in a footnote: the production quotas reported include "R41 and R23" in the aggregated totals for major manufacturers .
But the silence is itself informative.
China is the world‘s largest producer of R23, both as a primary product and as a byproduct of HCFC-22 manufacturing. Under the Kigali Amendment, China has committed to phased reductions in HFC production. Yet the 2026 quota document shows no specific, separate quota line for R23. It is subsumed into the "other HFCs" category.
What does this mean for supply?
It means R23 production is not explicitly capped, but it is **implicitly constrained** by the overall HFC quota cap and the flexibility rules. Manufacturers can shift up to 30% of their quota between HFC species in 2026 . This flexibility could theoretically allow increased R23 production if market signals justified it.
But the market signals do not justify it.
R23 demand in the refrigeration sector is flat-to-declining as R469A and R473A gain acceptance. R23 demand in semiconductor etching is stable but not growing explosively. And the CO₂-equivalent cost of placing R23 on the market under European quota is becoming prohibitive.
The rational producer allocates quota to R32, R134a, or R125—products with growing demand, lower regulatory headwinds, and better margins. R23 becomes a niche product, supplied to a shrinking installed base and a captive semiconductor customer.
This is the real 2026 story. R23 is not banned. It is not extinct. It is simply **no longer a growth business**.
Part VII: The Technician’s Problem—Legacy Equipment and Unfamiliar Cylinders
If you are a service technician who has spent your career working on supermarket racks and rooftop units, R23 has never crossed your bench.
If you work in **laboratory refrigeration**, **pharmaceutical storage**, or **environmental test chambers**, R23 is your daily reality. And that reality is becoming more difficult.
Consider the cylinder. Air Liquide supplies R23 in a steel cylinder with a **bright green shoulder (RAL 6018)** and a **DIN 477 No. 6 valve outlet** . This is not the same fitting as a standard refrigerant cylinder. If you do not have the correct adapter, you do not open that valve.
Consider the pressure. At room temperature, a full R23 cylinder is at approximately 41.8 bar—over 600 psi . This is higher than R410A, higher than R32, higher than almost any refrigerant you routinely handle except CO₂.
Consider the retrofits. Weiss Technik offers WT77 for older chambers, but the conversion requires replacing the expansion device, adjusting the charge, and validating the controller parameters . There is no generic "R23 to R469A" conversion procedure.
Consider the supply chain. As major distributors reduce R23 inventory in favor of low-GWP alternatives, procurement lead times extend. The cylinder that used to ship next day now ships in three weeks—if it is in stock at all.
The equipment owners do not see this. They see a freezer that is blinking a high-temperature alarm. They call you. You are expected to fix it.
This is the hidden cost of the refrigerant transition. The regulatory paperwork happens in Brussels and Montreal and Beijing. The operational burden lands on the technician with a manifold gauge set.
Part VIII: The Carbon Math That Won‘t Go Away
The environmental case against R23 is not abstract.
A single 8 kg cylinder of R23 contains, in its latent global warming potential, the equivalent of **117 metric tons of CO₂** . That is the annual emissions of 25 passenger vehicles. It is the carbon footprint of heating an average European home for 23 years.
Every kilogram of R23 that leaks from a cascade system—and ULT systems leak, because they operate under extreme pressure differentials and temperature swings—is a climate event.
This is why the European F-Gas Regulation, despite granting R23 a temporary reprieve, has created an economic structure that makes continued use increasingly untenable. The quota price for high-GWP refrigerants has risen every year since 2018. It will continue to rise.
The Total Equivalent Warming Impact (TEWI) methodology used in the 2025 Energies study accounts for both direct emissions (refrigerant leakage) and indirect emissions (energy consumption) . The alternative refrigerants, despite their slight COP penalty, achieve dramatic TEWI reductions because their direct emissions contribution is negligible.
This is the arithmetic that finally convinces facility managers. The efficiency gap between R23 and R469A is 1-2%. The GWP gap is 91%. Over a ten-year equipment lifetime, the carbon footprint advantage of the alternative is overwhelming.
Part IX: The Uncomfortable Question—Will R23 Ever Disappear?
I have been writing about refrigerants for fifteen years. I have watched R12 go from ubiquitous to contraband. I have watched R22 survive on reclaimed gas and dwindling stockpiles. I have watched R404A retreat from commercial refrigeration under the weight of its GWP.
R23 is different.
It will never be banned in the sense that R12 was banned, because there is no drop-in replacement that delivers -80°C capability with A1 safety classification. R469A stops at -70°C. R473A comes closer but still cannot match R23’s extreme low-temperature performance . The A3 flammables—R170, R290—can reach these temperatures, but they carry explosion risk that many users cannot accept.
So R23 persists. Not as a growth product. Not as a first-line specification for new equipment. But as the irreducible core of an installed base that will take another decade to retire.
The semiconductor etch application is even more persistent. There is no regulatory timeline for eliminating CHF₃ from chip manufacturing. The industry will reduce emissions through abatement and process optimization, but the molecule itself remains chemically optimal for certain dielectric etches.
CAS 75-46-7 will be manufactured, distributed, and consumed for the foreseeable future. It will simply become more expensive, more specialized, and less visible.
Conclusion: The Refrigerant That Doesn‘t Follow the Rules
There is a natural temptation to view every high-GWP refrigerant through the same lens. Regulation arrives. Industry panics. Alternatives emerge. The old refrigerant fades into history.
R23 resists this narrative.
It resists because its unique physical properties—that impossibly low boiling point, that sub-ambient critical temperature—are not easily replicated. It resists because it serves not one industry but two, and the second industry has no ready substitute. It resists because the A1 alternatives, while genuine achievements, cannot yet reach the deepest temperatures that certain applications require.
The future of R23 is not extinction. It is **stratification**.
New ultra-low temperature equipment will increasingly ship with R469A, R473A, or next-generation blends not yet commercialized. These systems will dominate the -70°C to -75°C range. They will satisfy 80% of the market demand.
The remaining 20%—the -80°C biological archives, the extreme thermal cycling chambers, the custom industrial cascade systems—will continue to use R23. They will pay premium prices. They will manage their leak rates obsessively. They will reclaim and recycle rather than vent. They will operate under a regulatory microscope.
And the semiconductor fabs will keep ordering electronic-grade trifluoromethane, keep feeding it into plasma reactors, keep etching the tiny features that make your phone work. Their abatement systems will capture and destroy the effluent. The net emissions will be small relative to the value created.
This is not the clean, linear phase-out that environmental advocates envisioned. It is messier, more complicated, and more honest about the trade-offs between climate protection and technological capability.
CAS 75-46-7 is not a villain. It is a molecule that does one thing extraordinarily well, and we have not yet figured out how to do that thing without it.
The cylinder with the green shoulder will remain on loading docks and in lab equipment rooms for years to come. It will be handled less frequently, specified more reluctantly, and priced more expensively. But it will remain.
R23 is the refrigerant that refused to follow the standard phase-out script.
And that refusal tells us something important about the limits of regulation, the inertia of industrial infrastructure, and the irreducible specificity of certain chemical problems.
We are not replacing R23 because we want to. We are replacing it because we must. And the process is taking longer than anyone expected.