Natural Refrigerant Decision

Ammonia (NH₃) vs CO₂ (R744) — a buyer's comparison for industrial cold storage

A vendor-neutral, buyer-focused comparison of the two natural refrigerants dominating new industrial cold storage: energy efficiency (COP), safety, refrigerant cost, climate suitability and CAPEX. Use it to shortlist the right technology before you brief suppliers.

Both ammonia and CO₂ are natural refrigerants with GWP ≤ 1, neither is subject to F-Gas or Kigali phase-down, and both are the strategic long-term choice for large-scale commercial cold storage investors. The right one depends on plant size, climate zone, operator availability and the safety envelope you can accept on site.

Side-by-side matrix

CriterionAmmonia (NH₃)CO₂ (R744)
GWP01
ODP00
COP @ –10 °C SST3.2 – 4.22.6 – 3.6
CAPEX vs HFC+15–30%+20–40%
Refrigerant costUSD 3–6/kgUSD 8–15/kg
System pressure10–15 bar (typical)80–120 bar (transcritical)
Safety regulationEN 378 / ATEX / OSHA PSM (>4.5 t)Pressure vessel code
Operator qualificationCertified NH₃ engineerStandard refrigeration technician
Best size>200 kW industrial50 kW – 2 MW
Ambient sensitivityStable across ambientEfficiency drops >30 °C w/o ejector
Regulatory futureNot phased downNot phased down

Energy efficiency (COP)

Ammonia COP is 3.2–4.2 at chilled duty (–10 °C SST). CO₂ transcritical COP is 2.6–3.6, dropping sharply above 30 °C ambient without ejectors. CO₂ cascade (with NH₃ or HFC high-stage) recovers efficiency in warm climates but adds capital.

Safety requirements

Ammonia is toxic at 25 ppm and mildly flammable — every jurisdiction requires leak detection, ventilation, ammonia-safe machinery rooms and (above 4.5 t charge in the US) OSHA PSM compliance. CO₂ is non-toxic below 5,000 ppm and non-flammable, so it can sit inside occupied buildings with standard pressure-vessel controls, at the cost of an 80–120 bar system and specialist transcritical know-how.

Refrigerant cost

Ammonia is the cheapest industrial refrigerant on the market at USD 3–6/kg. CO₂ sits at USD 8–15/kg, still an order of magnitude below phasing-down HFC blends. Neither is exposed to F-Gas quotas.

Climate suitability

Ammonia holds its efficiency across ambient temperatures and is the default for hot climates (GCC, Africa, LATAM, South-East Asia) at >200 kW. CO₂ transcritical loses 15–25% COP above 30 °C ambient without ejectors or parallel compression; cascade CO₂/NH₃ or CO₂/HFC restores efficiency in warm regions.

CAPEX & plant size

CO₂ transcritical CAPEX is typically 20–40% higher than HFC baseline due to 80+ bar pressure class, ejectors, parallel compression and gas cooler. Ammonia CAPEX is 15–30% higher than HFC, driven by low-charge safety architecture, ventilation and PSM compliance. At >500 kW the two converge.

Applications & operators

CO₂ transcritical: 50 kW – 2 MW, retail cold storage, food processing, mild climates, sites without qualified NH₃ operators. Ammonia: >200 kW industrial cold storage, meat/fish/dairy plants, ice plants, ports and 3PL distribution.

Ammonia (NH₃)

Advantages
  • Highest COP of any commercial refrigerant
  • Zero GWP, zero ODP
  • Very low refrigerant cost per kg
  • Mature technology with 100+ years of industrial deployment
Limitations
  • Toxic at concentrations >25 ppm — safety architecture required
  • Flammable at 15–28% concentration in air
  • PSM compliance mandatory above 4.5 t charge (US)
  • Requires qualified operators — labour scarcity in some regions

CO₂ (R744)

Advantages
  • No toxicity, no flammability — installable in occupied buildings
  • Standard-qualified technicians can service the plant
  • Natural refrigerant with zero phase-down risk
  • Excellent heat-recovery temperatures (>90 °C hot water)
Limitations
  • Efficiency degrades sharply above 30 °C ambient without ejectors
  • High-pressure components (80+ bar) increase CAPEX
  • Cascade with NH₃ or HFC needed for large low-temp plants in hot climates

Decision guidance

Choose CO₂ transcritical if your plant is 50 kW–2 MW, in a mild climate (<32 °C ambient), or in a location without qualified NH₃ operators. Choose ammonia if your plant is >200 kW, in a hot climate, and you have (or can hire) certified refrigeration engineers. Above 1 MW in a warm climate, CO₂/NH₃ cascade or pure NH₃ typically wins on lifetime cost.

Frequently asked questions

Is CO₂ safer than ammonia?
CO₂ is non-toxic below 5,000 ppm and non-flammable, so it's easier to install in occupied buildings. Ammonia is toxic at 25 ppm and flammable — requires safety architecture (leak detection, ventilation, PSM) and certified operators.
Which is more energy-efficient?
Ammonia — COP 3.2–4.2 vs CO₂ transcritical 2.6–3.6 at chilled duty. In hot climates without ejectors the gap widens. In mild climates with modern CO₂ (parallel compression + ejectors), the gap narrows to 5–10%.
Can I retrofit an HFC plant to CO₂ or NH₃?
Retrofit is rarely economical — compressors, piping, controls, safety systems and often panels are unsuitable. Plan for like-for-like replacement at end of life, or greenfield when possible.

Deep diveSafety

Is ammonia (NH₃) safe to use inside a food factory or distribution centre?
Ammonia is toxic — the OSHA PEL is 25 ppm over 8 hours and IDLH is 300 ppm — and mildly flammable between 15–28% in air. In practice, modern industrial plants keep the ammonia charge inside a dedicated, ventilated machinery room (ATEX / IIAR-2 compliant) and distribute cold to the halls with a secondary glycol or CO₂ loop. That design pattern has been the industry default for large food plants for 40+ years, and with a low-charge packaged chiller (20–100 kg NH₃ per system) the residual risk to occupied areas is very low. Above roughly 4.5 t of charge in the US you also fall under OSHA Process Safety Management (PSM) and EPA RMP, which adds documented risk assessments, operator training and mechanical integrity programmes.
Can CO₂ (R744) really be installed inside occupied buildings without special safety infrastructure?
Largely yes, and this is CO₂'s biggest single advantage over ammonia. CO₂ is non-flammable and only becomes an asphyxiation hazard above ~40,000 ppm (4%), versus 25 ppm for NH₃ — roughly three orders of magnitude more forgiving. You still need CO₂ leak detection (typically alarming at 5,000 ppm / 8-hour TWA), pressure-relief routed outside the building, and staff awareness training, but you avoid ammonia-safe machinery rooms, evacuation zones, and PSM/RMP paperwork. That is why CO₂ dominates new supermarket, convenience-retail and urban distribution builds where the plant sits close to people.
Which refrigerant carries the lower insurance premium and permitting burden?
CO₂ almost always wins on insurance and permitting. Underwriters price NH₃ against a toxic-release scenario and typically apply a 15–35% loading on property + business-interruption cover once charge exceeds ~1 t, plus require a documented emergency response plan with local fire services. CO₂ carries no toxicity loading and, outside the EU F-Gas registration, has no substance-specific permit regime. For a mid-size (500–1,500 kW) plant this can be worth €30–120k/year in premium and 3–6 months of permitting time — a real number to weigh against CO₂'s CAPEX premium.

Deep diveEnergy efficiency (COP)

Which refrigerant delivers the higher COP for industrial cold storage?
For medium-temperature duty (–10 °C evap, 35 °C condensing) ammonia typically delivers COP 4.5–5.2 versus 3.8–4.5 for CO₂ transcritical without ejectors — a 12–18% energy advantage that grows in hot ambients. For low-temperature blast/hardening (–35 to –40 °C) the gap narrows: NH₃ COP 1.6–1.9 vs CO₂ cascade 1.5–1.8, because CO₂'s volumetric capacity at low suction pressure is actually excellent. For small (<200 kW) subcritical CO₂ booster or cascade systems in cool climates, real-world COP can match or beat NH₃ once you account for the parasitic loads of NH₃'s secondary glycol loop.
How much does ambient temperature actually change the CO₂ vs NH₃ decision?
It is the single biggest variable. CO₂'s critical point is 31 °C, so any time the condenser cannot reject heat below ~28 °C the system flips into transcritical mode and COP drops 15–25%. In Gulf, sub-Saharan Africa, South Asia or Australian summer conditions (35–45 °C peaks), a plain CO₂ transcritical plant will burn 20–30% more electricity per year than a well-designed NH₃ plant. Fixes exist — parallel compression, multi-ejector packs, adiabatic gas coolers or a CO₂/NH₃ cascade — but they add 15–25% CAPEX and specialist maintenance. NH₃'s COP is essentially flat across ambient, which is why it remains the default for hot-climate industrial cold stores >200 kW.
Do CO₂ ejectors and parallel compression close the efficiency gap with NH₃?
In moderate climates (annual average <20 °C) — yes, almost entirely. A modern CO₂ transcritical booster with multi-ejector and parallel compression will land within 3–7% of NH₃ annualised COP in Northern Europe, the UK and the US Pacific Northwest. In hot climates the gap narrows from 20–25% down to about 8–12%, which is meaningful but no longer a knock-out argument against CO₂. The trade-off is a €40–90k CAPEX adder per 100 kW of MT duty and a smaller pool of service contractors qualified to commission the ejector logic — factor both into your TCO model, not just the COP number.

Deep diveRefrigerant cost

What do NH₃ and CO₂ actually cost per kg, and does the refrigerant charge matter to the business case?
Ammonia is the cheapest industrial refrigerant on the market at roughly USD 3–6/kg delivered. CO₂ (R744) is USD 8–15/kg. For context, phasing-down HFC blends like R404A are USD 60–120/kg in EU quota-constrained markets. Neither NH₃ nor CO₂ is exposed to F-Gas quotas, Kigali phase-down or CBAM. On a 500 kW cold store the total refrigerant fill is a rounding error on CAPEX (typically <0.5%), but it becomes very material over 15–20 years of operation as HFC systems face repeated recharges at rising prices — which is the core financial argument for switching to naturals now, not later.
Which system has the lower lifetime CAPEX + OPEX (TCO) for a mid-size project?
For 200–2,000 kW industrial cold storage in temperate climates, a well-designed CO₂ transcritical plant now lands within ±5% of NH₃ on 15-year TCO once you include insurance, permitting, machinery-room build-out and the NH₃ secondary loop. Below 200 kW, CO₂ almost always wins on TCO because you avoid the fixed cost of an ammonia-safe machinery room (€80–200k) and PSM compliance. Above 2 MW and/or in hot climates, NH₃ pulls ahead by 8–15% on TCO thanks to sustained COP and cheap refrigerant. Between those bands, run a project-specific model — the answer flips on ambient, electricity tariff and operator availability, not on the refrigerant itself.
Are there subsidies, tax breaks or green-finance instruments that shift the CO₂ vs NH₃ decision?
Yes, and they can be worth 8–20% of CAPEX. In the EU, both NH₃ and CO₂ qualify under the EU Taxonomy for substantial contribution to climate change mitigation, unlocking green-loan pricing typically 50–120 bps below vanilla senior debt. Several jurisdictions layer specific natural-refrigerant grants: US EPA GreenChill and IRA Section 179D, Canada NRCan RETScreen, Germany's Kälte-Klima-Richtlinie (up to 35% of eligible CAPEX for natural refrigerants), and the UK Industrial Energy Transformation Fund. HFC/HFO systems either don't qualify or qualify at a much lower rate. Model the after-subsidy CAPEX, not the sticker price, before you finalise the refrigerant decision — the incentive stack often turns a marginal CO₂/NH₃ business case into a clear yes.
Refrigerants hub
Explore all industrial refrigerant buyer guides →

Compare NH₃, CO₂, HFC and HFO options, phase-out timelines and refrigerant-adjacent design decisions in one place.

Related guides & tools

Pair your NH₃ / CO₂ decision with the applications, equipment and sizing tools that most often determine project outcomes.

Ready to brief suppliers?

Convert your NH₃ / CO₂ decision into a vendor-neutral RFQ with capacity, temperature class, refrigerant and delivery terms pre-filled.

Build my RFQ
Get Free QuotesFind the Best Solution