Why Roman Concrete Outlasted
Everything We've Built Since
The Pantheon's unreinforced concrete dome has stood for 1,900 years. The concrete seawall at Caesarea Maritima, built by Herod the Great around 37 BCE, has been submerged in the Mediterranean for two millennia and is still structurally intact. Meanwhile, modern reinforced concrete structures in marine environments begin to show cracking, spalling, and rebar corrosion within decades. In some coastal regions, reinforced concrete seawalls built in the 1970s have required replacement or major repair. The Romans were building more durable marine structures with less sophisticated chemistry, no rebar, and no computer modeling. How?
The answer was discovered only in the past decade through synchrotron X-ray and electron microscopy studies of ancient concrete samples. It involves a volcanic ash from a specific region near Naples, a reaction chemistry that modern Portland cement doesn't replicate, and a mechanism that not only resists damage but actively heals it over time. Roman concrete isn't just a historical curiosity. It's a material whose chemistry we're now trying to deliberately reproduce — because it points toward concrete that's both more durable and dramatically lower in carbon emissions than what we currently use.
"The Romans accidentally invented a self-healing material. We're only now figuring out how it works — and whether we can make it deliberately."
Portland cement and why it fails in seawater
Modern concrete uses Portland cement — invented by Joseph Aspdin in 1824 and refined through the 19th century. Portland cement is made by heating limestone and clay to about 1,450°C in a kiln — a process called calcination that releases large amounts of CO₂ and produces calcium silicate compounds. When mixed with water, these compounds undergo hydration reactions that produce the interlocking calcium silicate hydrate (C-S-H) crystals that bind aggregate together and give concrete its strength. The process is well understood, highly engineered, and produces a material with excellent initial mechanical properties.
The problem in seawater is a reaction called ettringite formation. Seawater contains sulfates and chlorides. The sulfates react with the aluminates in Portland cement to form ettringite — a hydrated calcium aluminum sulfate mineral that expands as it grows, cracking the concrete from within. Simultaneously, chlorides penetrate through the concrete matrix and corrode the steel reinforcement inside, causing rust expansion that spalls the surrounding concrete. Portland cement concrete in seawater is in a constant state of slow chemical attack, and the higher the exposure (wave splash zones are worst), the faster the degradation.
The Roman architect and engineer Vitruvius described the recipe for hydraulic concrete in his treatise De Architectura (c. 30 BCE) with remarkable precision. He specified the use of pozzolana — volcanic ash from the region of Pozzuoli near Naples — mixed with lime and seawater. He noted that the mixture "hardens under water and increases in strength with time." He was describing the correct recipe for what we now recognize as pozzolanic concrete, and he understood its key property — that it didn't just resist seawater but actually improved in it. The recipe sat in his text for 2,000 years. Modern materials scientists are only now fully explaining why it works.
The volcanic secret — pozzolanic chemistry
Roman maritime concrete used a specific volcanic ash called harena fossicia — from deposits near Pozzuoli, which sits in the Campi Flegrei caldera near Naples. This ash is rich in aluminum-rich silica minerals that don't behave like the compounds in Portland cement. When mixed with lime (calcium hydroxide) and seawater, they undergo a pozzolanic reaction — they react with the calcium hydroxide produced during lime hydration to form additional calcium-aluminum silicate hydrate (C-A-S-H) minerals. This C-A-S-H is denser and less porous than the C-S-H formed in Portland cement.
But the truly remarkable finding came from analysis of ancient concrete samples drilled from Roman harbor structures. The samples contained tobermorite — a rare, extremely stable calcium silicate mineral. In Portland cement, tobermorite is almost never found because the chemistry doesn't favor its formation at ordinary temperatures. In the Roman concrete, seawater percolating through the material was reacting with the volcanic ash over centuries to grow tobermorite crystals — filling microcracks, reinforcing grain boundaries, and strengthening the material from within. The seawater wasn't attacking the concrete. It was helping it.
There's a further dimension. Roman concrete also contains large aggregate pieces — chunks of volcanic rock (tuff or lava) as large as fist-sized — embedded in the cement paste. This aggregate wasn't just filler: the volcanic rock itself continued to react slowly with the surrounding paste over decades and centuries, releasing more reactive silica that contributed to continued tobermorite formation. The entire structure was designed — probably empirically, without understanding the mechanism — to keep reacting long after it was poured.
Portland cement concrete gains most of its strength in the first 28 days and then plateaus — reaching perhaps 110–120% of its 28-day strength over the following years. Roman pozzolanic concrete appears to have gained strength for centuries. The tobermorite crystals identified in 2,000-year-old samples are more ordered and more developed than any that could have formed in decades — they represent the product of continuous slow reaction. This is the opposite of how we currently design concrete, which is treated as a static material that achieves its final properties quickly and then degrades. The Roman approach was, unintentionally, a self-improving system.
The carbon problem and why this matters now
Portland cement production accounts for approximately 8% of global CO₂ emissions. The calcination reaction — CaCO₃ → CaO + CO₂ — is thermodynamically unavoidable with Portland cement chemistry. You cannot make the binder without releasing CO₂ from limestone. The world produces about 4 billion tonnes of Portland cement per year, and demand is growing as developing countries build infrastructure. Finding a cement chemistry that is both durable and lower in embodied carbon is one of the most important materials engineering problems of the current century.
Pozzolanic cements offer a partial solution. Supplementary cementitious materials (SCMs) — fly ash from coal combustion, slag from steel production, natural pozzolans from volcanic deposits — can replace 30–50% of Portland cement in concrete mixes, reducing CO₂ per tonne of concrete substantially. Some high-performance blended cements replace 70–80% of Portland cement. Geopolymer cements — aluminosilicate binders made entirely from industrial waste streams with no Portland cement — can achieve compressive strengths of 80+ MPa with CO₂ emissions 40–80% lower than Portland cement. The Roman pozzolanic recipe, refined over two millennia of research, is one of the key reference points for designing lower-carbon alternatives.
🤔 If Roman concrete is so good, why did we switch to Portland cement in the 19th century?
▼Several reasons. First, Portland cement is much stronger in early-age strength — it gains substantial strength in 24–72 hours, making it suitable for falsework removal and rapid construction cycles that modern schedules demand. Roman concrete gains strength slowly. Second, reinforced concrete — the dominant structural form of the 20th century — specifically requires Portland cement because the alkalinity it maintains (pH ~12.5) passivates steel reinforcement, preventing corrosion. Most pozzolanic cements are less alkaline, which means rebar in them corrodes more readily. Third, Portland cement can be made from widely available limestone and clay anywhere in the world; specific volcanic ash deposits for pozzolanic concrete are geographically limited. Fourth, when Portland cement was invented, the Roman concrete examples were admired but not scientifically understood — the chemistry wasn't analyzed until very recently. Engineers adopted the material that worked faster and more predictably by 19th-century standards, without realizing what they were trading away in long-term durability.
🤔 Can we now reproduce Roman concrete deliberately — has anyone actually made it?
▼Researchers at UC Berkeley and Lawrence Berkeley National Laboratory who identified the tobermorite mechanism have been working on formulations since the initial publications in 2013 and 2017. The challenge is that the specific volcanic ash from the Campi Flegrei region has particular chemical properties — aluminum content, glass content, reactivity — that aren't replicated by all pozzolanic materials. However, the general approach — using highly reactive aluminosilicate volcanic ashes or synthetic equivalents in combination with lime and seawater — has produced laboratory samples that show tobermorite formation. Industrial-scale production is a different challenge: seawater curing isn't practical for most construction, and finding volcanic ash deposits with the right chemistry at sufficient scale requires geological surveys. More practically, the research has informed the development of more reactive supplementary cementitious materials and blended cement formulations. The Roman concrete won't be commercially reproduced exactly, but it's actively informing the next generation of durable, lower-carbon marine concretes.
Roman vs. Modern Concrete
Drag to arrange these properties from Roman concrete (top) to modern Portland failure (bottom).
- Resists seawater attack through continued mineral reactions
- Uses volcanic ash (pozzolana) as the reactive ingredient
- Uses Portland cement — loses strength via ettringite in seawater
- Gains strength over centuries via tobermorite crystal growth