A short history of binders
Nine thousand years of trying to turn loose stone into structure, and one discovery in the 1840s that has never been superseded. The useful part of this history is not what changed — it is what didn't, because that is what tells you which parts of a modern plant are open to improvement and which are chemistry you cannot argue with.
On this page · 15 sections
- 01The one idea
- 02Nine thousand years, briefly
- 03First principles — why 1,450 °C
- 04What must be true
- 05The quantitative anatomy — how much is already gone
- 06What goes wrong here — the three claims everyone gets wrong
- 07Scenarios from the field
- 08The numbers that matter
- 09Build the model — how much is left
- 10Upgrade paths — what the history rules out
- 11The frontier, and who is running it
- 12What's changing now
- 13Check yourself
- 14Go deeper
- 15Carry forward
§ 01 The one idea
Every builder before 1824 faced the same wall. You could make a binder that set hard, or you could make one that set under water, or one that set in the middle of a thick wall — but not all three. The Romans got closest and then the knowledge was lost for a thousand years.
What broke the wall was not a recipe. It was the discovery that you have to burn the mix hot enough to partly melt it. That single insight — burn to the point of incipient fusion — is what separates Portland cement from everything before it, and it is why a modern kiln runs at 1,450 °C rather than the 900 °C that calcination alone requires. Everything else in this course is engineering built on top of it.
§ 02 Nine thousand years, briefly
| When | What | Why it mattered |
|---|---|---|
| ~7000 BCE | Lime concrete floors at Yiftahel, Galilee | The oldest known deliberate use of a burnt-lime binder. The chemistry of the lime cycle was in use nine thousand years before anyone wrote it down. |
| ~500 BCE | Greeks use Santorin earth — volcanic ash — with lime | The first hydraulic binder, arrived at empirically. |
| ~100 BCE onward | Roman opus caementicium with pozzolana from Pozzuoli | Volcanic ash reacts with lime and water to form calcium silicate hydrates directly — a genuinely hydraulic binder, and mechanically the ancestor of modern blended cement. |
| ~126 CE | The Pantheon dome completed | Still the world's largest unreinforced concrete dome, with graded aggregate — heavy tuff at the base, light pumice at the crown. Nineteen centuries standing. |
| 5th–17th C | The knowledge is lost | Europe reverts to air lime. Medieval mortars are weak, slow and non-hydraulic, which is why medieval marine works barely exist. |
| 1756 | Smeaton, Eddystone Lighthouse | Needed a mortar that would set under water on a wave-swept rock. Established that hydraulicity comes from the clay content of the limestone, not its purity — the counter-intuitive finding, and the first appearance of raw-mix design as a discipline. |
| ~1796 | Parker's "Roman cement" | A natural cement from septaria nodules. Fast-setting, hydraulic, commercially successful — and not Roman. |
| 21 Oct 1824 | Joseph Aspdin patents "Portland cement" | Named for its resemblance when hardened to Portland stone. See §06 — the product he patented was not Portland cement in the modern sense. |
| 1841 onward | William Aspdin sells a demonstrably stronger product | The first cement containing meaningful alite. He sheltered it under his father's patent to avoid disclosing how. |
| 1885 | Ransome patents the rotary kiln | Specified producer gas and fine rawmix. Commercially unsuccessful as filed. |
| 1895 | Hurry and Seaman patent coal firing and clinker cooling with heat recuperation | The two things Ransome lacked. This is the template every kiln since has followed — including yours. |
| 1901 | Rotary kilns operating in Britain | 20–30 t/day at 9–10 MJ/kg. Worse on fuel than a good shaft kiln. They won on throughput and consistency — the same reason every kiln change since has won. |
| 1953 | First multi-stage suspension preheater, Beckum | Humboldt, now KHD. The single largest efficiency step in the industry's history. |
| early 1970s | The precalciner, developed in Japan | Moves calcination out of the kiln. Triples output for the same kiln length. |
§ 03 First principles — why 1,450 °C
Here is the question that unlocks the whole history: if calcination happens at 850–900 °C, why does a kiln run at 1,450?
Because calcination only gets you the ingredients. Turning them into cement needs a second, much harder step, and that step needs a liquid.
Liquid phase is a design variable, not an accident. Modern clinker is engineered to have 20–25% liquid at 1,450 °C, and the alumina and iron in the raw mix are there largely to produce it. That is what the alumina modulus in lesson A3 controls. Too little liquid and alite will not form; too much and the kiln rings up.
This is why electrification is hard. Every 2026 decarbonisation route has to deliver 1,450 °C into a moving bed of solids, not just heat to 900 °C. Electrified calcination — the Coolbrook and Leilac routes in §12 — attacks the calcination step, which is the easier one. Nobody has electrified the burning zone at scale.
§ 04 What must be true
This lesson’s whole argument is that the useful part of the history is what did not change. So this section is the list. Each of these is a fact about chemistry or thermodynamics, not about engineering practice, and no plant anywhere will ever beat any of them.
| What must be true | Why it cannot change | The number | What it rules out |
|---|---|---|---|
| 1. The carbonate must be decomposed | CaCO₃ → CaO + CO₂ is an endothermic reaction with a fixed enthalpy. No catalyst, fuel or furnace design reduces it. Roughly 60% of a plant’s CO₂ is this reaction, not combustion. | ~1,760 kJ per kg CaO |
Every claim that a kiln technology can eliminate process CO₂. Only capture, or using less clinker, can. |
| 2. Alite forms around 1,450 °C and not below | C₃S is the phase that gives early strength, and it only forms in the presence of a melt, which requires that temperature. This is a phase diagram, settled before 1920. | 1,450 °C | Every “low-temperature clinker” proposal that still claims ordinary Portland performance. A different binder is possible; a cheaper alite is not. |
| 3. There is a thermodynamic floor on heat input | The theoretical minimum heat to make clinker from a typical raw mix is fixed by the reaction enthalpies. | 1.59–1.84 GJ/t clinker |
Any claimed heat rate below it. Best practice today is around 2.8–3.0 GJ/t, so the remaining theoretical headroom is about 40% and the remaining practical headroom is a small fraction of that. |
| 4. The heat must be recovered counter-currently or it is lost | Gas leaves the kiln hot and clinker leaves it hot. Every efficiency gain since 1950 is a different way of putting that heat back into the incoming material. | — | Any efficiency proposal that does not identify which heat stream it recovers. If it cannot name one, it is not an efficiency proposal. |
| 5. It takes about 1.5 tonnes of raw material to make a tonne of clinker | The CO₂ leaves as gas. The mass that goes in must exceed the mass that comes out by exactly the carbon dioxide and the water. | ~1.5 : 1 | Any quarry life, haulage or mining cost calculation that uses a 1:1 ratio. This single ratio is why a cement plant is built on a quarry rather than near a market. |
| 6. The product is a hydraulic binder, so it must be kept dry and used fresh | Cement reacts with atmospheric moisture. This is why cement is not stockpiled, not traded far, and not held as inventory in the way other commodities are. | — | Any strategy that treats cement as a storable commodity, and any import-threat analysis that ignores shelf life. |
These six are the ones to reach for when someone proposes something that cannot work. The most common version is a proposal that implicitly violates row 1 or row 3 — a claimed heat rate below the thermodynamic floor, or a process change presented as eliminating emissions that are chemistry rather than combustion.
The opposite failure is just as common and more expensive: treating something as fixed when it is not. Nothing on this list fixes the clinker factor, the fuel mix, the deviation of the kiln feed, or the electrical consumption. Those are all open, and they are where the money is. The point of knowing the floor is to stop arguing about it and go and work on the things above it.
§ 05 The quantitative anatomy — how much is already gone
This is the most useful thing in the lesson. Plot the industry's specific heat consumption from the first rotary kilns to today, against the thermodynamic floor, and the shape of the remaining opportunity becomes obvious.
Moving from the wet process to a long dry kiln saved about 420 kcal/kg in one step. That single change is larger than the entire remaining gap between India's best plant and the thermodynamic floor (259 kcal/kg) — and larger still than anything actually achievable, since much of that 259 is exhaust already reused.
Why was wet→dry worth so much? Because evaporating the slurry water consumes about 2,145 kJ per kg of clinker — more than the entire theoretical heat of reaction of 1,719 kJ/kg. A wet kiln spends more energy boiling water than it does making cement. Once you know that number, the whole 20th-century history of the industry explains itself.
§ 06 What goes wrong here — the three claims everyone gets wrong
Cement history is unusually infested with confident errors, and all three of these will come up. Getting them right is cheap credibility; getting them wrong in front of a technical audience is expensive.
| The common claim | What the record actually supports |
|---|---|
| "Joseph Aspdin invented Portland cement in 1824." | He patented and named it. The product he patented was not Portland cement in the modern sense — his double-burning process produced mainly belite and calcium aluminates, behaving like Roman cement, because his kiln never reached sintering temperature. William Aspdin, from about 1841, first sold a cement with meaningful alite, while sheltering under his father's patent to avoid disclosure. Isaac Charles Johnson independently established the role of hard-burnt clinker and is often given more credit than the record supports. |
| "Roman concrete is stronger than modern concrete." | It is not, and by a wide margin. Roman concrete is far weaker in compression. Its claim is durability and longevity at low strength, in unreinforced compression-only forms — no steel to corrode, no tension to carry. Comparing it to structural concrete is comparing two different products. |
| "Seawater / volcanic ash is what makes Roman concrete self-heal." | Neither, on the current evidence. The 2023 Science Advances work attributes self-healing to lime clasts — bright millimetre-scale inclusions long dismissed as sloppy mixing — produced by hot mixing with quicklime rather than pre-slaked lime. They act as calcium reservoirs: a crack lets water in, calcium dissolves and re-precipitates as calcite, and the crack seals. Pozzolana gives strength and hydraulicity; the clasts do the repair. The seawater work is a separate line of research on marine structures. |
And one that is nearly right. "We have lost the Roman recipe" — we have not. The 2023 team reproduced the effect in modern specimens, where lime-clast concrete stopped water flow through a crack completely within 30 days while controls kept leaking.
§ 07 Scenarios from the field
What the history says about how change actually arrives
Three patterns repeat, and all three are visible in today's decarbonisation debate.
- The winning technology is usually worse at first. Early rotary kilns ran at 9–10 MJ/kg — worse on fuel than a good shaft kiln — and made 20–30 t/day. They won on throughput and consistency, not efficiency. Every kiln change since has won on the same two, and efficiency has followed later. A consultant assessing a new technology on its current efficiency is using the wrong test.
- The patent is rarely the invention. Ransome's 1885 rotary kiln patent was commercially unsuccessful because it specified producer gas and omitted clinker cooling. Hurry and Seaman's 1895 patent added coal firing and heat recuperation from the cooler — and that is the machine that exists today. The valuable step was the second one.
- Adoption takes decades, not years. The first suspension preheater ran at Beckum in 1953. UK adoption ran from 1961 to 1977. The precalciner appeared in the early 1970s and took until the 1980s to become standard. When a 2026 vendor tells you their technology will be mainstream in five years, this is the base rate to check it against.
India's cement output went from 3.3 Mt in 1950–51 to 491 Mt in FY26 — roughly a 149-fold increase in 75 years, a compound rate near 6.9% a year. Over the same period world production went up about 19-fold. India grew roughly eight times faster than the world. That is why India's plant fleet is young, why it has already taken every step change, and why its improvement problem now looks like a mature market's rather than a growth market's.
§ 08 The numbers that matter
| Era / technology | kcal/kg clinker | GJ/t | Note |
|---|---|---|---|
| Early rotary, short wet kiln, c.1901 | 2,150–2,390 | 9–10 MJ/kg | 20–30 t/day output |
| Wet process, typical 20th century | ~1,670 | 6.98 | Range 4.9–8.8 MBtu/short ton |
| Long dry kiln | ~1,250 | 5.23 | The wet→dry step is worth ~420 kcal/kg |
| 3-stage preheater | ~812 | 3.4 | |
| 4-stage preheater | ~765 | 3.2 | |
| 5-stage preheater | ~741 | 3.1 | The Indian workhorse |
| 6-stage preheater | ~717 | 3.0 | |
| Best-practice precalciner | ~693 | 2.9 | |
| India average, 2025 | 740 | 3.10 | CII v7.0 |
| India best, 2025 | 670 | 2.80 | CII v7.0, 6-stage |
| Thermodynamic floor | 411 | 1.72 | Heat of reaction only, 1,719 kJ/kg at 25 °C |
| Your plant | — | — | And which era's technology is it? |
The ENERGY STAR guide states the preheater figures in both MBtu/short ton and GJ/tonne, and the two do not convert into each other: 3.8 MBtu/short ton × 1.163 gives 4.42 GJ/t, not the 3.4 GJ/t printed alongside it. The GJ/t series is consistent with the EU BREF and with Indian CII data and is the one used above. The MBtu series should not be quoted. A published benchmark from a reputable institution can still be internally inconsistent — check the conversion before you use either number.
§ 09 Build the model — how much is left
Step 1. The starting point: a 1901 wet rotary kiln at about 2,270 kcal/kg (midpoint of 2,150–2,390).
Step 2. The floor: 1,719 kJ/kg ÷ 4.187 = 411 kcal/kg, the heat of reaction and nothing else.
Step 3. The total prize ever available: 2,270 − 411 = 1,859 kcal/kg.
Step 4. India's best plant is at 670, so it has captured 2,270 − 670 = 1,600 kcal/kg.
Step 5. 1,600 ÷ 1,859 = 86% of everything that was ever available has already been taken.
The remaining 259 kcal/kg is not all addressable either — a large part of it is preheater and cooler exhaust that already dries raw meal and drives waste heat recovery. Call the genuinely recoverable remainder somewhere near 100 kcal/kg on a good plant, and you have the honest ceiling for any thermal-efficiency programme.
Move the sliders to place a specific plant on that curve.
§ 10 Upgrade paths — what the history rules out
The historical series is itself an upgrade argument, and its main use is ruling things out.
| Step change | Worth | Available to an Indian plant in 2026? |
|---|---|---|
| Wet → dry | ~420 kcal/kg | No. India has essentially no wet kilns left. Taken. |
| Dry → suspension preheater | ~440 kcal/kg | No. Taken. |
| Preheater → precalciner | ~50–120 kcal/kg | Rarely. Almost all Indian capacity is already precalciner. Where it isn't, this is a ₹780–1,100 crore project. |
| 5-stage → 6-stage | 8–24 kcal/kg | Yes, and it is the last one left — see A7 §09 for why the evidence conflicts and why it rarely pays. |
| The register | 30–70 kcal/kg | Yes, always, and it is what remains. Lesson A7 §06. |
| Electrified calcination / oxyfuel | CO₂, not kcal | Not yet, and note what it does. These attack emissions, not energy. Neither reduces the 411 kcal/kg floor. |
That table is the whole strategic point of studying the history. Every step change that was worth hundreds of kcal/kg has been taken. What is left is one marginal capital project and a register of small items — which is exactly what lesson A7 spends its length on, and exactly why the frontier moved only 1.8% in the last eleven years.
§ 11 The frontier, and who is running it
The honest version of this section is short. Nothing has superseded Portland cement in 180 years, and nothing in commercial production today is a serious candidate to. What has changed since 1950 is how the heat is recovered and how much of the clinker can be replaced — and both of those are now close to their limits on conventional technology. The genuine frontier is carbon capture, and that is being built, expensively, at a handful of named sites.
| Status | What | Where | The number |
|---|---|---|---|
| SETTLED | The step change that already happened | Suspension preheater, Beckum, Germany, 1953. Precalciner, Japan, 1970s. Grate cooler, progressively since the 1950s. | Wet process 5,512 MJ/t clinker → semi-dry 3,918 → preheater-only 3,700 → dry preheater-precalciner 3,515 MJ/t weighted world average, with a range of 3,000–4,000. Most of the total possible gain was taken between 1950 and 1990. |
| SETTLED | Where the curve has flattened | Indian benchmark, best plants, not named. | Best six-stage plant 2,824 MJ/t clinker and 56.1 kWh/t cement; the top ten Indian plants all sit below 2,887 MJ/t and 70 kWh/t. The gap between the best five-stage and the best six-stage plant is about 8 kcal/kg — roughly 1%. Adding a preheater stage is no longer a strategy. |
| OPERATING | The first large-scale oxyfuel cement kiln | Mergelstetten — catch4climate, Germany. Inaugurated 8 July 2026. | 450 t/d, more than €120m. The first time the kiln itself has been redesigned for capture rather than having capture bolted on. |
| OPERATING | The first full-scale capture plant on a cement kiln | Brevik — Heidelberg Materials, Norway. | 400 kt/yr design. About 185 kt injected over thirteen months, which is roughly 43% of design. Abatement cost around €100/t. The gap between design and delivered is the number worth carrying. |
| BUILDING | The next two | Lägerdorf — Holcim, Germany (1.2 Mt/yr). Padeswood — Heidelberg Materials, UK (800 kt/yr, 2029). | Lägerdorf delayed on German carbon-storage legislation. Neither is operating, and both are routinely cited as though they were. |
| ON HOLD | What happens when the carbon price does not arrive | Edmonton — Heidelberg Materials, Canada. | C$2bn project paused. The business case assumed a carbon price near C$170/t; the realised price was near C$45. A capture project is a carbon-price bet, and this is what losing it looks like. |
Roman concrete is the most frequently misused fact in this industry. Roman pozzolanic mortar is not cement and is not stronger than modern concrete — it is a lime-pozzolan binder that gains strength slowly, over years, and the surviving structures are a survivorship sample of massive compression-only forms. The interesting and true part is the chemistry: lime plus a reactive aluminosilicate, which is exactly what a calcined clay cement is. The Romans were doing the thing the industry is now returning to, for the same chemical reason and with none of the performance.
Use it that way and it is a useful bridge to the low-clinker argument. Use it the other way — as evidence that the industry has forgotten something — and you will be corrected by the first materials engineer in the room.
§ 12 What's changing now
Current as of August 2026 · refresh every six monthsSeymour, Masic and colleagues in Science Advances attributed Roman concrete's crack-sealing to lime clasts formed by hot mixing with quicklime, not to volcanic ash and not to seawater. Modern specimens made with clasts stopped water flow through a crack completely within 30 days.
What it means: a low-tech, low-clinker durability mechanism is attracting serious research money in the 2020s. It is also the single most mis-quoted result in the industry — which makes getting it right a cheap credibility marker in a technical room.Oxyfuel went to demonstration scale at Mergelstetten in July 2026, and electrified calcination is going into Adani's Boyareddypalli plant. Both are covered in A7 §12.
What it means, in historical terms: the kiln has burned a fuel in air since Hurry and Seaman in 1895. Changing the oxidant or the heat source is the first change to that arrangement in 130 years. Judge it the way the history says to judge these things — on whether it delivers throughput and consistency, not on its current efficiency, and expect two decades rather than five years.§ 13 Check yourself
§ 14 Go deeper
§ 15 Carry forward
- "Your kiln runs at 1,450 °C rather than 900 because alite only forms once part of the charge melts. That's been true since the 1840s and no technology on anyone's roadmap changes it."
- "About 86% of the total thermal efficiency ever available has already been captured — the wet-to-dry step alone was worth more than the entire gap that's left. So I'm not going to bring you a step change, because there isn't one. I'm going to bring you a register."
- "The first suspension preheater ran in 1953 and Britain finished adopting it in 1977. When a vendor tells me their technology is mainstream in five years, that's the base rate I check it against."