Calculate an Isotope Ratio (δ¹⁸O, δ¹³C, δD)
The ratio between two isotopes of the same element (¹⁸O/¹⁶O, ¹³C/¹²C…) varies slightly with the origin and history of a rock or geological fluid. Rather than the raw ratio, geochemists use the delta offset from an international standard, expressed in parts per thousand (‰): a readable figure, comparable from one lab to another.
Positive delta: the sample is enriched in the heavy isotope relative to the standard. Negative delta: it is depleted. That sign, and its size, is what points to the material's origin (evaporation, formation temperature, carbon source…).
Isotope system and measurements
Starred jargon is explained in the glossary that closes the article.
Why delta, not the raw ratio
R = heavy / light is the measurement. For oxygen it sits near 0.002. Natural shifts are in the fourth decimal place. Writing 0.002001 versus 0.002005 glazes a room. Two mass specs will also disagree on the fifth digit because each has its own discrimination. Raw R is a poor language for both problems.
R = Nheavy / Nlight
δ (‰) = (Rsample / Rstandard − 1) × 1,000
Delta* folds the tiny offset into parts per thousand against a shared zero. Positive: the sample is enriched in the heavy isotope relative to the standard. Negative: depleted. A granite quartz at δ18O = +8‰ and a polar snow at −40‰ are readable, plottable, and comparable between labs. The instrument still measures R; the lab then measures the standard in the same run and reports the difference. Without that second measurement the fifth decimal is a serial number, not a geology.
The calculator does the same arithmetic. Two abundances go in — cps, millivolts, atom counts, it does not care — and a standard R. What is published is δ, with the scale named. A number without VSMOW or VPDB is not an isotope result. It is a ratio with the label peeled off.
VSMOW, VPDB, VCDT: three zeros, three bottles
The zero of a delta scale is a bottle, or the ghost of a bottle, held by the IAEA and a handful of labs that know how to run it. Everyone else runs secondary waters, calcites, sulphides, and reports onto the primary scale through a calibration.
- VSMOW* (Vienna Standard Mean Ocean Water) — oxygen and hydrogen in water, and oxygen in silicates after fluorination. By definition δ18OVSMOW = 0‰ and δ2HVSMOW = 0‰. SMOW itself was a recipe; VSMOW and now VSMOW2 are the physical waters. Ocean water sits near this zero. Rain does not.
- VPDB* (Vienna Pee Dee Belemnite) — carbon, and oxygen in carbonates. The original Pee Dee belemnite is gone. The scale lives on through NBS-19 (and NBS-18, LSVEC, and whatever the IAEA currently blesses). Marine calcite clusters near 0‰ in both δ13C and δ18O on this scale. That is why palaeoceanographers like it.
- VCDT* (Vienna Canyon Diablo Troilite) — sulphur. The Canyon Diablo iron meteorite’s troilite was heterogeneous. IAEA-S-1 defines the Vienna scale. Sulphide versus sulphate, bacterial reduction, magmatic S: all reported as δ34SVCDT.
The same oxygen atom cannot be 0‰ on both VSMOW and VPDB. A water at 0‰ VSMOW is about −30‰ VPDB. Carbonate people and water people have spent decades talking past each other on that offset. Name the scale every time. The drop-down on this page already does, for the four common pairs.
δ18O in ice and in carbonate: two archives
Two archives, one heavy oxygen, two different stories. Ice is the precipitation. Carbonate is a mineral that grew from a water, at a temperature, with a fractionation. Mixing those stories is how a talk slide becomes a fight in the corridor.
In an ice sheet, snowfall keeps the δ18O of the cloud. Colder condensation, more depleted snow — a rough 0.6 to 0.7‰ per °C in polar regions, with source-region and season piled on. Greenland cores (GRIP, GISP2, NGRIP) cover the last glacial cycle in annual layers near the top. Antarctic cores (Vostok, EPICA Dome C) push beyond 800 ka, diffusion smearing the finest layers at depth. The y-axis is vs VSMOW. A jump of 5‰ is a climate event, not a lab hiccup.
Marine calcite — foraminifera, a bivalve, a speleothem if we are already in a cave — records a blend: δ18O of the water, plus a temperature-dependent fractionation between water and solid, plus whatever kinetic sins the organism or the drip committed. The old Epstein-style palaeothermometer is the idea, not a universal polynomial:
δ18Ocalcite = f (T, δ18Owater, kinetics)
T cannot be read from a foraminifer without a guess at ice volume (which itself shifts seawater δ18O). Rainfall cannot be read from a stalagmite without a guess at cave temperature and drip-rate. Ice cores skip that fork: they are the water. Carbonates keep the better chronology in the oceans and on land where no ice sheet ever sat. Use both. Do not swap the axes.
δ13C as a reservoir tracer
Carbon has two stable isotopes and a handful of reservoirs that do not overlap much. That is the whole trick. Measure δ13CVPDB, and the question is often “which reservoir?”, not “how hot?”.
- C3 organic matter — most trees, temperate crops: about −27‰. Photosynthesis is greedy for 12C.
- C4 organic matter — maize, sugarcane, many tropical grasses: about −13‰. A different pump, a smaller discrimination.
- Marine carbonate — near 0‰, by construction close to VPDB. A limestone that sits at −1‰ is not “organic”. It is a limestone.
- Marine organic carbon — roughly −20 to −22‰.
- Mantle / volcanic CO2 — about −5 to −8‰. A mid-ocean-ridge bubble, a crater lake, a soil gas with a magmatic spike.
- Thermogenic methane — −40 to −50‰; microbial methane can run −60 to −80‰. The gas window versus the bog.
A cement at −22‰ is not a marine calcite with a funny lab day. It has an organic carbon source, or it has mixed with one. A δ13C crash across a bed — Permian–Triassic, Paleocene–Eocene — is a carbon-cycle event: extra light carbon from organics, methane, or a flood-basalt plumbing. The same δ13C on a DIC sample from a borehole is how soil CO2 is told from dissolving marble. Pair it with calcite SI and the water starts to make sense.
Equilibrium and kinetic fractionation
Same element, different mass, slightly different habits. That is fractionation. Two flavours, and they do not sign the same way in every process.
Equilibrium fractionation is a bond story. The heavy isotope prefers the stiffer bond, the higher oxidation state, the condensed phase. Liquid water is a few permil heavier in 18O than the vapour it sits under. Calcite is heavier than its parent water by an amount that shrinks as T rises — the palaeothermometer in one sentence. The factor is α = RA / RB. For small offsets, 1,000 ln α ≈ δA − δB.
αA/B = RA / RB · 1,000 ln α ≈ δA − δB
Kinetic fractionation is a speed story. 16O and 12C move and react a little faster. Evaporation into dry air, a rapid degassing, an enzyme that hates 13C: the product goes light, sometimes by tens of permil, and equilibrium α is a souvenir. Speleothem people lose sleep over this when the cave is too windy. All the same, a Rayleigh rain-out of an air mass is mostly equilibrium condensation plus a reservoir that gets lighter as it snows. The GMWL is that process, smeared over a planet.
The global meteoric water line
Craig’s line, 1961, from rain and snow all over the map:
δ2H = 8 × δ18O + 10 (‰ vs VSMOW)
Slope 8 is roughly the ratio of the two equilibrium fractionations for hydrogen and oxygen during condensation. The +10 is deuterium excess on the mean ocean source. A rain that condensed from an air mass, without later evaporation, sits on or near this GMWL* — or on a local meteoric water line with a slightly different slope and intercept, which is what is actually fitted at a given station.
Deviations are the geology. A lake in a dry basin slides off with a slope of 4–6: evaporation. A geothermal water that swapped oxygen with silicate at 250 °C shifts δ18O to the right at nearly constant δD — the classic “oxygen shift”. A glacial melt sits very negative and usually still on the local line. Seawater itself is near (0, 0), which is not on the GMWL (the line at δ18O = 0 predicts δD = +10). That mismatch is how a coastal borehole tells rain from a marine mix. For that matter, plotting one isotope is half a story; the pair is the diagnostic.
What a working lab actually quotes as 1σ
IRMS* (and now cavity-ring-down lasers on waters) are good. They are not magic. A working lab, not a brochure:
- δ13C on carbonate or organic carbon: ±0.05 to 0.1‰ (1σ) on a good day; ±0.01‰ only with dual inlet, long counts, and a quiet source.
- δ18O on carbonate (phosphoric acid) or water (CO2 equilibration): ±0.05 to 0.1‰.
- δ2H on water: ±0.5 to 1‰. Hydrogen is a small ion and a leaky system; treat a ±0.05‰ claim on δD with the respect it deserves.
- δ34S: ±0.1 to 0.3‰, method depending.
Reproducibility is a protocol. Standards at the start, middle and end of the run. Drift correction. The same acid, the same bath temperature, the same crush. A secondary standard that is not the calibration standard, so a lie in the line is visible. Intercomparison waters (GISP, SLAP, a local bottled water that is trusted). A δ18O printed to 0.001‰ with no standard name and no n is a poster, not a datum. This form gives two decimals. That is already plenty until the 1σ of the batch is known.
A delta without its scale (VSMOW, VPDB, VCDT) is not usable. A water δ18O of “−6.2‰” might be ordinary rain on VSMOW or an impossible carbonate on VPDB. Write the scale, or the next lab will put the point in the wrong ocean.
Worked example: rain versus a coastal borehole
Two waters, both vs VSMOW, from a coastal plain. δ18O and δD are enough to be rude about origin. They are not enough to date the recharge.
| Sample | δ18O (‰) | δ2H (‰) | d-excess (‰) |
|---|---|---|---|
| Rain (this month) | −7.5 | −50 | 10 |
| Coastal borehole | −3.0 | −24 | 0 |
| Seawater (local) | 0.0 | 0 | 0 |
d-excess = δD − 8 δ18O. Rain: −50 − 8 × (−7.5) = 10, on the GMWL. Predicted δD = 8 × (−7.5) + 10 = −50. Exact hit. Meteoric, no argument.
Borehole: d-excess = −24 − 8 × (−3.0) = 0. Predicted GMWL δD = 8 × (−3.0) + 10 = −14. Measured −24. Off the line, pulled toward (0, 0). A two-component mix of the rain and local seawater would give, for a seawater fraction f:
δ18O = (1 − f)(−7.5) + f(0) ⇒ f ≈ 0.60 from oxygen
δDmix = (1 − 0.60)(−50) = −20 (measured −24)
Close, not perfect. A little evaporation on top of a mix, or a rain end-member that is not this month’s storm, would cover the extra 4‰ on hydrogen. What it is not is pure meteoric water sitting on the GMWL at a warmer-season composition: that water would still have d-excess near 10. SI on calcite, Cl−, and a bromide or sodium would settle whether the mix is seawater or a salty evaporite brine (the latter often leaves a different δD–δ18O fingerprint). Isotopes opened the file. They did not close it.
North Sea formation water on the GMWL plot
UKCS formation waters (Brent, Forties, Rotliegend) are a textbook English-language dataset. Many sit off the GMWL toward seawater (0, 0) or toward an evaporitic brine that is heavier still in δ18O. BGS and operator reports treat a Forties aquifer at δ18O ≈ −4 to −1‰ VSMOW, d-excess near 0, as a mixed or marine-influenced water — not as “warmer rain”. A Rotliegend brine can sit at positive δ18O after water–rock exchange, with δD still negative: the classic oxygen shift, not Craig’s line.
That is why a North Sea core-analysis pack quotes both isotopes. One number looks like a climate. The pair looks like a mix, an evaporite, or a silicate exchange. The same plot used on a Cornish granite spring (on the local meteoric line, d ≈ 10) would be a different water entirely.
Ice cores versus speleothems
Ice cores are the clean end-member. The precipitation itself is drilled. Dating is layer counting, then volcanic markers, then orbital tuning at depth. The isotope is vs VSMOW. The physics is condensation temperature plus the path of the air mass. Diffusion, melt layers, and a thinning column are the tax. An ice sheet is also required. Mid-latitude Holocene palaeoclimate is not in Greenland.
Speleothems (stalagmites, flowstone) grow in caves from drip water. The calcite is dated by U–Th, sometimes annually layered, and the δ18O is reported vs VPDB. The drip is meteoric water that has already gone through a soil and an epikarst. Cave temperature sets an equilibrium fractionation; a fast drip and a draught add kinetic offset; the “amount effect” in monsoon regions can dominate T. Hulu, Sanbao, Soreq, and British caves in the BGS palaeoclimate archive: they are why land-based millennial structure is aligned with ice cores, in places ice never reached.
Reading one as if it were the other is the regular mistake. A 1‰ shift in ice is not a 1‰ shift in a stalagmite, and neither is 1.5 °C without a paragraph of caveats. Convert the speleothem onto VSMOW if an overlay is required, then still remember a mineral is being plotted against a snow. The geochemistry is oxygen in both cases. The archive is not.
VSMOW versus VPDB for oxygen: two zeros, one conversion
Water labs speak VSMOW. Carbonate labs speak VPDB. The conversion is linear and not optional:
δ18OVSMOW = 1.03091 × δ18OVPDB + 30.91
A marine calcite at 0‰ VPDB is about +31‰ VSMOW. A polar ice at −40‰ VSMOW is about −69‰ VPDB — a number no carbonate person would ever write, which is the point of keeping two scales. Plug a VPDB carbonate value into this calculator’s VSMOW slot and a water that does not exist is invented.
Silicates (quartz, magnetite, a whole-rock granite) are fluorinated and reported vs VSMOW. A +8‰ quartz is a crustal rock, not a “+8‰ rainfall”. The mineralogy that goes with that number is still cheaper as a thin section than as another mass spec run. The drop-down, the two intensities, and the scale named on the figure: the arithmetic is the easy bit.
Deuterium excess: the GMWL intercept as a tracer
The +10 in Craig’s equation is a mean. The useful local number is deuterium excess:
d = δ2H − 8 × δ18O
Ocean evaporation into a humid air mass gives d around 10. A dry, windy source (eastern Mediterranean in winter) pushes d toward 20 or more: kinetic evaporation, 2H and 18O do not leave in the 8:1 ratio. Recycled continental moisture, a frozen sea-ice source, a different story again. Once the rain has fallen, evaporation from a lake or a soil column lowers d and flattens the slope.
d is a source-and-kinetic tracer, not a thermometer. Greenland interstadials, a monsoon speleothem’s drip, a desert aquifer that last saw rain in a different humidity regime: people plot d against time for that reason. A coastal mix toward seawater, as in the worked example, also drops d toward 0 because the ocean end-member itself has d ≈ 0. One extra plot after this page: d. δ18O alone cannot tell a warm rain from a seawater leak.
The mass spec will faithfully report the water that was posted. Fill the bottle — glass or dense HDPE — to the brim; a half-full vial in a hot van is a tiny evaporation experiment (both deltas climb, d-excess falls). Write the time, the T, and the pH on the label. The rock can wait on a shelf. The water cannot.
Glossary
- Delta (δ)
- The offset of a sample isotope ratio from a standard, in parts per thousand (‰): δ = (Rsample/Rstandard − 1) × 1,000. Positive = enriched in the heavy isotope; negative = depleted.
- VSMOW
- Vienna Standard Mean Ocean Water. The zero of the oxygen and hydrogen scales for water (and for oxygen in silicates after fluorination). Ocean water sits near 0‰; meteoric water does not.
- VPDB
- Vienna Pee Dee Belemnite. The carbon scale, and the oxygen scale for carbonates. Marine calcite clusters near 0‰. Not interchangeable with VSMOW.
- VCDT
- Vienna Canyon Diablo Troilite. The sulphur isotope scale, now defined by IAEA-S-1 rather than the original meteorite troilite.
- GMWL
- Global meteoric water line: δ²H = 8 × δ¹⁸O + 10 (‰ vs VSMOW). Rain and snow that condensed without later evaporation sit near this line. Seawater at (0, 0) does not.
- IRMS
- Isotope-ratio mass spectrometry. The laboratory method that measures R on CO2, H2 or SO2 prepared from the sample, then reports δ against a standard run in the same batch.