EN

Calculate Element Concentration (ppm, wt%)

A chemical analysis or mass spectrometry reading gives an element mass within a total sample mass: this calculator converts that ratio into weight percent or parts per million (ppm), the two units of concentration most used in geochemistry.

1% = 10,000 ppm. Weight percent suits major elements (silicon, aluminum, iron…), ppm suits trace elements (gold, uranium, rare earths…) whose content would be unreadable as a percentage.

Method

Element mass measured within the total mass of the analyzed sample.

Measured masses

An analysis result already reduced to a single figure, to convert to the other unit.

Value to convert

The glossary at the end of the article covers every starred term.

wt% or ppm: which unit for which element

Nobody reports silicon in a granite as 691,000 ppm. Nobody reports gold as 0.00000013 wt%. The number is the same mass fraction; the unit is a choice about whether it can be read. Weight percent* (wt%) is grams of analyte per 100 g of sample. ppm* is grams per million grams — milligrams per kilogram, micrograms per gram, same thing.

1 wt% = 10,000 ppm  ·  1 ppm = 0.0001 wt%

Majors live in wt%. Silicon, aluminium, iron, calcium, sodium, potassium, magnesium, titanium, phosphorus: the oxide is quoted, one or two decimals, and the column looks like a rock. Traces live in ppm. Barium, rubidium, nickel, uranium, the REE. Put Ba at 0.188 wt% on a certificate and a petrologist will ask whether a feldspar separate was meant.

The 1,000 ppm line (0.1 wt%) is the usual hand-off, not a law of nature. Strontium in a granite sits around a few hundred ppm; in a carbonate it can be 0.1 wt% and still get written as 1,000 ppm out of habit. The unit that matches the other analyses in the same table is the right one. Mixing wt% SiO2 with ppm Si in one column is how a spreadsheet “normalisation” quietly wrecks a TAS plot.

What an ICP-MS actually counts

An ICP-MS* does not weigh the element. It counts ions. The rock has already been dissolved (or fused and then dissolved), diluted, and pumped as a mist into an argon plasma at about 10,000 K. Molecules are gone. Most atoms are stripped of one electron. A quadrupole or a magnetic sector then sorts those ions by mass-to-charge, m/z, and a detector reports counts per second.

Counts are not a concentration. A 10 ppb indium spike, a calibration line from a multi-element standard, a dilution factor, and the original test-portion mass turn cps into µg/g in the powder. Miss the dilution and every number on the sheet is wrong by that factor — 50 mL from 0.100 g is not 100 mL from 0.050 g, even if the cps look similar.

ppm in rock = (µg/L in solution × dilution volume, L) ÷ (test portion, g) × 1,000

The isotope actually counted is a choice. 56Fe sits under 40Ar16O+. A collision cell or a high-resolution slit, or a jump to 57Fe and the smaller abundance. The certificate line that says “Fe 48,200 ppm” is several steps away from the ion the instrument saw. Isotope ratios are a different measurement again — two masses, a delta, a standard. This page is the bulk number: how much of the element is in the powder.

Majors, minors and traces

A silicate analysis is not a flat list of 50 elements. It is two jobs glued together. Majors* (typically > 0.1 wt% as the oxide) build the minerals named in thin section: quartz, feldspar, olivine, pyroxene, micas, oxides. Their sum, plus LOI, should land near 100%. That is the composition the rock is classified with, and the one that has to be consistent with the mode and with density.

Minors sit in the awkward band, roughly 0.01–0.1 wt% (100–1,000 ppm): Ba, Sr, Zr, V, Cr in many basalts, Mn if it was not already parked with the majors as MnO. Useful, not rock-forming. Traces are everything below that — Ni, Cu, Rb, Y, Nb, the REE, Th, U, Pb, Hf, Ta. They do not change the name of the rock. They speak to melting, source, contamination, and whether a magma sat in the crust long enough to grow zircon.

Compatible traces (Ni, Cr, Co in mafic systems) follow olivine and spinel; a picrite with 400 ppm Ni is doing what olivine does, a “basalt” with 5 ppm Ni has fractionated hard. Incompatible traces (Rb, Ba, Th, U, LREE, Nb, Ta) stay in the melt — hence 1,800 ppm Ba in a granite and 10–20 in a mid-ocean-ridge basalt. In practice the split is also analytical: majors by XRF or ICP-OES as oxides; traces by ICP-MS in ppm. One powder, two methods, one report. A WD-XRF Fe2O3T and an ICP-MS Fe should not be averaged “because they should match”. When they do not, the digest or the bead is the suspect, not the geology.

ppm, ppb, ppt: three more zeros

ppm is already a millionth. Push two more zeros and the register is environmental water, gold, mantle Os. The SI-friendly translations:

1 ppm = 1 mg/kg = 1 µg/g

1 ppb = 1 µg/kg = 1 ng/g  ·  1 ppt = 1 ng/kg = 1 pg/g

A modern quadrupole ICP-MS will see many elements at low ppt in a clean solution. That is not the same as ppt in the rock. Dilute a 0.1 g digest to 100 mL and a factor of 1,000 is already gone; 10 ppt in solution is 10 ppb in the powder. The blank from the acid and the beaker is often the floor, not the detector.

Gold in ordinary crust is about 1 ppb; uranium in a granite is a couple of ppm. The calculator converts 0.0002 wt% to 2 ppm. It will not say whether that 2 is a major, a trace, or a detection-limit artefact.

Oxide to element: SiO2 is not Si

XRF* and the classical wet majors are reported as oxides: SiO2, Al2O3, Fe2O3, MgO, CaO, Na2O, K2O, TiO2, P2O5, MnO. That is convention, not a claim that the rock is a bag of pure oxides. To get the element itself, multiply by the mass of the cation(s) over the mass of the oxide — the same molar masses used in a chemistry lab.

wt% element = wt% oxide × (Melement × n) / Moxide

Silicon: M(Si) = 28.085 g/mol, M(SiO2) = 60.085 g/mol, factor 0.4674. A granite at 69.1 wt% SiO2 holds 32.3 wt% Si, not 69.1 wt% Si. Mix those two and a mass balance against a mineral saturation calculation is nonsense.

OxideFactor → elementExample
SiO2 → Si0.467454.1 wt% SiO2 → 25.3 wt% Si
Al2O3 → Al0.529315.4 wt% Al2O3 → 8.15 wt% Al
Fe2O3 → Fe0.699413.8 wt% Fe2O3T → 9.65 wt% Fe
FeO → Fe0.7773keep FeO and Fe2O3 apart when both are reported
MgO → Mg0.60313.59 wt% MgO → 2.16 wt% Mg
CaO → Ca0.71477.12 wt% CaO → 5.09 wt% Ca
Na2O → Na0.74193.16 wt% Na2O → 2.34 wt% Na
K2O → K0.83014.48 wt% K2O → 3.72 wt% K
TiO2 → Ti0.59942.26 wt% TiO2 → 1.35 wt% Ti
P2O5 → P0.43640.35 wt% P2O5 → 0.15 wt% P

The reverse trip — element to oxide — is the reciprocal. ICP-MS Fe in ppm ÷ 10,000 ÷ 0.6994 gives Fe2O3T in wt%, if the XRF column is wanted back. A four-acid Fe (some resistate left behind) will not close with the bead; that is the digest, not a rounding error.

LOD and LOQ are method properties

The limit of detection* (LOD) is the lowest concentration that can be told from a blank with any honesty — classically three standard deviations of the blank. The limit of quantification (LOQ) is where a number is printed and stood on, often ten standard deviations of the same blank. Between LOD and LOQ the lab may write “detected, not quantified”, or a flagged figure, or just < LOQ.

Those limits are not a property of the element. They belong to this method, this matrix, this dilution, this day. WD-XRF on a fused bead will give SiO2 to a tenth of a percent without drama and then fall over for gold. Fire assay plus ICP-MS will see 1 ppb Au in a 30 g charge and is a stupid way to measure Si. A 0.1 g four-acid digest is a stupid way to measure Au in a vein sample: the nugget sits in the other 29.9 g.

A result printed as < LOD is not zero. It is “this method cannot see it here”. Re-run the same powder by a method whose LOD sits below crustal background, or say the content is unknown. Treating every less-than as 0.0 ppm in a spider diagram pulls the whole pattern to the floor.

Where geochemical numbers go wrong

The mass spec is rarely the loudest error. The rock is. Crushing in a steel jaw, a tungsten-carbide mill, a dusty crusher that just did a massive sulphide: Cr, W, Co, Ta, Nb contamination is real. Split a coarse reject badly and the next 50 g is not the same rock as the previous 50 g. That is sampling, not chemistry.

  • Incomplete digest. Zircon, chromite, cassiterite, barite shrug at aqua regia. Zr, Hf, Cr, Sn, Ba, the HREE come out low. Lithium-borate fusion or a properly fluxed bead gets them; a “trace package” on a 0.25 g aqua-regia leach does not.
  • Spectral overlap. ArO on 56Fe, ArCl on 75As, oxides of the LREE on the HREE. Collision cell, high resolution, or a different isotope. If the lab cannot say which, the number is a rumour.
  • Matrix and drift. A 2% TDS solution does not ionise like a 1% HNO3 standard. Internal standards (Rh, In, Re, Ge) mop up a lot of that; they do not fix a calibration built on the wrong matrix.
  • The wrong test portion. Majors on 0.4 g of fused bead, traces on 0.1 g of powder, gold on 30 g. Quote the charge. A 5 g “whole rock” Au on a quartz vein is still a lottery.

The most common spreadsheet error is still unit confusion: µg/L in the solution copied into a ppm-in-rock column, or wt% Fe2O3 compared with ppm Fe without the factor 6,994. This calculator exists for that afternoon.

Worked example: reading a granite certificate

A granite powder comes back from a commercial package. Majors by WD-XRF on a fused bead, traces by ICP-MS on a four-acid digest, gold by 30 g fire assay. Units, methods, and less-thans are the hunt — not a novel.

AnalyteValueUnitMethodLOD
SiO269.14wt%XRF bead0.01
K2O4.48wt%XRF bead0.01
Ba1,882ppmICP-MS0.5
Rb170ppmICP-MS0.1
Ni5ppmICP-MS0.2
Cu11ppmICP-MS0.2
U2.07ppmICP-MS0.05
Au< 0.001ppmfire assay0.001
  1. Units first. SiO2 69.14 is wt% oxide, not elemental Si. 69.14 × 0.4674 ≈ 32.3 wt% Si, or 323,000 ppm Si. Ba 1,882 ppm is 0.188 wt% — a minor, reported as a trace because everyone else does.
  2. Method next. Zr on this four-acid row would be a lower limit, not a concentration, if the granite is zircon-rich. Check whether Zr, Hf, Cr, Nb came from a fusion. If they came from the same digest as Cu, treat Zr as qualitative.
  3. LOD last. Au < 0.001 ppm means < 1 ppb. Continental crust sits near that. The rock has not been shown barren of gold; this fire assay did not see any. Ni at 5 ppm on a 0.2 ppm LOD is a real number, just a small one — granite, not a komatiite.

Convert 1,882 ppm Ba on the second tab: 1,882 / 10,000 = 0.1882 wt%. Convert 4.48 wt% K2O to elemental K: 4.48 × 0.8301 ≈ 3.72 wt% K. Those two operations are the whole of this page. The geology is in whether the pair looks like G-2, or like a sample that was milled in steel and then “gained” 80 ppm Cr.

XRF bead, four-acid digest, fusion: why two labs disagree

A fused lithium-borate bead plus WD-XRF is the workhorse for silicate majors: powder mixed with flux, melted, cast; volatiles gone (that is what LOI was for); Si, Al, Fe, Mg, Ca, Na, K, Ti, P, Mn as oxides. Four-acid (HF–HNO3–HClO4–HCl) ICP-MS is fast and cheap and misses the resistates. A lithium-metaborate fusion then ICP-MS takes Zr, Hf, Cr, Nb, Ta, Sn with it. Aqua regia is a leach, not a whole-rock granite. Portable ED-XRF on a sawn core face is a screening tool, not G-2.

Two labs, two numbers: 310 ppm Zr by fusion, 140 ppm Zr by four-acid. The rock did not change. The zircon did not dissolve in the cheaper package. Quote the method next to the number, or the next person will average them.

USGS G-2 granite and BCR-2 Columbia River basalt

Two USGS powders do a lot of that work in silicate geochemistry. G-2 is a granite (Westerly, Rhode Island). BCR-2 is a Columbia River basalt. Chinese IGGE GSR-1 is the granite many Asian labs run instead.

G-2 graniteBCR-2 basalt
SiO2 (wt%)69.154.1
Al2O3 (wt%)15.413.5
Fe2O3T (wt%)2.713.8
MgO (wt%)0.763.59
CaO (wt%)1.967.12
Na2O (wt%)4.083.16
K2O (wt%)4.481.79
TiO2 (wt%)0.482.26
Ba (ppm)1,880683
Sr (ppm)478346
Rb (ppm)17048
Zr (ppm)309188
Ni (ppm)513
U (ppm)2.11.7

Those are working figures, rounded, not a substitute for the current USGS certificate. Run G-2 with the batch. If SiO2 comes back 71.8 and Zr 140, the problem is flux or digest, not a new granite. BCR-2 is the sanity check for a mafic suite: high Fe–Ti, modest Ni–Cr (it is an evolved flood basalt, not a picrite), Ba and Sr in the hundreds. A “basalt” with G-2’s K2O and Rb is not a basalt.

BGS G-BASE and crustal background in Britain

The British Geological Survey G-BASE (Geochemical Baseline Survey of the Environment) mapped stream sediment and soil over much of the UK. The numbers are not whole-rock granite certificates; they are drainage and soil, with all the weathering and contamination that implies. They are still the English-language reference for “what does crust look like here?” — Cu a few tens of ppm in sediment, U a few ppm, Pb higher in the Pennine orefield than in the Scottish Highlands. A granite powder that looks like G-2 and a Cornish stream sediment that looks like G-BASE are two different samples of two different media.

Continental crust, in round numbers (Rudnick and Gao): Si ~28 wt% as the element, Cu ~27 ppm, Ni ~47 ppm, Cr ~92 ppm, Pb ~17 ppm, U ~2.7 ppm, Th ~10.5 ppm, Au ~1.3 ppb. A method is useful when its LOD sits comfortably below the background that matters. WD-XRF Cu at 5 ppm LOD is fine for ore, blind for a granite at 11 ppm with a 5 ppm floor. ICP-MS Cu at 0.2 ppm LOD sees crust without trying.

Gold is the usual trap. Crustal Au is about 1 ppb. A 0.005 ppm (5 ppb) LOD on a 5 g aqua-regia package will return < 5 ppb on almost every unmineralised rock in the drawer — a map of the method, not of gold. Fire assay at 1 ppb on 30 g is a mineralisation screen, not a mantle-Au study. Match the floor to the question, then convert units here if wt% and ppm are needed on the same line.

Closing the major-element total, LOI, and Fe as FeO or Fe2O3

Add the oxides. Add LOI*. Something like 98.5–101.5 wt% is the target on a silicate. Lower: a missing major (SO3 in a scapolite rock, F in a granite, Cl in a sodalite syenite), a weigh-in error, or a bead that crawled. Higher: LOI and Fe double-counted, or Fe2O3T and FeO in the same sum. Labs differ on whether they recast to 100% anhydrous; which was done belongs on the plot caption.

LOI is H2O+, CO2, organic carbon, some sulphur — and can go negative if FeO oxidises in the muffle. A limestone with 40 wt% LOI is a carbonate, not a bad analysis. A granite with 4 wt% LOI is altered, or still wet, or both. Fe2O3T assumes all Fe is ferric; an ICP-MS total has no opinion about Fe2+.

Glossary

Weight percent (wt%)
Grams of analyte per 100 g of sample. The unit for major-element oxides in a rock analysis. 1 wt% = 10,000 ppm.
ppm
Parts per million: milligrams per kilogram, or micrograms per gram. The unit for most trace elements. 1 ppm = 0.0001 wt%.
ICP-MS
Inductively coupled plasma mass spectrometry. The rock is dissolved, sprayed into an argon plasma, and the resulting ions are sorted by mass-to-charge. Counts become µg/g only after calibration, dilution and the test-portion mass are applied.
Major elements
The oxides that make the rock-forming minerals, typically quoted above ~0.1 wt%: SiO2, Al2O3, Fe2O3, MgO, CaO, Na2O, K2O, TiO2, P2O5, MnO. Traces sit far below that, in ppm.
XRF
X-ray fluorescence. Wavelength-dispersive XRF on a fused lithium-borate bead is the standard method for silicate majors. It does not replace a fire assay for gold.
LOD / LOQ
Limit of detection / limit of quantification. LOD is typically 3σ of the blank (seen, not trusted); LOQ is typically 10σ (printed as a number). Both belong to the method and the matrix, not to the element.
LOI
Loss on ignition: the mass change on heating, mostly H2O+, CO2 and organic carbon. Added to the oxide total when a silicate analysis is closed. Can go negative if FeO oxidises in the muffle.

Related calculators