Impregnated Activated Carbon: Which Chemistry Removes What

No manufacturer and no academic review currently publishes a structured impregnant × contaminant × mechanism × loading × reversibility matrix for activated carbon — and the numbers inside that gap are large ones: potassium hydroxide impregnation raised H₂S capacity on a coconut-shell carbon from 1.5 to 31.7 mg/g at 3,000 ppm and 30 °C, a 21-fold uplift. The closest academic work, Kiani and co-authors’ 2021 review in Environmental Science and Pollution Research, covers the copper–chromium–silver–zinc–molybdenum systems and TEDA in depth but gives no structured matrix and no quantitative BET data. The only genuine manufacturer table we could verify, General Carbon’s guide, pairs six impregnant families with their targets and carries no loadings, no capacities, no breakthrough data. That gap is why this article exists.

The short version

• KOH impregnation raised H₂S capacity from 1.5 to 31.7 mg/g at 3,000 ppm and 30 °C (JESTEC, 2013).
• The uplift is bought with surface area. Five impregnants on one carbon at the same 7 wt % solution took BET from 901.04 m²/g to about 30 m²/g — a decrease “by up to 96 %”.
• Where the impregnant sits matters more than how much there is: at ~10 wt % sulfur, BET ranged 634–824 m²/g on impregnation temperature alone, and mercury capacity tracked surface area, not sulfur content.
• Almost none of it is reversible. Palladium is the exception, and its regeneration route — 700 °C in air — is why the fielded catalyst sits on alumina, not carbon.
• Caustic impregnation carries a documented self-heating hazard: one supplier states the impregnants “decrease the ignition temperature”; UK guidance records two bed fires.

Twelve impregnation chemistries, with the published loading range where one exists and an explicit “not published” where one does not.

Here it is as text, so it can be quoted.

Impregnation adds a reagent that reacts with the target and converts it into something non-volatile. Every chemisorption gain is paid for out of physisorption capacity.

Hydroxides — KOH and NaOH for H₂S and acid gases

Caustic impregnation is the most widely deployed impregnation chemistry in odour control and biogas. The mechanism is acid–base neutralisation followed by catalytic oxidation, and the second half matters more than most datasheets admit. The neutralisation set published in JESTEC is H₂S + NaOH → NaHS + H₂O and H₂S + 2 NaOH → Na₂S + 2 H₂O, the potassium analogues following the same stoichiometry. But the alkaline surface also mediates oxidation of H₂S to sulfur and ultimately sulfate. That is why a caustic bed keeps consuming H₂S after the arithmetic says the hydroxide is spent, and why the heat release is larger and more sustained. Hold that for the safety section.

Loadings. US Patent 4,215,096 specifies about 0.5 to 20 % by weight of sodium hydroxide on a dry carbon basis; Environment Agency Technical Guide 12 gives the practical figure of 5–10 % caustic soda.

Capacity. At 3,000 ppm H₂S and 30 °C, KOH-impregnated coconut-shell carbon at impregnation ratio 1.0 delivered 31.7 mg H₂S/g against 1.5 mg/g unimpregnated; NaOH at ratio 0.25 gave 18.5 mg/g. Calgon Carbon’s ST 1X is specified at minimum 0.14 g H₂S per cc, which the bulletin states is approximately 25 % by weight. The two are not comparable: different challenge concentration, breakthrough criterion and units.

Reversibility: none — the products are sulfides, bisulfides, sulfur and sulfate, consumed stoichiometrically. Suracsh’s SURSORB ALKA (granular) and SURSORB P-ALKA (pellet) both publish CTC min 50 % and hardness min 95. Neither discloses the chemistry or publishes an H₂S capacity.

Potassium carbonate — the stoichiometric advantage

K₂CO₃ goes further per unit of impregnant for a reason that is arithmetic rather than chemistry: H₂S + K₂CO₃ → KHS + KHCO₃, and H₂S + K₂CO₃ → K₂S + H₂CO₃. The JESTEC authors attribute the advantage to the fact that “one molecule of K₂CO₃ will react with one molecule of H₂S”, against two NaOH per H₂S. At ratio 2.0 it reached 37.5 mg H₂S/g at 3,000 ppm and 30 °C — 25 times the unimpregnated carbon — and 62.5 mg/g at 1,000 ppm; ratio 0.5 gave 20.3 mg/g. Peak capacity was at 50 °C, the isotherm fitting Langmuir with R² = 0.9448.

The paper also gives a rare fresh-versus-spent comparison: 726.2 m²/g fresh against 607.4 m²/g spent, pore volume falling 0.40 → 0.30 cm³/g. Sulfur deposition in service costs a further 16 % of surface area on top of whatever impregnation cost. Not reversible.

Magnesium oxide — the weakest-sourced chemistry in this set

The honest version here is more useful than a confident one. The mechanism is alkaline-earth oxide chemisorption, analogous to the calcium oxide and potassium carbonate routes: a basic oxide reacting with an acid gas to give a metal sulfide and water. The key paper, a Materials & Design study on nano-MgO impregnated carbon for H₂S, is paywalled, so its loadings are unverified here.

There is also a trap. The open-access Scientific Reports paper most often cited in support of MgO-impregnated carbon is about CO₂ adsorption, not H₂S: its 12 wt % optimum and 94.02 mg/g capacity are carbon dioxide figures. So this article carries no MgO H₂S loading or capacity.

Suracsh’s SURSORB MAG is a magnesium-oxide impregnated pelletised carbon: CTC min 55 %, bulk density 0.550 ± 0.05 g/mL, hardness min 95, ash max 15 %. The ash figure is the interesting one — 15 % against the 8 % max on the non-impregnated pelletised grades, consistent with a substantial inorganic impregnant. No H₂S capacity is published, and we will not infer one.

Acid impregnation for ammonia

Phosphoric and other mineral acids neutralise ammonia to a non-volatile ammonium salt: 2 NH₃ + H₃PO₄ → (NH₄)₂HPO₄ and analogues; General Carbon’s guide confirms phosphoric acid for ammonia in vapour streams. Virgin carbon holds roughly 0.25–2 mg NH₃/g; acid-impregnated or surface-oxidised carbon holds 19–35 mg/g at comparable inlet. There is a live technical disagreement about why, so that gets its own article: ammonia, and why virgin carbon does almost nothing. Suracsh’s grades are SURSORB GAM (granular, impregnation 8–12 %, CTC min 55 %, cites EN 14387) and SURSORB PAM (pellet, iodine number min 1000 mg/g, CTC min 50 %).

Amine impregnation for formaldehyde

Formaldehyde’s carbonyl carbon carries a partial positive charge; an amine nitrogen attacks it and forms an imine — a Schiff base — releasing water:

HCHO(ad.) + R-NH₂ → R-N=CH₂ + H₂O

The open-access 2022 review in Applied Sciences adds the secondary chemistry — amine methylation ending in dimethylurea species — and lists the amines used: polyethyleneimine, hexamethylene diamine, urea, diethylenetriamine, ethylenediamine, melamine and p-aminobenzoic acid, attached by physical bonding or by grafting.

Its design insight generalises beyond formaldehyde: modification requires an optimised amount of polar functional groups specifically in order to limit water affinity. Load too much amine on and the carbon turns hydrophilic; water out-competes formaldehyde for the same sites, and the modification meant to raise capacity lowers it. The review quantifies no loadings or capacities, so neither does this article. Suracsh publishes SURSORB FORM (granular, proprietary reactive impregnation, CTC min 50 %) with no formaldehyde capacity figure.

Sulfur for mercury

Elemental sulfur chemisorbs mercury vapour as mercuric sulfide, Hg⁰ + S → HgS, the synthetic analogue of cinnabar. The product is a stable solid, so the process is irreversible and spent carbon is disposed of, not reactivated. The 1997 Department of Energy report from the University of Pittsburgh is the quantitative reference, and its central finding is counter-intuitive: more sulfur gave less mercury. The mercury article carries the full dataset. Suracsh publishes SURSORB GG-Hg (granular, mercury loading min 20 %) and SURSORB GP-Hg (pellet); the loading is stated on the granular sheet only.

KI versus TEDA for radioiodine — two different mechanisms

Suppliers who sell both collapse this into “iodine impregnation”.

Potassium iodide works by isotopic exchange. Non-radioactive ¹²⁷I in the impregnant exchanges with radioactive ¹³¹I in methyl iodide, so the radioiodine stays on the bed while ordinary iodine leaves. It is a swap, not a bond-forming reaction, and it depends on the iodide staying accessible.

TEDA works by nucleophilic substitution. Triethylenediamine’s tertiary amine nitrogen attacks the methyl carbon of CH₃I in an SN2 reaction, forming a non-volatile quaternary ammonium iodide salt. A 2024 paper in Environmental Science and Pollution Research confirms the SN2 route and reports iodine capacity rising from 180 to 1044 mg/g mL after TEDA treatment at 60 °C.

So TEDA survives conditions where the alternatives fail. Chebbi and co-authors’ study in Carbon Trends found something that deserves to be better known: at 35 °C and 26 % relative humidity, impregnation reduced performance, the impregnant blocking pores physisorption was using perfectly well. It paid off only at elevated temperature (75 °C, 30 % RH), where the authors report that TEDA impregnation was “required to enhance the trapping stability” and to capture methyl iodide with superior efficiency. If the duty is ambient and dry, an impregnated carbon can be the wrong specification.

TEDA loading percentages are not published in any source we could access; the commonly quoted 2–5 wt % has no primary citation. Suracsh publishes SURSORB NGC, nuclear-grade granular carbon with CTC (60 min) min 60 % and a methyl iodide removal efficiency of min 99.0 % — given without a test method. Request the qualification protocol before specifying against a regulatory requirement.

ASZM-TEDA

ASZM-TEDA is copper, silver, zinc and molybdenum with triethylenediamine on a carbon substrate, the chromium-free successor to ASC whetlerite. It sits against MIL-DTL-32101, which DEVCOM Chemical Biological Center records as the sole testing requirement from 1992, when the Department of Defense moved off chromium-containing carbon after roughly 50 years. The metal weight percentages are controlled by that specification and are not public; this article publishes none.

Suracsh’s SURSORB CBRN+ publishes a breakthrough table — AC, CK, CG, DMMP and HCl, unaged and aged — with BET min 1000 m²/g and hardness min 90. It is export-controlled: an identified end user and end use are required before quotation. The ASZM-TEDA article covers the rest.

Palladium for carbon monoxide — with a correction

No carbon adsorbs carbon monoxide in useful quantity at ambient temperature. CO is catalytically oxidised — CO + ½O₂ → CO₂ — with palladium the catalyst of choice.

Here is the correction, which we would rather publish than be challenged on. The fielded norm for CO removal in respiratory protection is palladium on alumina, not on carbon. US Patent 4,623,637 specifies 0.5 wt % palladium on high-purity alpha-alumina of better than 99 % purity, plus 1 wt % vanadium as the oxide. The reason is regeneration. Work in Nature Communications on atomically dispersed palladium shows deactivation proceeds by ionic palladium collapsing into metallic clusters, and that “the catalyst activity can be regenerated by oxidation at 700 °C in air”. No carbon support survives that. The same paper gives the mechanism as Mars–van Krevelen.

Carbon-supported palladium for CO oxidation is real and is sold; it is simply not the fielded configuration. The alternative catalyst is hopcalite, the copper–manganese oxide mixture; the patent’s rationale for the palladium–vanadium formulation is that hopcalite “loses its effectiveness under the influence of moisture”, whereas the palladium formulation works dry, at 80 % RH and 23 °C, and at trace and percent-level CO alike. Moisture tolerance, not activity, is the discriminator.

Suracsh publishes SURSORB PAL, a palladium-impregnated pelletised carbon: CTC min 55 %, bulk density 0.550 ± 0.05 g/mL, hardness min 95, ash max 15 %. No CO conversion figure is published.

Silver — a media preservative, not a water disinfectant

Silver-impregnated carbon is sold on the oligodynamic effect: silver ions inhibit bacterial growth. What it does not do is disinfect water. The University of Kentucky Cooperative Extension’s guidance on home water treatment is unusually direct. Because such a filter contains a chemically active ingredient intended to inhibit microbial growth, the unit and its manufacturer must register with the EPA as a pesticide device before sale. Then the caveat, in the guidance’s own words: registration does not mean the unit is “in any way endorsed or approved by EPA as a water treatment device.” A registration number is a licence to sell, not an endorsement. On efficacy, it reports that “silver-impregnated AC made little difference when compared to untreated AC” on plate count bacteria, the advantage confined to roughly the first month of filter life.

Silver is a media preservative — it slows the colonisation of a wet bed by its own biofilm. Suracsh publishes SURSORB AQUA (granular, CTC min 50 %, moisture max 7 %, ash max 4 %) and SURSORB P-AQUA (pellet, CTC 60 min min 60 %), both explicitly not certified for potable water in the United States.

Permanganate on alumina

Potassium permanganate oxidises what it captures. Purafil’s Chemisorbant media is spherical pellets of activated alumina and binders, permanganate distributed through the pellet volume at a minimum 4 % KMnO₄ by weight, with published capacities per unit volume:

General Carbon’s GC PPA12 runs 12 % minimum KMnO₄ on 4–6 mm alumina spheres, positioned for polar or low-molecular-weight compounds where activated carbon is ineffective.

The selling point is irreversibility. Purafil describes a process combining adsorption, absorption and chemical oxidation in which contaminants are “converted into harmless solids” retained in the pellet, “eliminating the possibility of desorption”. In an occupied space, a medium that cannot give back what it took beats a higher-capacity medium that can.

One thing this article will not tell you: that activated carbon reduces permanganate, and that this is why alumina is used. The explanation is repeated throughout the trade and is chemically plausible, but we found no citable primary source. What can be said accurately is narrower. Alumina is the dominant commercial support; carbon-supported permanganate products do exist and are sold; and the documented manufacturer rationale is an application one — permanganate alumina is specified where carbon is ineffective.

Suracsh’s permanganate products are on alumina, not carbon: SURSORB BAL (spherical activated alumina, static adsorption min 170 mg/g, cavity volume min 0.4 cm³/g) and SURSORB BLEND, an alumina-and-carbon blend with permanganate on the alumina fraction.

What impregnation costs

One carbon, five impregnants, the same 7 wt % impregnation solution. Surface area fell by up to 96 %.

Zulkefli and co-authors, in PLOS ONE in 2019, impregnated one commercial carbon five ways, all at a 7 wt % solution. The raw carbon measured 901.04 m²/g; sodium carbonate left it untouched at 901.58; KOH took it to about 700, zinc acetate to about 350, copper sulfate to about 30. Their summary: the modified adsorbents’ surface area “decreased by up to 96 %”.

The DOE sulfur data explains why that spread is possible: at comparable loading, the BET surface area retained — and therefore the mercury capacity — varied by a factor of four on impregnation temperature alone, because low-temperature impregnation deposits ring-structured S₈ molecules that plug pores while high-temperature impregnation deposits shorter chains that spread as a thinner film (the full dataset is in the mercury article).

So the variable that matters is not how much impregnant you put on but where it ends up. Two carbons with identical loading percentages can differ by a factor of four in service, and nothing on either datasheet tells you which is which.

The counter-example, for balance: acid modification, as distinct from acid loading, can raise BET. In Frontiers in Environmental Science in 2023, 10 M nitric acid at 90 °C took surface area from 525.20 up to 581.27 m²/g, etching opening porosity faster than the acid filled it.

The safety section: caustic impregnation and self-heating

The mechanism has three legs, each independently sourced. First, the reaction is oxidative, not merely acid–base: H₂S on an alkaline surface is oxidised to sulfur and ultimately sulfate, releasing far more heat than neutralisation and continuing as long as oxygen and H₂S are supplied. Second, the impregnant catalyses secondary exothermic reactions — Technical Guide 12 states that with potassium hydroxide or caustic-based impregnants “these impurities catalyse secondary reactions” that generate localised hot spots, and records suppliers moving to copper-impregnated alternatives because caustic-based carbons are more problematic. Third, Calgon Carbon’s bulletin for ST 1X states that “the impregnants on this carbon have been shown to decrease the ignition temperature”, and that the product can “self-heat and ignite” in air at low convective airflow.

Together: exothermic catalytic oxidation, plus a lowered ignition temperature, plus loss of convective cooling, equals self-heating to ignition.

Airflow is the control variable. TG12 is explicit that under normal operation the air flow dissipates the heat of adsorption so that fire temperatures cannot be reached, and that at low flow it may take only a gentle breath of air entering the filter to start a fire. The dangerous states are shutdown, standby, storage and maintenance. TG12 records two incidents: a filter disconnected during maintenance that then smoked and caught fire; and oxygen injection into a carbon filter resulting in a fire within 24 hours. Its monitoring recommendation is that carbon monoxide is a better early indicator than temperature: a smouldering front is localised, so a thermocouple elsewhere reads normal long after CO evolution has begun.

Do not trust the published ignition temperature. This table, from an IChemE Hazards symposium paper, is the most useful safety item in the literature we reviewed:

The ASTM D-3466 figure overstates real-world exothermic onset by 200–300 °C. The paper’s caveat is that the value “should not be interpreted as the probable ignition temperature encountered in an application” unless conditions are similar. It also records a bed fire in fresh wood-based carbon with no adsorbed solvent at all — which ends any argument that adsorbed solvent is a precondition.

One attribution correction: the EPA chemical safety alert on carbon adsorption deodorising systems and the OSHA bulletin on carbon bed fires concern crude sulfate turpentine, mercaptans, ketones and aldehydes. Neither mentions caustic impregnation, NaOH, KOH or H₂S. Good citations for carbon bed fires in general; the wrong citation here.

What to take from this

Impregnation is a specification decision with three parts, and datasheets publish only the first: the chemistry says what the carbon removes, the loading sets the stoichiometric ceiling, and the impregnation process decides how much of that ceiling you reach. Where a figure does not exist we have said so rather than filling the cell. The full specification for every grade named here is published in HTML at suracsh.us, and every datasheet downloads without a form.

Sources

• Kiani et al., “Impregnation on activated carbon for removal of chemical warfare agents (CWAs) and radioactive content,” Environmental Science and Pollution Research 28(43):60477 (2021)
• General Carbon, “A Guide to Impregnated Activated Carbon”
• Choo, Lau, Mohamed & Lee, Journal of Engineering Science and Technology 8(6):741–753 (2013)
• US Patent 4,215,096 — caustic-impregnated carbon, loading range
• Environment Agency / Organics Recycling, Technical Guide 12, “Fire risk of activated carbon”
• Calgon Carbon ST 1X product bulletin
• Zulkefli et al., PLOS ONE 14(2):e0211713 (2019)
• Vidic, DOE report DE-FG22-96PC96212, University of Pittsburgh (1997) — sulfur impregnation and mercury capacity
• Li et al., Frontiers in Environmental Science (2023), doi:10.3389/fenvs.2023.976113
• “A Brief Review of Formaldehyde Removal through Activated Carbon Adsorption,” Applied Sciences 12(10):5025 (2022)
• Environmental Science and Pollution Research (2024) — SN2 mechanism and iodine capacity on impregnated carbons
• “Unravelling the potential of iodine isotopic exchange in CH₃¹³¹I capture by K¹²⁷I-impregnated activated carbons”
• Chebbi et al., Carbon Trends 7:100164 (2022)
• DEVCOM Chemical Biological Center on ASZM-TEDA and MIL-DTL-32101
• US Patent 4,623,637 — palladium/vanadium on alpha-alumina for CO oxidation
• Peterson et al., Nature Communications 5:4885 (2014), doi:10.1038/ncomms5885
• University of Kentucky Cooperative Extension IP-6, home water treatment
• Purafil Chemisorbant Media Bulletin
• General Carbon GC PPA12 — 12 % permanganate on alumina
• IChemE Hazards symposium paper — thermal stability and ignition temperature of activated carbon

Questions