Siloxanes in Biogas: Engine Damage, Limits and Carbon Beds

Deposits from eight landfill gas engines, analysed by Konkol and co-workers, came back at roughly 81 % silicon in the exhaust manifold, 68 % on the piston face and 60 % in the combustion chamber. On untreated biogas, the same study found operation limited to around 100 hours before lubricant degradation. That is what siloxanes do, and why every serious biogas gas-quality specification names them.

The short version

• Measured engine deposits from landfill gas ran about 81 % silicon at the exhaust manifold, with untreated biogas giving roughly 100 hours before lubricant degradation (Konkol et al., Materials 15:2408).
• Concentration depends entirely on feedstock: EPA reports 0–400 mg/m³ for wastewater digesters, 0–50 for landfills, 0–0.2 for livestock and dairy.
• Three unit bases — mg/m³ total siloxane, mg Si/m³ and ppmv — are routinely conflated. D4 and D5 are 37.9 % silicon by mass, so the bases differ by about 2.6×.
• Published equipment limits span nearly three orders of magnitude — about 900 to 1 — from Caterpillar at 28 mg/m³ to Capstone microturbines at 0.03 mg/m³; SoCalGas Rule 30’s siloxane trigger level is 0.01 mg Si/m³, on a silicon-only basis.
• Siloxane-loaded carbon is a throwaway: the retained species polymerise into heavier polydimethylsiloxanes that block pores and defeat thermal regeneration.

What siloxanes are, and how they get into the gas

Siloxanes are silicon–oxygen compounds with methyl groups on the silicon. The species that matter in biogas are the cyclic volatile methylsiloxanes D3, D4, D5 and D6, the linear L2–L5, and trimethylsilanol. ASTM D8230-19 covers nine target compounds, “spanning trimethylsilanol through dodecamethylcyclohexasiloxane (D6)”.

They arrive from consumer and industrial products. The Hinkley Center’s review describes their extensive use in health care, personal hygiene and industrial products — cosmetics, lotions, paints and waterproofing — with shampoos and detergents carrying them into sewers.

In landfill gas the distribution is skewed: SCS Engineers reports D4 as the largest single contributor at about 60 % of the total, followed by L2, D5 and L3 — corroborated in the Energies review by Nkuna and co-workers.

How much is actually in the gas

The single most citable set is EPA’s, from An Overview of Renewable Natural Gas from Biogas (456-R-24-001):

The dairy row matters most to project developers: GTI’s California field testing found no detectable siloxanes in dairy digester gas, so on a manure feedstock siloxane treatment may be a question you close out with data rather than equipment.

Field measurements support the spread: roughly 10–160 mg/m³ across 28 US landfill facilities (SCS), and 9.5 ± 0.4 mg/m³ total at a Polish landfill.

One note on the literature: SCS states that siloxane concentrations are generally higher in digester gas than landfill gas, while the Energies review compares landfill and sludge gas against agricultural feedstock. The two comparisons are not interchangeable — municipal sludge digestion is the high-siloxane case, agricultural and dairy the low one.

Three units, three different numbers

This is where trade literature does the most damage. Siloxane concentrations circulate in three bases: mg/m³, the mass of the whole molecule; mg Si/m³, the mass of the silicon atoms only; and ppmv, a species-specific volume fraction.

D4 and D5 are 37.9 % silicon by mass; L2 is 34.6 %; trimethylsilanol is 31.1 %. So 1 mg/m³ of total siloxane is roughly 0.35–0.38 mg Si/m³ — about 2.6× between the two most commonly quoted bases. On a volume basis, 1 ppmv of D4 is roughly 12.1 mg/m³, or about 4.6 mg Si/m³, at 25 °C and 1 atm. That is straightforward arithmetic from molecular weights, but somebody has to actually do it.

There is a second trap. Engine specifications are frequently written per Nm³ of methane, not per Nm³ of raw gas. At 50 % methane landfill gas, a limit per Nm³ CH₄ is roughly half that value per Nm³ of the gas your analyser sees, so two engineers can quote the same OEM limit and be a factor of two apart without either being wrong.

What silica actually does to the machine

Siloxanes combust to silicon dioxide: amorphous, abrasive silica that deposits wherever the gas goes.

From shampoo to seized turbine wheel. The deposit compositions are measured values from eight landfill gas engines.

Reciprocating engines. The Environment Agency and SEPA guidance for landfill gas engines, sourced from engine OEMs, describes cylinder-head deposits as “white to light grey, somewhat laminar”, severely reducing engine life and requiring manual scraping from pistons, heads and valves. Silica also partitions into the lubricating oil, forcing shorter oil-change intervals; Konkol found it reducing piston-to-liner separation, with abrasive wear of pistons, rings and bearings. Caterpillar adds exhaust valve deposits, valve guttering, contaminated spark plugs, turbocharger turbine wheel buildup and muffler residue — and states that “high-compression-ratio engines … are less tolerant of siloxane contamination”.

Turbines and microturbines. SCS describes progressive performance loss as silica builds in the combustor and recuperator, until a larger mass breaks off and seizes the turbine wheel — a step failure, which is why microturbine limits sit nearly three orders of magnitude below the most tolerant engine limits.

Fuel cells and aftertreatment. Argonne reports that 10 ppm of D5 caused total SOFC failure in 30 hours at 1,000 °C; CCST notes silica clogging catalytic fuel processing reactors and porous electrodes. SCS reports SiO₂ deposits deactivating catalysts in hours or days, and CCST adds that SCR catalysts are highly susceptible to silica fouling of the bed pores.

One damage mode we will not claim: siloxane poisoning of lambda or oxygen sensors. It is widely repeated in trade material, but we found no biogas-specific OEM or peer-reviewed source for it.

What it costs when you get it wrong

• A US survey put the annual O&M cost of siloxane damage at $17,300 to $190,400 per year, depending on capacity across the 10–2,900 acfm range.
• At Calabasas Landfill, microturbines on 2 mg/m³ inlet gas carried a carbon operating cost of 0.3 ¢/kWh against $85/kW of carbon capital. A Waukesha engine on 34 mg/m³ inlet gas ran 1.5 ¢/kWh — described in the SCS paper as excessive.
• At the Trecatti landfill in the UK, activated carbon was being changed weekly, at nearly €2,000 per change.

Carbon cost scales with inlet loading, and at high loading the media bill stops being a consumable line and becomes the reason the project does not work.

Published limits — and why they are all secondary citations

The primary documents — INNIO Jenbacher TI/TA 1000-0300, MWM Technical Circular 0199-99-3017, Caterpillar’s gaseous fuel application and installation guides, Capstone document 410002 — are customer and dealer documents, not publicly downloadable. Lucideon confirms the first two as governing test specifications but publishes no values. Everything below is a secondary citation: a starting point for a conversation with the OEM, not a contract limit.

About nine hundred to one between the most and least tolerant equipment — Caterpillar at 28 mg/m³ against Capstone at 0.03. The two pipeline tariff values are on a silicon-only basis, not total siloxane, so they are not directly comparable with the equipment limits on the same axis.

The SCS paper does not state whether its mg/m³ figures are total siloxane or silicon-only — resolve that before designing to them. CCST’s own conclusion is that California’s specification is more stringent than most manufacturer requirements.

Measuring siloxanes, and why two labs disagree

ASTM D8230-19 exists as of 2019 — a standard test method for volatile silicon-containing compounds in gaseous fuel by gas chromatography with spectroscopic detection. It specifies the sampling media (fused-silica-lined stainless canisters, 14-day hold; Tedlar bags, 72-hour hold; or dual-bed hydrophobic carbon sorbent tubes), a five-point minimum calibration and per-component precision limits.

Before it, there was no common method. GTI names the root cause: “Variations to ‘in-house’ methods developed by individual laboratories” in sampling media, instrumentation, target analytes, reporting units and QC.

Eurofins’ case study quantified the consequence:

• Impinger samples gave D4 and D5 concentrations 2.5 to 6 times higher than canister results from a different laboratory.
• Canister recoveries of D4, D5 and D6 were erratic over 7 days; methanol impingers held above 80 % over 21 days.
• “Sample bags utilizing silicone septa and adhesives resulted in high blank levels” — the sampling hardware is itself a siloxane source.

The same uncertainty reaches the regulator: biomethane developers argued to CCST that California’s specification sits below reliable detection limits, making projects hard to finance because of the perceived risk of shut-in on measurement error, and several laboratories claim detection limits of 0.1 mg Si/m³ or lower, but CCST stated it had “not been able to independently test these claims”.

One detail captures the state of the art: the Canadian Gas Association’s 1 ppmv siloxane level was derived by taking an agreed 0.5 ppmv detection limit and doubling it. That is a specification derived from measurement capability, not from engine damage.

What activated carbon does — and the one thing it cannot do

Capacity, at laboratory conditions. The best single dataset, from Cabrera-Codony and co-workers in Environmental Science & Technology, screened twelve commercial carbons. The best performer, a wood-based chemically activated carbon, reached 1,732 ± 93 mg/g for D4 at 1,000 ppmv in dry nitrogen. That is a ceiling, not a design basis: at biogas-typical D4 concentrations the capacity halved, and real biogas conditions — methane, CO₂, humidity — reduced it further.

Capacity, at engineering conditions. Papurello’s figures are the ones to design against: C64 carbon at 37.3 mg/g and CKC at 25.2 mg/g, removing 4.59 mg/m³ of D4 at 25 °C, with D4 dosed at 1, 2, 5 and 10 ppmv in a 60/35 vol% CH₄/CO₂ carrier at a GHSV of 11.9 kh⁻¹. A non-activated biochar managed 3.5 mg/g. Capacity tracked surface area and pore volume rather than elemental composition; C64 outperformed CKC by 48 %.

Pore structure. Papurello identifies 1–2 nm micropores and pores below about 2.5 nm as optimal. Gaj notes that many researchers point to a 1.7–6.0 nm range for molecules roughly 1 nm across, against typical carbon average pore diameters of 1.4–2.0 nm.

Humidity is brutal. Siloxane adsorption falls tenfold or more at 50–70 % relative humidity — moisture is the single most significant factor, and water must be removed upstream of the carbon.

VOCs win on mass. In a real sewage biogas stream, hexane was present at 375 mg/m³, limonene at 220 and toluene at 24, against D4 at 54 and D5 at 102 — roughly four to one in favour of the VOCs. Carbon is not selective; H₂S, halides, ammonia and water vapour take the same sites.

The hard limit. Cyclic siloxanes do not simply adsorb — they polymerise on the carbon surface. Gaj describes the “polymerization of cyclic VMSs into less volatile, heavier and hardly desorbable polydimethylsiloxanes” which “block the adsorbent’s pores and prevent its effective thermal regeneration”. Oxidative regeneration fails for a different reason: the product is SiO₂, which also blocks pores. Cabrera-Codony observed D4 polymerisation on all twelve carbons after prolonged contact, the extent correlating with phenolic and carboxylic surface groups.

The contrast case proves the mechanism rather than merely asserting it: polymer resins regenerate effectively at around 100 °C with no evidence of siloxane polymerisation, and activated alumina behaves the same way. Regeneration is not hard in general — this particular chemistry happens on this particular surface. The siloxane bed is a throwaway.

Why a siloxane bed is not sized like an H₂S bed

No published source states this contrast directly, and no published EBCT guidance for siloxane duty could be found. What follows is derived reasoning with each input separately sourced — a framework for asking better questions, not a design rule.

1. Loading differs by two to three orders of magnitude. Siloxanes arrive at roughly 1–160 mg/m³; H₂S at 10–3,000 ppm and above, i.e. hundreds to thousands of mg/m³. The H₂S bed is mass-limited; the siloxane bed is not.
2. Working capacity is inverted relative to that. Impregnated and catalytic carbon holds roughly 20–25 wt % H₂S — Calgon’s ST 1X bulletin states approximately 25 % by weight — while real-biogas siloxane capacity is roughly 2.5–3.7 wt % from Papurello’s figures. Both of those are gravimetric, per unit mass of carbon; the design-literature bands quoted for the same duties, such as 0.25–0.35 g H₂S/cc for catalytic carbon, are volumetric, per unit packed volume, and the two convert into one another only through the apparent density on the same datasheet — see why H₂S capacity figures are not comparable. The H₂S bed is a reactor consuming its own volume of reagent; the siloxane bed is a physical adsorber sized on contact time and mass-transfer-zone shape.
3. The two mechanisms want opposite gas conditions. Catalytic H₂S carbon requires free oxygen and RH above 10 %, per Calgon’s own bulletins; siloxane adsorption collapses with humidity. In practice the H₂S stage runs wet and the siloxane stage runs after chilling — a plant layout consequence, not a media selection one.
4. The endpoints differ by orders of magnitude. H₂S breakthrough is tolerated at ppm levels. A microturbine limit of 0.03 mg/m³ or a pipeline trigger of 0.01 mg Si/m³ is a near-zero endpoint, so mass-transfer-zone length, not equilibrium capacity, sets bed depth.

The alternatives, and why carbon survives anyway

Chilling condenses only about 26 % of siloxanes at −25 °C and needs about −70 °C for above 99 % — energy-intensive, and still short of a microturbine or pipeline specification on its own. Silica gel has good capacity but requires the gas below 10 % relative humidity. PDMS membranes are selective for siloxanes but are damaged by other biogas contaminants and need frequent replacement.

That is why carbon persists as the polishing step, and why Environment Agency guidance lists the standard configurations as activated carbon plus chilling, and activated carbon plus a heat exchanger.

Where Suracsh stands on this

Plainly: the SURSORB grades carrying a biogas application — CAG and MAG — along with PAG-10, PAG, ALKA, P-ALKA and PAM, are specified for hydrogen sulfide, acid gas and ammonia duty. Suracsh does not currently publish a siloxane capacity figure for any grade, and nothing here should be read as a siloxane performance claim for a Suracsh product.

The physically relevant starting point is the non-impregnated range. SURSORB SA is a coconut-shell granular carbon published at BET 1000–1500 m²/g with ball pan hardness min 98. The SURSORB PAC pellets publish both CTC and iodine number for the same carbons — PAC-900 at CTC min 50 % and iodine min 950 mg/g, PAC-1050 at min 60 % and 1050 mg/g, PAC-1250 at min 75 % and 1250 mg/g — so pore development can be reasoned about from published numbers.

There is also a specification reason not to make one bed do both duties. Gaj notes that alkaline impregnation reduces the adsorbent’s active surface area and pore volume, which reduces its capacity for volatile methylsiloxanes: the impregnation that makes a carbon good at H₂S makes it worse at siloxanes. Two beds, two media, two sets of gas conditions.

Every grade named here publishes its full specification in HTML at suracsh.us, and every datasheet downloads without a form or an NDA. The selection guide maps contaminant to grade to specification, and our companion article on why H₂S capacity figures are not comparable covers the upstream stage.

Sources

• Konkol et al., Materials 15(7):2408 (2022) — Biogas pollution and mineral deposits on landfill gas engines
• EPA 456-R-24-001 — An Overview of Renewable Natural Gas from Biogas (Jan 2024)
• SCS Engineers — Siloxanes in Landfill and Digester Gas Update, 27th SWANA LFG Symposium (2004)
• Tansel & Surita — Hinkley Center Report #10329 (May 2016), Siloxanes in Biogas; project number 2014-13
• Nkuna et al., Energies 13(22):6088 — Biogas upgrading approaches with special focus on siloxane removal
• Gaj, Energies 13(10):2605 — Adsorptive biogas purification from siloxanes: a critical review
• Papurello et al., Processes 7(10):774 (2019) — D4 removal on commercial activated carbons
• Cabrera-Codony et al., Environ. Sci. Technol. — Optimal activated carbon properties for siloxane removal
• Santos-Clotas et al., Chemosphere 228:242–251 (2019)
• Environment Agency / SEPA EP165 Part 1 — Gas treatment technologies for landfill gas engines
• Caterpillar — Biogas Comes of Age (white paper)
• Lucideon — Jenbacher TI 1000-0300 / MWM TC 0199-99-3017 testing
• Papadias, Ahmed & Kumar — ANL/CSE/FCT/FQ-2011-11, Fuel Quality Issues in Stationary Fuel Cell Systems
• CCST — Biomethane in California Common Carrier Pipelines (2018) · CPUC workshop slides
• SoCalGas Rule 30 tariff
• ONE Gas / Texas Gas Service — RNG Quality and Testing Standards
• ASTM D8230-19 · GTI ASTM D8230 siloxanes update (2020)
• Eurofins — A summary of available analytical methods for the determination of siloxanes in biogas
• Calgon Carbon — ST 1X product bulletin

Questions