Carbon Capture Solvent Regeneration Column Molecular Sieve Polishing: MDEA-DEA Mixed Amine Carryover into 13X

The Failure Mode: Why Amine Carryover Kills 13X

Post-combustion CO2 capture using aqueous MDEA, MDEA-DEA blends, or piperazine-activated MDEA (PZ-MDEA) is now the default process for amine-based CCUS on coal-fired flue gas, cement kiln exhaust, and steel blast furnace gas. The capture train ends at a solvent regenerator column, where the rich amine is boiled at 105-125 degrees C and 1.5-2.5 bara to liberate the absorbed CO2 as product gas. The CO2 product leaves the top of the regenerator saturated with water vapor and, critically, with a measurable quantity of amine solvent that has either entrained as fine droplets or vaporized at the reboiler temperature.

Field experience from operating amine units shows that amine carryover from a properly designed regenerator is rarely below 5 ppmw and frequently lands between 15-40 ppmw in the gas phase. Foaming events from degraded amine (heat-stable salts, suspended solids, hydrocarbon contamination) push carryover to 80-120 ppmw for hours or days at a time. Every gram of amine that crosses the top of the stripper and reaches a 13X polishing molecular sieve bed is one gram of strongly basic nitrogenous liquid that reacts with the zeolite acid sites and permanently degrades the bed.

A 13X polishing bed placed directly downstream of the regenerator without protection will lose 30-60% of its CO2 working capacity within 6 months of normal service. The failure mechanism is not a simple loss of pore volume. MDEA and DEA oligomerize on the zeolite surface under regeneration temperature (110-130 degrees C), forming high-molecular-weight nitrogenous polymer that cannot be removed by thermal regeneration at any practical temperature. Once 15-20% of bed mass is converted to this polymer, the bed must be physically replaced. Industry data places the cost of a 13X polishing bed replacement on a 1 Mt/year CCUS plant at USD 380,000-520,000 for the adsorbent alone, plus 7-10 days of polishing bed downtime at 25,000 MT CO2/day throughput loss. For the broader CO2 capture adsorbent landscape and the chemical-solvent vs PSA cost trade-off that drives polishing bed selection, see our Molecular Sieve 13X for CO2 Capture: Flue-Gas PSA vs Chemical Absorption TCO comparison. For an in-depth look at how amine poisoning manifests on the molecular sieve surface during PSA regeneration, see Acid Gas Breakthrough: 13X H2S / SO2 Poisoning and Thermal-Swing Reversibility. For the upstream flue-gas H2O loading case where 13X polishing beds fail before amine carryover is even introduced, see Why Your Regeneration Cycle Keeps Getting Longer: 7 Root Causes Ranked. For comparison to upstream hydrogen-train dryer sizing where MDEA carryover from syngas treatment crosses 13X beds in blue-hydrogen service, see PEM Electrolyzer Stack Hydrogen Recirculation Dryer: 4A vs 5A at 35 bara DP.

Surface Chemistry of Amine-13X Reaction

The 13X zeolite crystal carries Brønsted acid sites on its internal surface in the form of residual Si-OH and Al-OH groups. These acid sites, while weak (pKa approximately 8-9 in liquid phase), are concentrated at the 8 angstrom pore mouth and provide reactive anchoring points for basic amine molecules passing through the bed. The reaction is fast: at 110 degrees C and an amine partial pressure of 0.1 Pa (corresponding to 1 ppmw in the gas phase), the residence time of an amine molecule in the 13X pore is 0.4 seconds, but the chemisorption half-life is approximately 18 seconds - so 95% of amine entering the bed is captured permanently.

The reaction proceeds in two stages. Stage 1 is a fast proton transfer that forms an alkylammonium aluminosilicate salt on the pore surface. Stage 2 is a slow oligomerization where the alkylammonium condenses with neighboring surface hydroxyls to form secondary and tertiary amines, which then couple with additional alkylammonium to build nitrogenous polymer chains exceeding 500 Da amide linkages. The polymer is brown, viscous at regeneration temperature, and completely blocks 8-12% of pore volume per 1000 hours of exposure. The surface area drops from the design 800 m2/g to 380-450 m2/g after 6 months of unprotected operation.

Why piperazine-activated solvents are worse than straight MDEA

Piperazine (PZ) is a cyclic secondary amine with a pKa1 of 9.8, considerably more basic than MDEA (pKa 8.5). When PZ-MDEA carries over at the typical 3-8 ppmw, each PZ molecule poisons 13X 2.5x faster than MDEA by mass because the secondary amine has two reactive N-H bonds rather than one. PZ also forms a tighter coordination complex with the zeolite Al site, requiring 50% higher regeneration temperature (180 vs 120 degrees C) to desorb even partially. Field data from 4 PZ-MDEA installations show 13X polishing bed life of 4-5 months without protection vs 6 months for straight MDEA, and the PZ protection case must use FTIR detection at the 1240 cm-1 PZ ring band in addition to the standard 1120 cm-1 C-N stretch used for MDEA.

Two-Stage Guard Bed Design (3A + Activated Alumina)

The field-proven layout is a two-stage guard bed placed immediately downstream of the regenerator overhead condenser and upstream of the 13X polishing adsorber. Stage 1 is a 6-inch (150 mm) deep bed of 3A molecular sieve, 8x12 mesh, operated at the same temperature as the regenerator overhead (100-115 degrees C). The 3A pore aperture of 3 angstroms admits water (kinetic diameter 2.6 angstroms) but rejects both MDEA (kinetic diameter 5.5 angstroms) and DEA (kinetic diameter 5.0 angstroms) on a steric basis. Stage 2 is a 12-inch (300 mm) bed of activated alumina, 1/4-inch spheres, BET 300-360 m2/g, for bulk water knock-down and isocyanate hydrolysis.

Why two stages rather than one? The 3A stage provides the molecular selectivity that prevents amine from reaching the 13X, but its water capacity is limited (approximately 18% by mass at 110 degrees C). The activated alumina stage acts as a water buffer, taking the bulk of the moisture load off the 3A so the 3A can be regenerated on a longer cycle (12-16 hours) without exhausting its amine-sieving capacity. The activated alumina also catalytically hydrolyzes trace isocyanates (R-N=C=O) that form as MDEA thermal degradation intermediates at 110 degrees C, converting them to carbon dioxide and primary amine that the 3A can capture more easily.

Vessel sizing worked example: 1 Mt/year capture plant

For a 1 Mt/year post-combustion capture plant operating at 1.4 Mt CO2/day nameplate and 95% online factor, the regenerator overhead gas flow is approximately 38,000 Nm3/hour at 100 degrees C and 1.8 bara. The guard bed vessel must be sized for a superficial velocity of 0.18 m/s through the bed, which sets the vessel inner diameter at 2,400 mm for a single vessel, or 1,700 mm for a parallel duplex (one operating, one regenerating). The 3A bed depth of 150 mm and activated alumina bed depth of 300 mm give a total guard bed depth of 450 mm. The guard vessel height is approximately 4,800 mm tangent-to-tangent to accommodate the bed plus 1,200 mm of top head space for distribution and 600 mm of bottom support beam.

Online Detection: FTIR, TN, DP Trending

Three online measurements are recommended for monitoring guard bed health before the 13X polishing bed reaches saturation. First, an FTIR analyzer at the guard bed outlet tuned to the C-N stretch band at 1120 cm-1 (MDEA) or 1240 cm-1 (piperazine) provides continuous amine detection with detection limit 0.1 ppmv. Second, a total nitrogen (TN) analyzer on the condensate from the guard bed knockout drum provides a slower but more reliable measurement with detection limit 0.5 ppmw. Third, differential pressure (DP) trending on the guard bed provides a mechanical saturation indicator independent of the analytical instruments - polymer buildup raises DP from 0.3 kPa (clean) to 1.5 kPa (saturated) within 8-12 weeks.

Alarm setpoints for the FTIR are tiered: warning at 0.3 ppmv amine (means regenerator foaming is increasing carryover - inspect regenerator trays and antifoam dosing), high alarm at 0.5 ppmv (means 3A bed amine capacity at 60% - schedule change-out within 4 weeks), and high-high alarm at 1.0 ppmv (means 3A bed exhausted - 13X polishing bed is being poisoned NOW, schedule immediate change-out). The DP alarm is similarly tiered: warning at 0.8 kPa, high at 1.0 kPa, high-high at 1.2 kPa. The TN alarm on the condensate is single-tier at 1 ppmw because the condensate sample lags the guard bed by 30-60 minutes.

In-Place Regeneration Limits

The 13X polishing bed, once amine-poisoned, can only be partially regenerated in place, and only in the early stage of poisoning. The procedure is documented in 8 field campaigns between 2020 and 2024. Stage 1 of the regeneration: raise regeneration temperature from 130 to 200 degrees C for 12 hours under N2 purge (50% of design flow) to volatilize free, unreacted amine from the pore surface. This step recovers 30-50% of lost capacity if the bed has less than 2000 hours exposure. Stage 2: cooldown to 80 degrees C and pass 5% O2 in N2 for 4 hours at GHSV 200 h-1 to oxidize surface oligomers - this raises CO2 working capacity by another 8-12%. Stage 3: resume normal regeneration at 130 degrees C.

Beyond 5% capacity loss or 2000 hours exposure, the polymer is graphitic and cannot be burned off without damaging the zeolite crystal structure. Treatment at 230 degrees C with 100% O2 begins to compress the zeolite framework (the 13X unit cell parameter a0 contracts from 24.7 angstroms to 24.4 angstroms within 48 hours), which reduces BET surface area from 800 m2/g to 560 m2/g - a 30% permanent capacity loss from regeneration damage alone. Once capacity drops 25% from the initial value, full bed replacement is required. Industry practice is to schedule replacement at 20% capacity loss to leave a 5% safety margin against unplanned shutdown.

Economic Impact: Compression Penalty & Replacement Cost

A failing 13X polishing bed imposes three direct costs on the CCUS plant operator: (1) replacement adsorbent cost, (2) polishing bed replacement downtime, and (3) downstream CO2 compression penalty from elevated H2O content in the product. For a 1 Mt/year capture plant, the adsorbent cost is USD 380,000-520,000 for a full 13X polishing bed (approximately 65 MT at USD 6,000-8,000/MT typical 2026 pricing for 8x12 mesh 13X HP grade). The bed-replacement outage is 7-10 days at 25,000 MT CO2/day throughput, equivalent to 175,000-250,000 MT of lost CO2 capture revenue at USD 80/MT CO2 offtake value (2026 EU ETS plus offtake premium), totaling USD 14-20 million of foregone revenue per avoided bed change-out.

The compression penalty from elevated H2O is more subtle but no less real. Each 10 ppmv of residual H2O in the CO2 product depresses downstream pipeline CO2 dewpoint by approximately 4 degrees C. In a 70-bar sequestration pipeline, a dewpoint depression of 10 degrees C raises the recompression work by 0.10-0.15 kWh per kg of compressed CO2, equivalent to 0.8% of compressor power for the 1 Mt/year plant. A 100 ppmv H2O excursion sustained for 30 days adds roughly 280 MWh of avoidable compression work, equivalent to 110 MT of CO2 emissions from the parasitic compression load - a perverse outcome for a carbon capture facility.

Alternative Guard Materials: Why Activated Carbon Fails

Engineers sometimes propose activated carbon as a guard bed material on the basis of its high surface area (900-1100 m2/g) and apparent amine affinity (10-15% by mass at saturation). Activated carbon alone is NOT recommended as a guard bed for amine carryover in CCUS service. The failures are documented across 4 case studies between 2015 and 2023. Failure mode 1: activated carbon catalytically converts MDEA to dimethyldiethylammonium oligomers at 110 degrees C. The oligomer is a sticky, nitrogenous tar that fouls downstream piping and the 13X polishing bed more aggressively than the parent amine. Failure mode 2: the carbon surface is not regenerable at temperature - thermal swing regeneration at 200 degrees C leads to 8% mass loss per cycle from carbon-steam reaction (the carbon gasifies). Failure mode 3: the carbon has 10x higher pressure drop than 3A at equivalent bed depth, requiring larger vessels.

The combination of these three failure modes makes activated carbon a poor economic choice over a 5-year operating horizon. A 3A + activated alumina combination has lower pressure drop (0.3 vs 3.0 kPa at design velocity), is fully regenerable (the alumina wash extends to 5+ years at water-only service), and costs USD 38k per initial fill vs USD 65k for the equivalent carbon guard. Over 5 years, the activated carbon guard is projected to cost USD 415k in replacement and downtime vs USD 145k for the 3A + alumina guard - a USD 270k penalty that buys worse protection.

Procurement Specification Template

A typical procurement specification for a 3A + activated alumina guard bed for amine carryover service in CCUS should reference the following standards and acceptance criteria. The 3A sieve must conform to ASTM D5357 sampling plan with certificate of analysis for crystal phase (verified by XRD as 3A, no detectable 4A or 5A impurity), particle size distribution (8x12 mesh, less than 0.5% fines below 8 mesh per ASTM E11), bulk density (0.70-0.75 g/cm3), water capacity at 25 degrees C and 75% relative humidity (greater than 20% by mass per ASTM D1413), crush strength (greater than 12 N per bead per ASTM D4179), and attrition loss (less than 0.05% per ASTM D4058).

The activated alumina must conform to ASTM D5306 sampling plan with BET surface area 300-360 m2/g (ASTM D3663), pore volume 0.45-0.55 cm3/g (ASTM D4641), crush strength greater than 8 kg per 1/4-inch sphere (ASTM D4179), and attrition loss less than 0.1% (ASTM D4058). Both products must be supplied in 1 MT sealed (heat-sealed polyethylene liner) steel drums with lot traceable Certificate of Analysis dated within 60 days of shipment, and must carry a 24-month shelf life warranty from the date of manufacture. Request SDS from All-Inclusive CCSDS in advance of shipment for plant HSE review.

Next Steps

If you are sizing a new CCUS polishing train or diagnosing an existing 13X polishing bed failure, the fastest path is a 30-minute engineering call with our team. Send the 6 data points above plus your current polishing bed CO2 working capacity (kg CO2 / 100 kg sieve) and any guard bed installation details, and we will return a recommended guard bed sizing, FTIR/DP monitoring setpoints, and a 5-year operating cost projection within 48 hours.

Sample quantities of 3A sieve (1 kg, 8x12 mesh) and activated alumina (1 kg, 1/4-inch spheres) are available for laboratory amine-loading qualification testing. Standard production lead time is 15-25 days for orders up to 50 MT. Engineering documentation package (PFD, P&ID, vessel datasheet) is included at no charge with every guard bed quote.

Frequently Asked Questions