Mercury Removal Ahead of Cryogenic Units: Adsorbent Selection and Aluminium Exchanger Protection
Mercury in natural gas isn't a nuisance contaminant. It's a direct threat to plant integrity. Send it into the cryogenic section of an LNG or NGL facility and it goes straight for the aluminium brazed plate-fin heat exchangers through liquid metal embrittlement (LME) — a failure mode that can put a cold box beyond economic repair. The bill that follows: unplanned shutdown, extended lead times for exchanger replacement, and the safety hazard of a brittle fracture in a cryogenic, high-pressure environment. Getting mercury removal right at the design stage is one of the highest-leverage decisions a process engineer makes on a gas plant project.
Why Mercury Damages Aluminium Exchangers
The aluminium alloys used in cryogenic brazed plate-fin exchangers are susceptible to LME whenever elemental mercury contacts the metal surface, particularly under tensile stress. Mercury forms an amalgam with the aluminium, disrupting the oxide layer that normally protects the base metal. Once that oxide is compromised, crack propagation can run fast under the stresses imposed by thermal cycling and operating pressure. What you get is a brittle fracture with no prior deformation warning — it behaves nothing like the ductile failure modes engineers typically size against.
Even extremely low concentrations of elemental mercury in the gas phase are enough to initiate this mechanism. Speciation matters. Elemental mercury (Hg⁰) is the primary concern for LME, because it is the form that amalgamates. Where the outlet specification is set depends on the cold box design and operator risk tolerance — the consequence of exceedance is severe enough that it drives the need for a dedicated Mercury Removal Unit (MRU) positioned upstream of the main cryogenic exchanger, and in many designs a secondary guard bed positioned immediately ahead of the cold box.
Adsorbent Selection
Sulphur-Impregnated Activated Carbon (SIAC)
SIAC is the workhorse — the most widely deployed non-regenerative adsorbent for gas-phase mercury removal. Elemental mercury reacts with the sulphur impregnant to form mercuric sulphide (HgS), which is thermally stable and immobile at ambient and moderate temperatures. Capacity for elemental mercury is high, and the technology is commercially mature.
The limitations are worth knowing. SIAC is sensitive to liquid hydrocarbon carryover and free water. Liquid loading blocks pore access, reduces effective capacity, and can cause channelling. Heavy hydrocarbons co-adsorb and compete with mercury uptake sites. So upstream liquid separation — a correctly sized inlet scrubber or coalescing filter — is not optional. It is a prerequisite for reliable SIAC performance.
Nor is SIAC suitable where regeneration is required. Thermal regeneration mobilises the captured mercury and creates a concentrated mercury waste stream that requires specialised handling.
Metal Sulphide Adsorbents (Non-Carbon Supports)
Metal sulphide formulations on alumina or other inorganic supports offer improved resistance to liquid hydrocarbon contact compared with SIAC. They are the preferred pick where the feed gas carries heavier condensate fractions, or where slug flow from the upstream gathering system cannot be fully excluded. Mercury capacity per unit mass is generally lower than SIAC — account for that in bed sizing.
Silver-Based Adsorbents
Silver-on-alumina adsorbents capture mercury through amalgam formation. They are regenerable, which makes them attractive for high-mercury-loading streams where a non-regenerative bed would require frequent changeout. The trade-off: the regeneration step releases concentrated mercury vapour, so you need a condenser and mercury collection system. Extra capital, extra operational complexity. Silver-based systems are more commonly found in liquid-phase mercury removal (condensate treating) than in gas-phase applications ahead of cryogenic units.
Adsorbent Selection Summary
| Property | SIAC | Metal Sulphide (Inorganic Support) | Silver-on-Alumina |
|---|---|---|---|
| Primary capture mechanism | HgS formation | Metal sulphide reaction | Amalgam formation |
| Regenerable | No | No (typical) | Yes |
| Liquid HC tolerance | Low | Moderate | Moderate |
| Preferred application | Dry gas, well-separated feed | Wet or condensate-laden gas | High-loading or liquid streams |
| Waste mercury form | Stable HgS solid | Stable sulphide solid | Recovered liquid Hg |
Bed Sizing Principles
Bed sizing for a non-regenerative MRU is governed by three factors: inlet mercury loading, required outlet specification, and design service life before changeout.
It comes down to a mass balance across the bed:
Mass of mercury to be captured = (Inlet concentration − Outlet specification) × Volumetric flow rate × Design service life
Get the units consistent before the calculation proceeds. If inlet concentration is expressed in µg/Nm³, flow in Nm³/day, and service life in days, the result is in µg — convert to kg for bed mass calculation using adsorbent mercury capacity (kg Hg per kg adsorbent, from the vendor's isotherm data).
Inlet mercury concentrations in natural gas are notoriously variable and difficult to measure accurately — particularly the speciation split between elemental and organic forms. So bed sizing should incorporate a conservative multiplier on the design inlet loading. Relying on a single field measurement from a reservoir that has not been fully characterised is a recognised source of undersized beds and premature breakthrough.
Superficial velocity through the bed must be controlled. Too low and you get channelling; too high and you get excessive pressure drop or adsorbent attrition. Vendor-specific velocity limits apply and should be confirmed at the design stage.
For LNG applications, a lead-guard (or primary-polishing) two-vessel configuration is strongly recommended. The lead vessel carries the primary mercury load. The guard vessel provides a safety margin and can be sampled independently to confirm breakthrough has not propagated to the cryogenic section. When the lead vessel reaches end-of-life (confirmed by outlet sampling), it is replaced while the guard vessel continues to protect the cold box.
Operational Challenges and Hidden Failure Modes
Several mechanisms cause performance degradation that is not captured by routine pressure drop monitoring alone.
Mercury speciation shift: Organic mercury species present in the feed may not be efficiently captured by sulphur-impregnated adsorbents optimised for elemental mercury.
Liquid carryover: Inlet separation equipment must be maintained to the same standard as the MRU itself. Coalescing filter elements should be on a defined inspection and replacement schedule, not changed only on pressure drop.
Channelling: Mercury breaks through the channel while the bulk of the bed remains unsaturated. This makes outlet sampling the only reliable indicator of true bed status — pressure drop alone will not detect channelling.
Temperature excursions: SIAC capacity is temperature-dependent. Elevated inlet temperatures reduce equilibrium capacity. If the MRU is located downstream of compression without adequate cooling, the effective bed life will be shorter than the design basis.
Illustrative Sizing Scenario
This scenario is illustrative and does not represent a specific plant.
A gas plant processes a feed stream with a design inlet mercury concentration of 100 µg/Nm³ (elemental basis, confirmed by speciation analysis) and a required outlet specification of 0.01 µg/Nm³ ahead of the cold box. Flow rate is 10 × 10⁶ Nm³/day. Design service life for the lead vessel is set at 3 years (1,095 days).
Mercury to be captured = (100 − 0.01) µg/Nm³ × 10 × 10⁶ Nm³/day × 1,095 days ≈ 100 µg/Nm³ × 10 × 10⁶ Nm³/day × 1,095 days = 1.095 × 10¹² µg = 1,095 kg of mercury
Apply a design safety factor (typically 1.5 to 2.0, vendor-specific) to arrive at the installed bed mass. The guard vessel is sized independently using the same methodology for a shorter polishing service interval.
Pre-Design and Troubleshooting Checklist
- Obtain mercury speciation data (elemental, inorganic, organic fractions) — a total mercury figure alone is insufficient for adsorbent selection.
- Confirm inlet gas conditions: temperature, pressure, water dewpoint, and heavy hydrocarbon content at MRU inlet.
- Verify that inlet separation (scrubber and/or coalescing filtration) is sized and maintained to prevent liquid carryover to the adsorbent bed.
- Select adsorbent type based on feed gas cleanliness, not solely on mercury capacity.
- Size the lead bed on a conservative inlet loading with an explicit design safety factor; document the basis.
- Specify a lead-guard two-vessel arrangement for any LNG cold box application.
- Define an outlet sampling programme — frequency, analytical method, and detection limit — before commissioning. Pressure drop monitoring alone is not sufficient.
- Establish a changeout trigger based on outlet mercury concentration, not on elapsed time or pressure drop alone.
- Confirm spent adsorbent classification and disposal route before first loading — mercury-laden SIAC is a hazardous waste in most jurisdictions.
- For vessel entry, changeout, or inspection: full isolation, depressurisation, nitrogen purge, zero-energy verification, LOTO, continuous atmospheric monitoring for mercury vapour and hydrocarbons, and confined-space entry procedures are mandatory.
Conclusion
Protecting cryogenic aluminium exchangers from mercury-induced LME is non-negotiable in LNG and NGL processing. The engineering response is a correctly designed MRU — adsorbent selection matched to feed gas speciation and cleanliness, bed sizing built on a defensible mass balance with conservative loading assumptions, and a two-vessel lead-guard configuration that provides independent verification ahead of the cold box.
On a new project, the next step for a process engineer is to commission a mercury speciation study on representative wellhead samples before fixing the adsorbent type or bed size. For a brownfield troubleshooting exercise, the first action is to implement an outlet sampling programme with a detection limit well below the cold box specification — because without that data, the MRU's actual performance is unknown, regardless of how clean the pressure drop trend looks.