Apr-2023

COS and mercaptans removal from gases

Challenges with removing trace sulphur species can be better understood and resolved with a new kinetic model validated by plant data in its ability to predict COS removal.

Prashanth Chandran, Harnoor Kaur, Jeffrey Weinfeld and Ralph Weiland
Optimized Gas Treating, Inc.

Article Summary

Gas and liquid hydrocarbon streams from refineries and gas plants must be well cleaned of sulphur compounds such as H2S, COS, and mercaptans (RSH), dictated mainly by environmental concerns. LPG can be subtle because although a copper strip test may indicate acceptable sulphur content today, COS slowly reverts to H2S and CO2 in the presence of water, so the same test administered tomorrow may fail.

This contribution offers a new model for COS absorption into alkaline solvents based on mass transfer rates enhanced by reaction kinetics – the first time commercial software has had the ability to simulate this aspect of COS absorption accurately. As part of the reported work, a compilation of plant performance test data on mercaptans removal is also presented. It shows plenty of room for improvement in reliability and accuracy of simulation tools.

Amines are excellent solvents for H2S, but, by and large, they are horrible for removing other less acidic, trace sulphur species such as COS and mercaptans. Until now, no simulator has been able to model COS and mercaptans adequately. With mercaptans, the basic problem seems to be insufficient, inaccurate phase-equilibrium data. Almost all the public domain mercaptans solubility data are academic in origin, which may explain why there is so little of it.

Very few academic institutions welcome researchers who handle mercaptans – academia is generally ill equipped to handle them safely and therefore avoids the risk associated with their malodorous and toxic nature. Good-quality data in the range of commercial interest are hard to come by. For COS, one of the main issues has been that simulators have ignored its reactive nature in aqueous amine solutions, treating its chemistry in an over-simplified way as a purely physically dissolved, non-reacting solute. The COS absorption rate is thus wrongly computed because the calculations fail to account for significant absorption rate enhancement that results from the chemical reactions of COS with non-tertiary amines.

Reactions of COS and mercaptans
Reaction kinetics of H2S and CO2 in aqueous amines are too well known to benefit from further discussion here. RSH merely dissociates in aqueous media. But to describe the decomposition of COS in water just by the reaction COS + H2O ® CO2 + H2S is deceptively oversimplified. The reaction mechanisms and kinetics of COS in amines are much more complex than that and can benefit from a brief explanation:

RSH ⇌ H+ + RS–     (1)
COS + H2O ⇌ H+ + HCO2S–     (2)
HCO2S– + H2O  H+ + HCO₃– + HS–     (3)

Reaction (1) is a simple dissociation involving a single hydrogen ion and, as such, is known to be essentially instantaneous. Thus, it is always at equilibrium. The limitation with RSH is that it is an extremely weak acid, so even a low level of acidification of the solvent will drive Reaction (1) back towards the formation of molecular RSH, and mercaptans have very low physical solubility in water. Significant acidification can be had even with a modest amount of dissolved CO2 or H2S. In regenerative caustic solutions, the CO2 and H2S spend the caustic from its intended purpose of RSH removal. The significance of these effects is discussed in the next section by looking at the vapour-phase profile of mercaptans in a typical absorber.

COS reacts in aqueous solutions first to form thiocarbonate Reaction (2), which further hydrolyses to bicarbonate and bisulphide Reaction (3). The combined form of Reactions (2) and (3) along with other speciation reactions of CO2 and H2S is equivalent to the overall simplified hydrolysis of COS to CO2 and H2S already mentioned. Reactions (2) and (3) are very slow unless a base is present in the solution to catalyse them. In the presence of amines, it is postulated that COS reacts by a base-catalysed mechanism according to:

COS + Am + H2O ⇌ AmH+ + HCO2S–     (4)
HCO2S– + Am + H2O  AmH+ + HCO₃- + HS–     (5)

In addition to these reactions, COS forms thiocarbamate with primary and secondary amines via a zwitterion mechanism:

COS + AmH+ ⇌ AmH + COS–     (6)
AmH+ COS– + B  AmCOS– (thiocarbamate) + BH+     (7)

Reaction (6) represents zwitterion formation (AmH+ in Reaction (6) stands for the primary or secondary amine with at least one mobile hydrogen). Reaction (7) describes the zwitterion’s deprotonation to thiocarbamate, AmCOS–. Any base, B, present in solution deprotonates the zwitterion. These reactions are responsible for quite significant COS absorption rates into primary and secondary amines, but they do not occur with tertiary amines. Reaction (4) is known to be equilibrium limited. The rate of the reverse reaction is observed to be practically zero for Reaction (5), thus indicating that, for any amine, COS will completely hydrolyse to CO2 and H2S in the fullness of time.

Thiocarbamate formation is significantly limited by the rate of deprotonation, Reaction (7). In fact, for several amines, the COS absorption rate is almost completely determined by the rate of deprotonation. This is unlike CO2, where the zwitterion deprotonation rate has much less influence on the overall conversion. As a result of these factors, the COS-amine reaction rate is much slower than amine-CO2.

Nevertheless, COS reaction rates are significant enough for a substantial fraction of the COS in a typical feed gas to be removed by primary and secondary amines. However, such is not the case for mercaptans beyond MeSH because they are very weak acids and easily displaced by co-absorbed CO2 and H2S.

Recently, we finished developing a COS absorption model that treats COS as a rigorous mass transfer rate-controlled component and incorporates it along with its reaction kinetics into the OGT ProTreat simulator. The model results were validated against some 20 proprietary sets of field-performance data for various amine systems. They showed the model accurately simulates COS removal in amine absorbers for the first time. What follows is a case study showing:
•    Mass-transfer and reaction-rate control in the COS removal model
•    A comparison between various amines’ performance for COS removal in a simple absorber
•    Comparison between simulation and actual plant performance in RSH removal
•    Summary of literature renditions of VLE in RSH-amine systems.

Because of the role played by reaction kinetics, different types of amines (primary, secondary, tertiary) have quite different COS removal effectiveness. For mercaptans removal, on the other hand, it is mainly the pKa of the amine that determines RSH removal – kinetics plays no role at all. Therefore, it makes sense to treat COS and mercaptans removal in separate ways.