Alumina-based Coating for Coke Reduction in Steam Crackers

Alumina-based coatings have been claimed as being an advantageous modification in industrial ethylene furnaces. In this work, on-line experimentally measured coking rates of a commercial coating (CoatAlloy™) have pointed out its superiority compared to an uncoated reference material in an electrobalance set-up. Additionally, the effects of presulfiding with 500 ppmw DMDS per H2O, continuous addition of 41 ppmw S per HC of DMDS, and a combination thereof were evaluated during ethane steam cracking under industrially relevant conditions (Tgasphase = 1173 K, Ptot = 0.1 MPa, XC2H6 = 70%, dilution δ = 0.33 kgH2O/kgHC). The examined samples were further evaluated using online thermogravimetry, scanning electron microscopy and energy diffractive X-ray for surface and cross-section analysis together with X-ray photoelectron spectroscopy and wavelength-dispersive X-ray spectroscopy for surface analysis. The passivating coating illustrated a better performance than the reference Ni-Cr Fe-base alloy after application of an improved pretreatment, followed by piddling changes on the product distribution. Presulfiding of the coating affected negatively the observed coking rates in comparison with the reference alloy, so alternative presulfiding and sulfur addition strategies are recommended when using this barrier coating.


Coke Formation Mechanisms
The carbon deposition process during thermal cracking in the presence, or not, of steam, is a quite complex phenomenon. It has been extensively described in literature [1][2][3][4][5] that it mainly consists of three mechanisms; catalytic coke formation, pyrolytic coke growth from existing carbon layers onwards and gas phase coking.
The initial phase of carbon deposition is the formation of a porous network of carbon filaments, catalyzed typically by Ni and Fe present at the surface of the reactor walls. It is widely accepted that catalytic carbon formation [6][7][8][9][10] on metallic surfaces involves surface reactions, diffusion and precipitation of carbon. As a first step, a hydrocarbon molecule is chemisorbed on a metal crystallite of the surface. Dehydrogenation of the R-CH groups takes place with the hydrogen atoms recombining and desorb into the gas phase. As a result, the carbon atoms are formed at the surface, dissolve in and diffuse through the metal particle. The carbon accumulation in the particle causes a pressure build up at the dislocations and the grain boundaries, which may exceed the tensile strength of the metal. Potentially, the metal particle is then lifted from the surface and carbon crystallizes at the rear end of the particle. Growing stems are developed that carry crystallites on their top. The precipitation of the carbon can give rise to structural deficiencies in the carbon lattice, thereby creating reactive carbon sites along this layer. Additional hydrocarbon molecules from the gas phase are incorporated at these sites causing addable filamentous coke growth. In that way, a porous layer of interwoven filaments is developed. The metal particles at the top of the whiskers are becoming more accessible from the gas phase, therefore the deposition rate increases, while the diffusion rate through the metal remains stable. Carbon migration over the metal surface occurs, surrounding the carbon stems. At this point, surface carbon can occur, encapsulating the metal and terminating any further coke growth.
Dissociative chemisorption of water molecules on the metal particles produces highly reactive oxygen atoms that react with surface carbon to from carbon monoxide, which desorbs in the gas phase. This reaction prevents the fast encapsulation of the metal tip of the filament and therefore the termination of this reaction, in other words the "deactivation of the metallic active site". How quickly these phenomena are occurring is a matter of relative kinetics of carbon growth, gasification, dissolution and diffusion. Certainly, the properties of the alloy are very important in all the mechanisms, however in this study the composition was kept stable to isolate the effect of roughness.
The heterogeneous non-catalytic mechanism, also known as pyrolytic coking, is the major source of coke in an industrial cracker, since it takes place over the complete run length. The coke in contact with the gas phase looks like a succession of several discrete layers, deriving from the coke-gas interface [11,12]. Coke formation is described by the reaction of gas phase precursors with active centers of the surface. Kopinke et al [1] have provided a general scale for the coking tendency of several precursors. For example, acetylene, anthracenes, cyclic naphthenes and aromatics possess a high tendency for coke formation. Ethylene is not the most reactive one, however due to its high concentration is very important. No significant differences of the relative constants of a particular hydrocarbon were found for different materials [1], supporting that the active centers are radical in nature and positioned in the coke matrix. The radicals can be generated by hydrogen abstraction from the partially dehydrogenated carbon layer being determined by the gas phase composition and the available coked reactor surface [13]. The coke radicals react by addition with unsaturated molecules and radicals from the gas phase. After dehydrogenation, graphitic layers are formed [14].
The last mechanism, also known as homogeneous non-catalytic coking results from a sequence of molecular and/or radical reactions in the gas phase. These lead to high molecular weight polynuclear aromatic compounds which remain solid even at high temperatures. As a result, they collide at the wall and integrate in the coke layer [15]. In this work, the latter mechanism contributes very little, since there are not a lot heavy compounds generated in the gas phase due the light feedstock (ethane).

Coking Rates and Coking Curve
The coke deposition on each sample is measured over time by continuously weighing the mass of the sample. This allows the determination of the total amount of coke after every cracking cycle, as well as the calculation of the initial catalytic coking rate and the asymptotic pyrolytic coking rate. Similarly with previous work [16], the coking rate is determined as: where Rc is the coking rate in kg·s −1 ·m −2 , mtj the mass of coke at time j in kg, tj the experimental time at instant j in s and S the surface area of the coupon in m 2 .
In this work, the mean value of the coking rate between 15 min and 60 min is defined as the initial or catalytic coking rate, i.e. characterizing the catalytic coking behavior of the sample. The asymptotic or pyrolytic coking rate is related to the pyrolytic coking mechanism and is reported as the mean measured coking rate between the 5 th and 6 th hour of cracking (see Figure 3). Figure S1. Processing sequence for the coking rate determination. Figure S2. Cross section build-up of the CoatAlloy™ passivating coating [16].