Industrial scale experience on steam reforming of CO2-rich gas

Applied Catalysis A: General 495 (2015) 141–151

Contents lists available at ScienceDirect

Applied Catalysis A: General journal homepage: www.elsevier.com/locate/apcata

Review

Industrial scale experience on steam reforming of CO2 -rich gas Peter Mølgaard Mortensen ∗ , Ib Dybkjær Haldor Topsoe A/S, Nymøllevej 55, DK-2800 Lyngby, Denmark

a r t i c l e

i n f o

Article history: Received 28 November 2014 Received in revised form 12 February 2015 Accepted 13 February 2015 Available online 23 February 2015 Keywords: Carbon formation CO2 CO2 -reforming Dry reforming Industrial scale reforming

a b s t r a c t The following article summarizes experience on application on industrial scale reforming of CO2 -rich gas, showing how nanoscale science and detailed catalyst information have been bridged to large scale reforming plants. Reforming of methane with CO2 alone (“dry methane reforming”, DMR) is closely related to steam methane reforming (SMR), and reaction mechanism and kinetics are comparable in the two reactions. This implies that much of the knowledge from SMR can be applied on DMR as well, including catalyst development. The primary challenge of reforming of CO2 -rich gas is carbon formation, as the low H/C ratio of the feed implies that a high potential for carbon formation exists. Thus, catalysts resistant to carbon formation are required; where noble metals, partly passivated nickel catalysts, and promoted nickel catalysts have good potential. In an industrial perspective, reforming of CO2 -rich gas will require a co-feed of water to decrease the severity of the gas for carbon formation and for conversion of any higher hydrocarbons. Use of traditional nickel catalyst has been demonstrated at industrial scale of dry synthesis gas production up to 133,000 Nm3 /h, but this requires a co-feed of large amounts of water. Better success has been demonstrated with the SPARG (sulfur passivated reforming) process or noble metal catalysts, where large-scale operation has been done under very severe conditions to produce synthesis gas with a relative low H2 /CO ratio. © 2015 Elsevier B.V. All rights reserved.

Contents 1. 2. 3. 4. 5.

6.

7.

8.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Thermodynamic considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reaction mechanism of dry reforming of methane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carbon deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Deactivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Sulfur poisoning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2. Sintering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Catalysts for dry reforming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1. Catalyst support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2. Catalyst promotion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Industrial experience with large scale reforming of CO2 -rich gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1. Nickel catalysts for CO2 -reforming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2. SPARG process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.3. Noble metal catalysts for CO2 -reforming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

141 142 143 144 145 145 145 145 146 146 146 148 149 150 150 150

1. Introduction

∗ Corresponding author. Tel.: +45 2552 9483. E-mail address: [email protected] (P.M. Mortensen). http://dx.doi.org/10.1016/j.apcata.2015.02.022 0926-860X/© 2015 Elsevier B.V. All rights reserved.

Synthesis gas production is one of the largest industries in the world with its development stemming back from 1930 [1]. Important bulk chemicals such as hydrogen, ammonia, and methanol are produced on the basis of this, but also synthetic fuels (via methanol

142

P.M. Mortensen, I. Dybkjær / Applied Catalysis A: General 495 (2015) 141–151

or by the Fischer–Tropsch synthesis), acetic acid, and other commodity chemicals are produced on this platform. Among the reforming reactions, steam reforming of methanerich natural gas (SMR) must be considered as the principal reaction for production of synthesis gas and hydrogen. Steam reforming of CO2 -rich gas (in the following referred to as “CO2 -reforming”) or reforming of methane-rich feedstock with carbon dioxide alone (“dry methane reforming”, in the following referred to as DMR) are receiving much attention as they in theory offer ways of using CO2 , which in many industries is considered as a waste product and environmentally is a polluting greenhouse gas. Notice, however, when considering use of CO2 in reforming, that the potential impact on CO2 emission is not very significant [2]. Making a simple back of the envelope calculation will show that if all carbon in the yearly methanol demand of close to 65 million tons/year in 2013 [3] (a top 10 worldwide commodity chemical) were assumed to come from CO2 , a net capture of 89 million tons/year could be achieved. Comparison of this with the annual CO2 emission of 34,500 million tons in 2012 [4] shows that it would only correspond to 0.26% reduction in CO2 emission. CO2 -reforming or DMR related to the production of chemicals should therefore not be done with focus on greenhouse gas reduction, but rather be considered as a means to process hydrocarbon feeds with high CO2 contents or processing of inexpensive CO2 (waste) streams. Many natural gas resources and biogas come with a natural high content of CO2 [5,6], raising request for direct reforming of these. CO2 -reforming also offers a direct route to production of synthesis gas with a relatively low H2 /CO ratio (in the order of 1 or lower), which is suitable for production of e.g. higher alcohols or acetic acid [7]. One of the primary tasks in the development of CO2 -reforming or DMR is to find operating conditions in combination with a suitable catalyst to avoid carbon formation. Nickel, cobalt, and noble metal catalysts have been tested as catalyst for CO2 -reforming, with nickel being the most investigated system, as this is the conventional choice in SMR and a relatively cheap catalyst in comparison to noble metals [7]. Haldor Topsoe A/S has many years of experience in reforming. The first startup of a steam reformer designed by Haldor Topsoe dates back to 1956. In the current work, our experience within the field of CO2 -reforming and DMR is summarized, starting with a short review on the science behind the CO2 -reforming process in industrial scale. The review section describes aspects of thermodynamics, reaction mechanism, deactivation, and catalysts. The discussion will to a large extent originate in the SMR reaction and extrapolate to CO2 -reforming and DMR. This is done because SMR is better understood and the three reactions in reality are closely connected. The key point of the current work is not to give a detailed review of CO2 -reforming and DMR, but only to supply the background information needed to understand the work done to bridge

the gap between laboratory research and full scale plant operation. For detailed reviews, references are instead made to Papadopoulou et al. [6], Fan et al. [8], or Bradford and Vannice [9]. Also, the autothermal reforming (ATR) technology, where a hydrocarbon feed is reacted at high temperature with steam and oxygen, is a route which may be used to reform larger amounts of methane with carbon dioxide. However, ATR will not be the focus in this article as this approach is somewhat different from more traditional tubular reforming. For information on the ATR technology, see references [10–12]. 2. Thermodynamic considerations Traditional SMR is the endothermic reaction between steam and methane at elevated temperatures to produce synthesis gas: CH4 (g) + H2 O(g)  CO(g) + 3H2 (g)

(1)

DMR takes place in a similar way, but with CO2 : CH4 (g) + CO2 (g)  2CO(g) + 2H2 (g)

(2)

The stoichiometric DMR reaction yields a synthesis gas with a H2 /CO ratio of 1, compared to 3 for SMR. The close relation of the two reactions can be illustrated by considering the reaction equation: CH4 (g) + xH2 O(g) + (1 − x)CO2 (g)  (2 − x)CO(g) + (2 + x)H2 (g) (3) where x = 1 is SMR (reaction 1) and x = 0 is DMR (reaction 2). Besides the reforming reactions, also water gas shift (WGS) takes place during reforming; specifically reverse water gas shift (RWGS) is relevant for DMR: CO2 (g) + H2 (g)  CO(g) + H2 O(g)

(4)

The reforming reactions and the RWGS are all endothermic reactions, and temperatures in the order of 750 ◦ C are required for an effective conversion of methane and carbon dioxide [8,13]. Fig. 1 shows equilibrium composition of the product gas as a function of temperature for the DMR reaction at CO2 /CH4 feed ratios of 1 and 2. High conversions require high temperatures, due to the endothermic nature of the DMR reaction. Methane conversion is additionally increased slightly by operating with excess CO2 . This will also influence the H2 /CO ratio of the product, which drops with excess CO2 due to a larger amount of CO from the RWGS in the product. Besides the endothermic reforming reactions, the choice of operating conditions is additionally influenced by the Boudouard

Fig. 1. Equilibrium gas composition as a function of operating temperature at CO2 /CH4 feed ratio of 1 (a) or 2 (b) at a total pressure of 20 bar. Calculations were done using Excel on the basis of thermodynamic data for Eqs. (2) and (4).

P.M. Mortensen, I. Dybkjær / Applied Catalysis A: General 495 (2015) 141–151

reaction (also called CO disproportionation), the CO reduction reaction, and the methane decomposition reaction: 2CO(g)  C(s) + CO2 (g)

(5)

CO(g) + H2 (g)  C(s) + H2 O(g)

(6)

CH4 (g)  C(s) + 2H2 (g)

(7)

as these may be responsible for carbon formation during the reaction. Generally, the tendency for carbon formation will increase with decreasing H2 O/CH4 and CO2 /CH4 ratios of the feed gas and increasing temperature [14]. However, it should be emphasized that carbon formation is a non-linear phenomenon and prediction of the risk for carbon formation will require a thermodynamic evaluation for a given feed composition. The reforming reactions are favored at low pressure due to the Le Chatelier’s principle (cf. Eqs. (1) and (2)). However, as compression of synthesis gas downstream the reformer is costly, it is desirable to operate the reformer at elevated pressure [2]. 3. Reaction mechanism of dry reforming of methane With regard to reduced metal catalysts, the adsorption of methane is generally agreed to take place on metallic sites, adsorbing either directly as CH4 * or dissociative as CH3 * [15–17]. Additional CHx−1 * species are formed as consecutive reactions of the CHx * species [15]. Density functional theory (DFT) calculations on nickel surfaces have shown that the methane activation step has a reaction barrier height on facet sites of ca. 100 kJ/mol and ca. 80 kJ/mol on a step site [16]. Additionally, all methane derivatives are more stable on the step sites than on the facet sites [16]. This shows that the reaction is structure sensitive with reactions favored at low coordinated sites. Adsorption of CO2 on supported reduced metal catalysts is believed to take place either by direct adsorption on a metal particle or on an oxygen vacancy site on an oxide support [6,18]. Direct adsorption on the metal particle is structure sensitive (similar to CH4 ) and preferably occurs at low coordinated sites. After adsorption, dissociation to CO and O, oxidation to CO3 (most likely on noble metals), or disproportionation with gaseous CO2 to CO and CO3 have been described to take place [6,19]. On a catalyst as Ru/Al2 O3 , the Ru sites are able to activate CO2 , but CO2 is more efficiently activated on the Al2 O3 support [20]. This is linked to that CO2 readily adsorbs on basic sites in an oxidic support [21] and generally has a larger binding energy on metal oxides compared to metal surfaces; an effect which is even more pronounced when comparing oxygen vacancy sites and metal sites [22]. Overall, the interface between the metal particle and the support is believed to be the most active sites for CO2 adsorption, and the dissociation of the CO2 molecule may proceed by a hydrogenspillover-like mechanism [22–24]. A key step in the reaction mechanism of DMR is that the oxygen species (O and OH), produced from the CO2 dissociation are required for the oxidation of the CHx species on the metal sites to produce CHx O species, which can decompose to produce H2 and CO [6,9]. Table 1 summarizes the elementary steps associated with the DMR reaction, also indicating the prospective reaction path for CO2 on the support. Comparing DMR to SMR, only the CO2 adsorption steps deviate, as both the CH4 adsorption and dissociation steps and the CO and H2 formation have been reported in SMR [15,16,25–28]. In both cases, OH and O species are believed to be oxidants for the CHx species. An important difference is that H2 O is produced in DMR, but consumed in SMR. Kinetic investigations of the DMR reaction have shown that the RWGS reaction is practically equilibrated under almost all relevant operating conditions [6,9,28,29]. Thus, water will be produced

143

Table 1 Elementary reactions of the DMR reaction on supported metal catalysts [6,8,9]. CH4 adsorption and dissociation CH4 + 2*  CH3 * + H* CH4 +*  CH4 * CH3 * +*  CH2 * + H* CH2 * +*  CH* + H* CH* +*  C* + H* CO2 adsorption and dissociation CO2 +#  CO2 # CO2 # + H*  CO# + HO* CO#  CO + # CO2 # +*  CO# + O* H2 O formation O* + H*  *+ HO* HO* + H*  H2 O + 2* CO and H2 formation CHx * + HO*  CHx O* + H* CHx * + O*  CHx O* +* CHx O*  CO* + xH* C* + O*  CO* CO*  CO +* 2H* +*  2* + H2 * Indicates metal sites and # indicates support sites.

readily at the inlet to the reactor during DMR. Based on this, it has been argued that the DMR reaction to some extent can be considered as a combination of the RWGS and SMR reactions [13,30,31]. Rostrup-Nielsen and Bak Hansen [32] compared SMR and DMR at 500 ◦ C and 1 bar for a series of transition metal catalysts and concluded that replacing H2 O with CO2 did not have a significant impact on the reforming mechanism. The major difference was found in the turnover frequencies (TOF), as DMR was found to be a significantly slower reaction compared to SMR. Avetisov et al. [27] later compared kinetic models for both SMR and DMR with nickel catalysts and concluded that mechanisms and kinetic expressions are similar for the two reactions. The lower rate of reaction observed for DMR compared to SMR was concluded to be due to a larger surface coverage of CO which to some extent occupy and block the active sites. Additionally, Rostrup-Nielsen and Bak Hansen [32] showed that the metals adsorbing CO the strongest have the largest drop in activity for DMR compared to SMR, as shown in Fig. 2 [32]. Wei and Iglesia [28] also compared SMR and DMR and found similar rate of reactions for both reactions, concluding that both reactions could be described by the same rate term.

Fig. 2. Rate of dry reforming relative to the rate of steam reforming at 550 ◦ C on the same catalysts depicted relative to the CO adsorption energy. Data from Rostrup-Nielsen and Bak Hansen [32].

144

P.M. Mortensen, I. Dybkjær / Applied Catalysis A: General 495 (2015) 141–151

Fig. 4. Equilibrium constant for the methane decomposition reaction on various metal catalysts. Ni-a is a nickel catalyst with a maximum nickel particle size of 300 nm, Ni-b is a nickel catalyst with a maximum nickel particle size of 10 nm, and Ni-I is a sulfur-passivated nickel catalyst. Data from Rostrup-Nielsen and Bak Hansen [32,36]. Fig. 3. HRTEM image of whisker carbon (C) formation from a nickel (Ni) particle. Image adopted from Helveg et al. [34].

The similarities in SMR and DMR kinetics can probably be traced back to that the rate determining step in both processes is considered to be the methane adsorption step for many catalysts [6,8,9,33]. 4. Carbon deposition The low H/C ratio in DMR makes carbon formation a major challenge [6]. Carbon formation during primary reforming of methane will principally be due to whisker carbon formation. This takes place as a growth of a carbon nanotube from a nickel particle as illustrated in Fig. 3. The phenomenon has been extensively studied by Helveg et al. [34,35]. In the case of whisker formation, the catalyst pellets will break and fractionate due to the high mechanical strength of the whiskers. This will result in an increased pressure drop over the reactor and eventually a shutdown of the process. Thus, for practical application of the reforming reactions it is a mandatory to ensure that carbon formation does not take place [36]. The methane decomposition reaction (reaction 7) and the Boudouard reaction (reaction 5) are the primary sources of carbon on the catalysts during DMR and these reactions can to some extent be predicted on the basis of thermodynamic calculations. Rostrup-Nielsen [36] used the principle of equilibrated gas to predict the potential for carbon formation. Here, the carbon formation potential is evaluated after establishing the reforming and shift equilibrium of the feed gas and states that carbon formation is expected if the equilibrated gas shows affinity for carbon formation. This can be expressed by the carbon activity (ac,eq ), which for the methane decomposition reaction will be given by: ac,eq = Keq ·

   

pCH4  p2H

2

(8) eq

Here, Keq is the equilibrium constant for carbon formation, and pi is the partial pressure of component i. pCH4 and pH2 should be inserted as calculated from the equilibrium gas phase composition. By definition, this term states that carbon deposition will be expected when ac,eq ≥ 1, as an ac,eq of 1 corresponds to the thermodynamic potential of carbon. Bengaard et al. [16] described that carbon nucleation takes place on the step sites on the nickel crystals and builds carbon layers from

these, forming whisker carbon. However, these layers are only thermodynamically stable when the carbon layer is larger than ≈25 A˚ in diameter (ca. 80 atoms), and therefore, the associated nickel facet, on which the carbon layer is build, should be larger than this. Increasing the carbon layer size decreases the energy of the system and stabilizes the carbon formation further. Thus, large nickel particles are generally found more prone to carbon formation [15,16,37]. The critical size for when carbon deposition is accelerated is in many studies believed to be about 10 nm [6,8,38]. Rostrup-Nielsen and Bak Hansen [32] compared the tendency for carbon formation during DMR on a series of transition metals on MgO-stabilized Al2 O3 at 500–650 ◦ C and 1 bar and found the following order for carbon formation: Ni > Pd  Ir > Pt > Ru ≈ Rh From a series of thermo-gravimetric analysis (TGA) experiments with these catalysts, the equilibrium constants for the methane decomposition equilibrium were derived and are shown in Fig. 4 as a function of temperature. All catalysts had equilibrium constants lower than graphite, which reflects that the carbon formed on the catalysts has a different free energy of formation, and that the most stable form of carbon is graphite. These observations can be used in combination with, e.g., Eq. (8) and the principle of equilibrated gas to predict carbon formation limits. Nickel had the highest potential for carbon formation, with noble metals being more resistant. The Pd catalyst was found most sensitive to temperature and generally deviated somewhat in trend compared to the other catalysts. On a more general note, noble metals are more resistant toward carbon formation than nickel, as shown from Fig. 4, and have low tendency for formation of whisker carbon [2]. Carbon deposition was evaluated on three types of nickel catalysts in the work of Rostrup-Nielsen and Bak Hansen [32] (cf. Fig. 4): nickel catalysts with nickel particles up to 300 nm (Ni-a), nickel catalysts with nickel particles up to 10 nm (Ni-b), and nickel catalyst passivated with sulfur (Ni-I). Following the trends described by Bengaard et al. [16], the potential for carbon formation was significantly decreased by lowering the nickel particle size (comparing Ni-a to Ni-b). In general, it has been described that the largest nickel particle size on a catalyst defines the potential for carbon formation [36]. Even lower potential for carbon formation was achieved by Rostrup-Nielsen and Bak Hansen [32] by passivating the nickel catalysts with sulfur (cf. Ni-I) giving an equilibrium constant for methane decomposition comparable to Ru and Rh catalysts (cf. Fig. 4).

P.M. Mortensen, I. Dybkjær / Applied Catalysis A: General 495 (2015) 141–151

145

Overall, it is apparent that the potential for carbon formation is dictated by thermodynamics, but it can be markedly influenced by the choice of active material on the catalyst and the particle size of this, as this determines the stability of the formed carbon. Besides carbon formation on catalysts, also carbon formation on/in metal surfaces (as the reactor wall and equipment downstream the primary reformer) can be a problem during reforming, a phenomenon known as metal dusting. This causes disintegration of metal alloys of Fe, Ni, and Co, observed as loss of surface material as a metal dust due to carbon formation in the metal [39]. This is usually a problem on reactor walls of plants operating with aggressive carbonaceous gases with high CO partial pressure and low water partial pressure. CO and CH4 are usually reported as the source of metal dusting by reactions 5 and 7 [39–41]. 5. Deactivation DMR and SMR are subject to the same deactivation mechanisms. Sulfur, alkali metals, silica, arsenic, and phosphor have all been reported to be poisonous to the catalyst [15,33,42], with sulfur being the most common poison [42]. Additionally, sintering can be a problem due to the high temperatures required for DMR [6].

Fig. 5. Two-dimensional volcano-curve of the TOF for SMR as a function of the oxygen and carbon adsorption energies developed on the basis of DFT calculations. T = 500 ◦ C, P = 1 bar. Figure from Jones et al. [26].

5.1. Sulfur poisoning The content of sulfur in hydrocarbon feeds is normally in the order of 5–20 ppm for natural gas and up to 500 ppm in liquid hydrocarbons [42]. Under reforming conditions, all sulfur will be transformed to H2 S. This constitutes a potential threat for reforming catalysts, as all group VIII–XI metals can react with H2 S, forming surface adsorbed sulfur, and nickel is among the materials with the highest affinity for sulfur adsorption when comparing to other traditional metals used in reforming catalysts [36]. Sulfur adsorption is additionally preferred at the low coordinated nickel sites relative to facet sites [16]. Rostrup-Nielsen [43] showed by comparing the intrinsic reforming reaction rate of a sulfur-passivated nickel catalyst to a sulfur-free nickel catalyst that the reaction rate scales with a factor of (1 −  S )3 , where  S is the fractional coverage of the Ni-surface by sulfur atoms. This means that the sulfur efficiently deactivates the catalysts. As sulfur is known to preferentially adsorb on low coordinated sites relative to facet sites [16], the deactivation observed by sulfur of Rostrup-Nielsen [43] can probably be linked to the preferential adsorption on the low coordinated (most active) sites. The adsorption of sulfur is in principle reversible, but the driving force for the reaction is low, especially at low sulfur coverage [14]. Thus, it is in general preferential to remove sulfur prior to SMR, CO2 -reforming, and DMR. This is normally done by initial hydrodesulfurization of the hydrocarbon feed followed by absorption in ZnO. In this way, the sulfur content can be decreased to low ppb levels [44]. 5.2. Sintering Sintering is the coalescence of small particles into larger particles at elevated temperatures, which reduces the surface area and thereby decreases the total energy of the system [25]. This is dependent on a number of parameters: atmosphere over the catalyst, catalyst composition, catalyst structure, support morphology, and temperature [15,45,46]. Sehested [47,48] described that the rate of sintering scales with the partial pressure of water. Additionally, the rate of sintering will be accelerated with increasing temperature [49]. The growth of particle size results in a decrease in the total amount of active sites, but it also decreases the fraction of low

coordinated sites. Thus, it is desirable to limit and ideally avoid sintering as both activity and stability of the catalyst are affected. 6. Catalysts for dry reforming Rostrup-Nielsen and Bak Hansen [32] compared the activity for DMR of a series of transition metals on MgO-stabilized Al2 O3 at 500–650 ◦ C and 1 bar and found the following order of activity: Ru > Rh ≈ Ni > Ir > Pt > Pd. From this study, Ru appeared as the best choice of a DMR catalyst. Ferreira-Aparicio et al. [20] also compared the activity of a series of transition metal catalysts on both ␥-Al2 O3 and SiO2 at 450 ◦ C and 1 bar and found the following order of activity on ␥-Al2 O3 : Rh > Ni > Ir > Ru > Pt > Co and on SiO2 : Ni > Ru > Rh > Ir > Co ≈ Pt. As the activity for reforming is sensitive to dispersion of the active metal and support–metal interaction, comparing the activity of the active metal on the catalyst can be difficult and therefore difference in the data from Rostrup-Nielsen and Bak Hansen [32] and Ferreira-Aparicio et al. [20] is seen, but some general trends are that Ru, Rh, Ir, and Ni rank among the best performing metals for DMR. Jones et al. [26] used DFT calculations to show that the activity of SMR catalysts based on transition metals scaled with the adsorption energies of O and C on the respective metals due to the Brønsted–Evans–Polanyi (BEP) relationship, as shown in Fig. 5. On this basis, the apparent order of activity for SMR catalysts was found as: Ru ≥ Rh > Ni ≥ Ir > Co > Pt ≈ Pd > Fe > Cu > Ag > Au. This result is very similar to the observations of Rostrup-Nielsen and Bak Hansen [32] and also resembles the result of Ferreira-Aparicio et al. [20]. Assumably, the activity of DMR catalysts will therefore also scale with the O and C adsorption energies, due to the similarities in reaction mechanisms and kinetics of SMR and DMR (as discussed in Section 3), and the volcano plot developed by Jones et al. [26] is probably applicable for DMR catalysts as well. Screening work of both SMR and DMR catalysts indicates Ru and Rh as some of the best metals, followed by Ni and Ir. However, comparing the metal prices of these shows that Ru, Rh, and Ir are a factor of 200–2000 more expensive than Ni [50]. Thus, despite the fact that Ni may not be the most active metal for DMR, it represents a Pareto optimal catalyst (for further information on Pareto

146

P.M. Mortensen, I. Dybkjær / Applied Catalysis A: General 495 (2015) 141–151

Fig. 6. General flow sheet of a reforming process. Dotted lines indicate optional configurations. CO2 can be added to adjust the feed gas composition or recycle from the CO2 removal unit. H2 S import is only used for the SPARG process, which is discussed in Section 7.2.

optimality see references [51,52]) when considering activity compared to metal price.

adsorption and will therefore promote gasification of carbon and coke precursors on the catalyst [36,56].

6.1. Catalyst support

7. Industrial experience with large scale reforming of CO2 -rich gas

Strong metal–support interaction and metal oxides capable of supplying vacancy sites for adsorption of the CO2 molecule in the bifunctional reaction mechanism described in Section 3 are desirable for the DMR reaction, as the choice of support influence both the activity and tendency for carbon formation [6]. Supports such as Al2 O3 , MgAl2 O4 , CaAl2 O4 , TiO2 , MgO, CeO2 , ZrO2 , and La2 O3 have been investigated for DMR [8]. Generally, supports with strong basicity facilitate CO2 -adsorption and have been indicated to suppress carbon deposition [6,9]. CO2 adsorption capacity of Al2 O3 is generally considered moderate [6]. Instead, the basic properties can be improved by adding alkaline earth materials as CaO and MgO. Adding small amounts of CaO to a Ni/Al2 O3 catalyst can improve activity and decrease carbon deposition. The lower tendency for carbon formation has been linked to a larger potential for adsorption of CO2 , which promotes the reverse Boudouard reaction [53]. 6.2. Catalyst promotion Blocking low coordinated sites has been found as an aid in preventing carbon formation, which effectively means that this process can operate under more severe conditions than a conventional catalyst with respect to carbon formation. Promotion of Ni catalysts has been tried, with selective deposition of ad-atoms on the low coordinated sites trying to inhibit carbon formation. Gold and boron have both been shown to effectively suppress the formation of whisker carbon in comparison to an unpromoted case by both stabilizing small nickel particle sizes and blocking the low coordinated sites [6,54,55]. However, blocking the low coordinated sites with Au, B, or similar substances will also decrease the activity of the catalyst because these sites also serve as the most active sites for the reforming reactions with CH4 . Moreover, potassium is known for its ability to suppress the carbon formation. This promoter is believed to enhance steam

A typical layout for an industrial scale reforming plant is illustrated in Fig. 6. The feedstock, which may range from lean natural gas to heavy naphtha, is heated to around 400 ◦ C and then cleaned for sulfur species in a desulfurization section [33]. Subsequently any higher hydrocarbons in the feedstock are reformed to a mixture of H2 , CO2 , CH4 , and traces of CO in a prereformer. This is done at relative low temperatures of 400–550 ◦ C to avoid carbon formation from the higher hydrocarbons [42]. Details on prereforming can be found in references [33,42,44,57]. The actual reforming takes place in a tubular reformer (primary reformer) for the conversion of the CH4 /H2 O/CO2 mixture to synthesis gas. In this configuration, many tubes are placed in a row (or parallel rows) in a furnace box where heat is delivered to the endothermic reaction by combustion of fuel. The tubes typically range from 100 to 200 mm in outer diameter and 10 to 13 m in length [30]. The configuration of many tubes with relative small diameter is done to optimize heat transfer as this becomes the limiting factor of the primary reformer, especially toward the exit of the tubes [15]. Typically, the inlet temperature is adjusted to 400–600 ◦ C prior to the primary reformer and then heated to an exit temperature up to 950 ◦ C in the reformer tubes. The high exit temperature is required for sufficient conversion of the methane (cf. Fig. 1). When the process gas flows through the reformer tubes in plug flow it will be close to equilibrium. Thus, carbon formation can be expected if the principle of equilibrated gas predicts carbon formation at any temperature between 400 and 1000 ◦ C for a given feed gas composition. As already discussed in Section 4, it is essential to design a reforming plant to completely avoid carbon formation. If a potential for carbon formation exist, it will only be a matter of time before a shutdown will be forced due to too high pressure drop. In industrial context this will be expensive due to lost time on stream and loading of a new batch of catalyst. It is emphasized that carbon formation at reforming conditions is as whisker carbon. This is destructive in nature toward the catalyst pellet (cf. Fig. 3) and

P.M. Mortensen, I. Dybkjær / Applied Catalysis A: General 495 (2015) 141–151

regeneration is therefore not an option. Thus, the possible operating range for a tubular reformer will be defined by the conditions which will not have a potential for carbon formation. When sufficient knowledge about the thermodynamics of carbon formation for a specific catalyst is known the exact limit for carbon formation can be calculated and this can be illustrated by the carbon limit curve depicted in Fig. 7 [58,59]. Both the carbon limit for graphite

147

and an industrial nickel catalyst are shown. As the limits have to be defined in a worst case scenario, industrial nickel catalyst represent a nickel catalyst aged for several years in a reforming plant where the catalyst has been severely sintered. The curves are derived from the principle of equilibrated gas and show the most severe conditions (as a function of initial H2 O/CH4 and CO2 /CH4 ratio or O/C and H/C ratio) which can be tolerated in the entire temperature

Fig. 7. Carbon limit diagram at different H2 O/CH4 and CO2 /CH4 feed ratios. The curves are the carbon limits at 25.5 bar for graphite and a typical industrial nickel catalyst. The left-hand sides of the curves are the areas where carbon formation is expected for temperatures between 400 and 1000 ◦ C. The product H2 /CO ratio is evaluated at TExit = 950 ◦ C. The stoichiometric reforming region is shown for reference. The points refer to specific large scale tests with nickel catalysts (green), SPARG (orange), and noble metal catalysts (blue), further information on these is found in Table 2. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

148

P.M. Mortensen, I. Dybkjær / Applied Catalysis A: General 495 (2015) 141–151

Table 2 Comparison of full size monotube pilot experiments (HOU) and industrial plants (referenced with country were they are constructed) of CO2 rich reforming with different catalysts and under different conditions. Operating conditions

Primary reformer feed

Outlet

TOS [h]

Catalyst

Feed

P [barg]

TExit [◦ C]

H2 O/CH4 [−]

CH4 [dry mole%]

CO2 [dry mole%]

H2 [dry mole%]

H2 /CO [−]

Ni cat. HOU-1 HOU-2 Iran Indonesia Malaysia South Korea India United Kingdom Netherlands

Ni/MgAl2 O4 Ni/MgAl2 O4 Ni/MgAl2 O4 Ni/MgAl2 O4 Ni/MgAl2 O4 Ni/MgAl2 O4 Ni/MgAl2 O4 Ni/MgAl2 O4 Ni/MgAl2 O4

NG NG NG NG NG Naphtha Naphtha Naphtha NG

23 24 20 12 25 25 22 20 15

945 945 960 960 950 900 920 900 815

1.4 1.6 1.3 2.1 1.5 1.7 1.7 2.0 2.0

74 56 70 74 63 47 40 32 28

18 44 12 23 20 38 41 56 71

9 0 17 2 16 15 17 12 1

2.6 1.8 3.0 2.7 2.7 2.1 1.9 1.7 1.2

400 380 – – – – – – –

SPARG HOU-3 HOU-4 HOU-5 HOU-6 USA

Ni/MgAl2 O4 Ni/MgAl2 O4 Ni/MgAl2 O4 Ni/MgAl2 O4 Ni/MgAl2 O4

NG NG NG NG NG

6 6 14 15 7

890 930 930 945 900

1.0 0.4 0.7 0.7 0.9

57 40 38 28 61

37 59 61 67 33

6 1 1 5 6

1.9 0.9 0.9 0.7 1.8

500 500 160 400 –

Noble cat. HOU-7 Japan

Ru/MgAl2 O4 Ru/MgAl2 O4

NG NG

23 2

940 850

0.9 1.8

51 10

44 56

5 32

1.4 1.0

490 –

SPARG: sulfur-passivated reforming, NG: natural gas, TOS: time on stream.

range from 400 to 1000 ◦ C at a pressure of 25.5 bar. Carbon formation will be expected on the left of the curves and “safe” operation on the right. This shows that the tendency for carbon formation increases with decreasing CO2 /CH4 and H2 O/CH4 ratios. The severity of operation can be defined relative to the placement compared to the carbon limit curves; operation far beyond the carbon limit curve is considered very severe. As shown in Fig. 4, the limit for carbon formation can be pushed by choice of catalyst. Thus, more severe conditions can be tolerated with a nickel catalyst compared to formation of graphitic carbon (cf. Fig. 7). However, performing “traditional” DMR, without H2 O in the feed, over a nickel catalyst in a tubular reformer will result in carbon formation as this is placed far beyond the carbon limit curve for nickel catalysts in Fig. 7. Instead, CO2 -reforming is possible as long as the H2 O/CH4 ratio is balanced accordingly. The dotted lines in Fig. 7 show the equilibrated H2 /CO ratio of a synthesis gas produced at 950 ◦ C and 25.5 bar as a function of the O/C and H/C ratio (or inlet H2 O/CH4 and CO2 /CH4 ). Increasing the CO2 /CH4 feed ratio will decrease the H2 /CO ratio accordingly. However, when producing a synthesis gas with a very low H2 /CO ratio an accompanied high H2 O/CH4 ratio will be needed to balance the severity of the gas to avoid carbon formation on a nickel catalyst. Specifically in these cases it is desirable to operate in the severe region to the left of the carbon limit curves. This would decrease the requirement for addition of large amounts of steam, which is expensive both with respect to operating costs (increased expenses to steam preheating) and construction costs (increased reactor sizes due to the large feed streams). In Table 2 representative references of what can be done with CO2 -reforming in industrial scale are summarized, comparing operating conditions with different combinations of catalysts. The severity of these cases is additionally illustrated with the points in Fig. 7. The cases in Table 2 cover operation at full scale plants (referred to on the basis of the nations where the plant was constructed) and tests in a pilot plant placed in Houston, TX, United States (referred to as HOU). The pilot plant is a full size monotube reactor with a tube of similar dimensions to an industrial scale reforming plant and a furnace with radiant wall burners in six levels. This single tube pilot was operated with a natural gas feed flow in the order of 60–120 Nm3 /h. This configuration makes it possible to

simulate the exact conditions of any typical side fired reforming plant. The industrial references in Table 2 are all examples of plants which have been designed around the principle flowsheet in Fig. 6, but adapted to the specific needs of the site and the downstream requirements for the synthesis gas utilization. The sizes of these plants range from production of 2,400 Nm3 /h up to 133,000 Nm3 /h of dry synthesis gas. A differentiation is made between plants operated with a natural gas feed (NG in Table 2) or a naphtha feed. Natural gas is considered as a gaseous feed consisting mainly of methane, but can also contain CO2 , higher hydrocarbon (C2 –C6 ), among other gas species. Naphtha is a heavier liquid feedstock which will consist mainly of C6 –C12 hydrocarbons. For naphtha based plants the prereforming section is an important an integrated part of the plant as the very reactive higher hydrocarbons should be converted prior to the primary reformer to avoid carbon formation from these. Several of the plants additionally operated with a CO2 import or recycle stream to give a high CO2 /CH4 ratio in the feed gas to the primary reformer. 7.1. Nickel catalysts for CO2 -reforming As already discussed, nickel based catalysts are industrially preferred. Table 2 lists several examples of how nickel based catalysts have been used for CO2 -reforming. A few references were made at the full size monotube reforming reactor in Houston in the beginning of the 90s (tests denoted HOU1-2), but several full scale plants have been designed and constructed relative to the principle of equilibrated gas and the carbon limit curve shown in Fig. 7. Common for all of these references is that they were operated over long periods (several years) without problems with carbon formation. As long as the plant is operated as intended and not poisoned by external sources, lifetimes of more than a decade can be obtained for industrial reforming catalysts [60]. During this period only traces of carbon formation will be deposited when operation is based on thorough thermodynamic considerations. A key parameter to monitor during operation is the approach to equilibrium of the reforming reaction at the outlet of the primary reformer. This is defined as the difference between the temperature at which the actual gas composition would have been at equilibrium compared to the actual temperature measured out of the reactor. This should

P.M. Mortensen, I. Dybkjær / Applied Catalysis A: General 495 (2015) 141–151

ideally be within say 20 ◦ C, larger approaches would usually mean that the catalyst has been deactivated. When prereforming heavy feedstock, such as naphtha, the product gas will have a relatively high fraction of CO2 due to the combination of higher hydrocarbon reforming, water-gas shift, and methanation. This means that naphtha based reforming plants will be cases of CO2 -reforming. The plants in India, South Korea, and the United Kingdom are examples of naphtha based CO2 -reforming plants with CO2 recycle, in these cases the feed to the tubular reformer contains 41 dry mole%, 38 dry mole%, and 56 dry mole% CO2 , respectively. Using the CO plant in India as example, at the inlet to the prereformer the naphtha is mixed with steam to a S/C (steam to hydrocarbon–carbon ratio) of 1.8 at 480 ◦ C to convert all the higher hydrocarbons. This produces a gas with 55 dry mole% CH4 , 20 dry mole% CO2 , 1 dry mole% CO, and 23 dry mole% H2 . Recycled CO2 is added to this, producing a feed to the primary reformer containing 40 dry mole% CH4 , 41 dry mole% CO2 , and 17 dry mole% H2 at a H2 O/CH4 ratio of 1.7. Reforming this gas at an exit temperature of 920 ◦ C produces a synthesis gas containing 2 dry mole% CH4 , 12 dry mole% CO2 , 30 dry mole% CO, and 56 dry mole% H2 . Despite having a high conversion of methane, there is a significant slip of CO2 , which is due to the thermodynamics as shown by Fig. 1. Thus, any CO2 -reforming plant usually features a gas purification step such as pressure swing adsorption, chemical absorption, or similar (see Rufford et al. [61] for further details on CO2 separation) to separate excess CO2 from the synthesis gas. Large fractions of CO2 in the feed were demonstrated in the Houston pilot plant demonstration run 2 and the plants in India, South Korea, and the United Kingdom. However, as seen from both Fig. 7 and Table 2, significant amounts of water were needed in the feed to comply with the limitations of the carbon limit curve and additionally to adjust the composition so the synthesis gas is produced with the desired H2 /CO ratio. The plant in the Netherlands operated under very severe conditions for a nickel based catalyst with 71 dry mole% CO2 in the feed, producing a synthesis gas with a H2 /CO ratio of 1.2. As seen from Fig. 7, this plant was operated slightly beyond the conventional border for carbon formation for nickel catalysts. After 2 years of operation at these conditions analysis of the spent catalyst revealed insignificant carbon formation, varying between 500 and 1100 ppmw along the length of the reactor tube. Analysis of the spent catalyst showed that operation under the quite severe conditions only was possible because the nickel particles were stabilized at a significantly lower particle size than what usually is seen during steam reforming at industrial conditions, which was helped by the relative low exit temperature of 815 ◦ C used for the current plant. However, it should still be noted that a H2 O/CH4 ratio of 2 was used at the given operating conditions. Operation at the same conditions with a nickel catalyst with larger nickel particles later proved to cause carbon formation, supporting the fundamentals on carbon formation described in Section 4. All the examples presented for nickel based CO2 -reforming show how important it is to know where the limit for carbon formation is, as this confines the freedom for designing reforming processes. 7.2. SPARG process In the SPARG (Sulfur PAssivated ReforminG) process, sulfur is used to selectively poison the most active sites and in this way prevent formation of carbon while maintaining some activity for reforming. Rostrup-Nielsen [43] described that coke formation requires a larger ensemble of nickel atoms compared to SMR and that one sulfur atom quenches the four neighboring nickel atoms. Thus, sulfur will more effectively inhibit carbon formation than

149

SMR/DMR activity. This was used to explain why sulfur has a beneficial effect in preventing carbon formation. Ideally, the sulfur coverage should be larger than 0.7 to prevent carbon formation [43]. On the basis of the observations of the beneficial effect of sulfur passivation, the SPARG process was developed. Here, a small co-feed of H2 S was added to prevent carbon formation on nickelbased catalysts. This should be accompanied by a small co-feed of hydrogen (H2 S/H2 < 0.9) to avoid formation of bulk nickel sulfides. Furthermore, having sulfur in the gas helps prevent carbon formation in the downstream section of the reforming plant [36]. The SPARG approach offers a route to circumvent the carbon limit curve in Fig. 7. The first large scale test of this concept was made in the full size monotube pilot plant in Houston at a CO2 /NG feed ratio of 0.65 and H2 O/NG ratio of 1.0, and an outlet temperature of 890 ◦ C. These conditions would result in carbon formation on a conventional nickel catalyst as illustrated by point 10 (HOU-3) in Fig. 7. However, with the sulfur-passivated nickel catalyst no carbon was observed after 500 h of operation at the conditions. Building on the success of HOU-3, “dry” DMR tests were made in the pilot plant with the sulfur-passivated Ni catalyst, a CO2 /NG feed ratio of 2.5, and 890 ◦ C as outlet temperature. However, at these conditions, carbon deposition was observed in the first part of the reactor tube after a short period of operation. This was concluded to be due to cracking of the larger hydrocarbons in the natural gas on the sulfur passivated catalyst, which now had insufficient activity for higher hydrocarbon reforming. To avoid the severe carbon deposition observed from higher hydrocarbons, it was realized that a prereformer was essential prior to the tubular reformer to remove the higher hydrocarbons in the natural gas. On the basis of this experience, a general flow sheet for CO2 -reforming using the SPARG technology should therefore be as illustrated in Fig. 6. Initially sulfur is removed from the natural gas to enable prereforming of the higher hydrocarbons over a nickel based catalyst, which is intolerant to sulfur. A controlled amount of H2 S is then added to the gas mixture of H2 O/CO2 /CH4 prior to the primary reformer. The SPARG configuration in Fig. 6 was used in three pilot-scale tests (HOU-4-6) of the SPARG process with the prereformer operated at a minimum S/C ratio of 0.35 with a sulfur-free catalyst; CO2 and sulfur were added after the prereformer. HOU-4 was a lowpressure test at 6 barg, and HOU-5 and HOU-6 were at 14–15 barg. These three runs demonstrated that the SPARG process could be used for operation under severe conditions for carbon formation, as the points in Fig. 7 are far beyond the conventional carbon limit of the traditional nickel catalyst. The three tests commonly demonstrated that the SPARG technology was applicable in a wide pressure range interesting for reforming plants and that the technology can be used to produce synthesis gas with a H2 /CO ratio below 1. During unloading after the pilot experiments, samples were taken with 0.5 m intervals along the length of the 12 m reactor tube. Analyzing the spent catalyst for carbon, less than 500 ppmw (detection limit for combustion infrared detection technique at time of measuring) on all catalyst samples was quantified, documenting carbon free operation. Additional analysis of the catalyst samples also showed that the strength of the catalyst pellets was very similar to the strength of the fresh catalyst pellets. Analysis of nickel crystal sizes showed that sintering had taken place, but that the sintering followed conventional particle growth rates compared to more traditional reforming conditions. The SPARG technology was demonstrated in industrial scale by revamping a reforming plant in the United States (see also Fig. 7 and Table 2) from SMR to CO2 -reforming. Originally the plant operated at a S/C of 1.9 and a CO2 /C of 0.4 with an exit temperature of 900 ◦ C, but to increase the production of CO the plant was modified

150

P.M. Mortensen, I. Dybkjær / Applied Catalysis A: General 495 (2015) 141–151

Table 3 Comparison of the synthesis gas composition from the plant in USA pre-SPARG and during SPARG.

H2 [dry mole%] CO [dry mole%] CO2 [dry mole%] CH4 [dry mole%] N2 [dry mole%] H2 /CO

Pre-SPARG

SPARG

65 24 8 3 0.1 2.7

60 33 5 3 0.1 1.8

with the SPARG technology to operate at a S/C of 0.9 and a CO2 /C of 0.5 instead. Table 3 shows a comparison between the synthesis gas composition out of the primary reformer of the plant prior to the revamp and at SPARG conditions. A significant increase in the CO concentration was achieved by implementing the SPARG conditions. The only major difference between original operation and SPARG operation was that a pre-reformer had to be installed to remove the higher hydrocarbons prior to the primary reformer, as also observed in the pilot experiments. The primary reformer and the surrounding equipment could all be used directly for the SPARG process. Carbon free-operation was demonstrated throughout 4 years of continuous operation on one batch of catalyst on the SPARG conditions. The revamp of the plant from SMR to SPARG was calculated to give 23% savings in production costs at that time [62]. 7.3. Noble metal catalysts for CO2 -reforming As already discussed in Section 4 and shown in Fig. 4, noble metals generally have a lower tendency for carbon formation compared to nickel catalysts. This group of catalysts therefore offers a route for operation at severe conditions without carbon formation. This was demonstrated in large scale with the HOU-7 pilot test where a ruthenium catalyst was used to operate at a H2 O/CH4 of 0.9 and a CO2 /CH4 of 0.8. These conditions are far beyond the carbon limit curve of the nickel catalyst, as shown in Fig. 7. With an exit temperature of 940 ◦ C, the product synthesis gas was 52 dry mole% H2 , 36 dry mole% CO, 8 dry mole% CO2 , and 4 dry mole% CH4 . These conditions were tested for 490 h of continuous operation. After operation, 0.2–0.9 wt% of carbon was found on the spent ruthenium catalyst. This could all be traced back to carbonate, which was a leftover from the production of the spinel support. An additional 1500 h bench-scale test confirmed, with the same catalyst under similar conditions, that carbon did not form on the catalyst. However, these tests did also show that the quite aggressive synthesis gas product led to some metal dusting in the cold part of the reactor outlet. This underlines the importance of managing metal dusting potential when considering CO2 -reforming reactions, despite finding a catalyst that can deal with the severe conditions. The experience from HOU-7 and the bench-scale tests were used to demonstrate the durability of the catalyst prior to installation in a convection reformer in Japan. The plant should produce CO from mixture of natural gas, a CO2 import stream, and a recycle stream, which ultimately meant that high amounts of CO2 should be processed. Use of the noble metal catalyst enabled operation at quite severe conditions while maintaining a syngas H2 /CO product ratio of 1. This plant operated for several years without any catalytic problems. 8. Conclusions The DMR reaction is closely related to the SMR reaction. Despite having origin in different reactants, the fast equilibration of the WGS results in similar reaction conditions in the two processes

at an early stage in a reactor. For this reason, kinetic models, rate controlling steps, reaction mechanism, etc., can to a large extent be handled similarly across the two reactions. The similarities in the SMR and DMR reactions also entail that tendencies for catalyst activity are similar, if not identical. Nickel is the most investigated catalyst due to the attractive price compared to the more active noble metal catalyst. Much work has focused on improving nickel catalysts by proper choice of support and promoters. Carbon deposition must be considered as the biggest challenge of DMR, as the stoichiometric DMR reaction implies severe conditions for carbon formation. This phenomenon is to a large extent controlled by thermodynamics and the principle of equilibrated gas is an important method to determine suitable operating conditions. Water cannot be completely omitted from the reaction, as this specifically will be needed for removal of higher hydrocarbons. Thus, “dry” DMR is difficult to realize in large scale with realistic feedstocks, but high severity CO2 -reforming can be done. Noble metals are generally less prone to carbon deposition than nickel because the equilibrium constant for carbon formation is found closer to graphitic carbon on nickel. Pushing the potential for carbon formation can be aided by approaches such as the SPARG process or by proper control of nickel particle size. When detailed information of the thermodynamics of a given system is available, the principle of equilibrated gas can be used to make quite precise predictions on which operating ranges can be covered without carbon formation on the catalyst. On the basis of the principle of equilibrated gas, several CO2 reforming plants with nickel based catalysts have been designed. These have all been confined to also having significant amounts of water in the feed to decrease the severity for carbon formation as this is dictated by the limit for carbon formation. For more severe conditions, the SPARG process or noble metal catalysts have been demonstrated in full scale reforming plants, where only small amounts of water were needed as co-feed. These cases show how CO2 -rich gas can be reformed to produce a synthesis gas with a low H2 /CO ratio.

References [1] J.R. Rostrup-Nielsen, Stud. Surf. Sci. Catal. 147 (2004) 121–126. [2] S.E.L.W. Madsen, J.-H. Bak Hansen, J.R. Rostrup-Nielsen, Industrial aspects of CO2 -reforming, in: Presented at the AIChE Spring Meeting, Houston, TX, USA, 1997. [3] Methanol Institute, The Methanol Industry, 2014 http://www.methanol.org/ Methanol-Basics/The-Methanol-Industry.aspx [4] J.G.J. Olivier, G. Janssens-Maenhout, M. Muntean, J.A.H.W. Peters, Trends in Global CO2 Emissions: 2013 Report, PBL Netherlands Environmental Assessment Agency, 2013. [5] S. Holloway, J.M. Pearce, T. Ohsumi, V.L. Hards, A Review of Natural CO2 Occurrences and their Relevance to CO2 Storage, British Geological Survey, CR/05/104, 2005. [6] C. Papadopoulou, H. Matralis, X. Verykios, in: L. Guczi, A. Erdôhelyi (Eds.), Catalysis for Alternative Energy Generation, Springer, New York, 2012, pp. 57–127. [7] J.R. Rostrup-Nielsen, L.J. Christiansen, Concepts in Syngas Manufacture, Imperial College Press, London, 2011, pp. 3–72. [8] M.-S. Fan, A.Z. Abdullah, S. Bhatia, ChemCatChem 1 (2009) 192–208. [9] M.C.J. Bradford, M.A. Vannice, Catal. Rev. 41 (1999) 1–42. [10] T.S. Christensen, I.I. Primdahl, Hydrocarb. Process. 73 (1994) 39–46. [11] I. Dybkjær, Fuel Process. Technol. 42 (1995) 85–107. [12] K. Aasberg-Petersen, T.S. Christensen, C.S. Nielsen, I. Dybkjær, Fuel Process. Technol. 83 (2003) 253–261. [13] M.K. Nikoo, N.A.S. Amin, Fuel Process. Technol. 92 (2011) 678–691. [14] J.R. Rostrup-Nielsen, Handbook of Heterogeneous Catalysis, Wiley-VCH, Weinheim, 2008, pp. 2882–2905. [15] J.R. Rostrup-Nielsen, J. Sehested, J.K. Nørskov, Adv. Catal. 47 (2002) 65–139. [16] H. Bengaard, J.K. Nørskov, J. Sehested, B.S. Clausen, L.P. Nielsen, A.M. Molenbroek, J.R. Rostrup-Nielsen, J. Catal. 209 (2002) 365–384. [17] J. Xu, G.F. Froment, AIChE J. 35 (1989) 97–103. [18] A. Slagtern, Y. Schuurman, C. Leclercq, X. Verykios, C. Mirodatos, J. Catal. 172 (1997) 118–126. [19] H.-J. Freund, R.P. Messmer, Surf. Sci. 172 (1986) 1–30.

P.M. Mortensen, I. Dybkjær / Applied Catalysis A: General 495 (2015) 141–151 [20] P. Ferreira-Aparicio, I. Rodrı´ıguez-Ramos, J. Anderson, A. Guerrero-Ruiz, Appl. Catal. A: Gen. 202 (2000) 183–196. [21] A. Cimino, F.S. Stone, Adv. Catal. (2002) 141–306. [22] U. Burghaus, Catal. Today 148 (2009) 212–220. [23] A. Topalidis, D.E. Petrakis, A. Ladavos, L. Loukatzikou, P.J. Pomonis, Catal. Today 127 (2007) 238–245. [24] Z.X. Cheng, X.G. Zhao, J.L. Li, Q.M. Zhu, Appl. Catal. A: Gen. 205 (2001) 31–36. [25] I. Chorkendorff, J.W. Niemantsverdriet, Concepts of Modern Catalysis and Kinetics, 2nd ed., Wiley-VCH, Weinheim, 2007. [26] G. Jones, J.G. Jakobsen, S.S. Shim, J. Kleis, M.P. Andersson, J. Rossmeisl, F. Abild-Pedersen, T. Bligaard, S. Helveg, B. Hinnemann, J.R. Rostrup-Nielsen, I. Chorkendorff, J. Sehested, J.K. Nørskov, J. Catal. 259 (2008) 147–160. [27] A.K. Avetisov, J.R. Rostrup-Nielsen, V.L. Kuchaev, J.-H. Bak Hansen, A.G. Zyskin, E.N. Shapatina, J. Mol. Catal. A. Chem. 315 (2010) 155–162. [28] J. Wei, E. Iglesia, J. Catal. 224 (2004) 370–383. [29] M.C.J. Bradford, M.A. Vannice, Appl. Catal. A: Gen. 142 (1996) 97–122. [30] J.R. Rostrup-Nielsen, L.J. Christiansen, Concepts in Syngas Manufacture, Imperial College Press, London, 2011, pp. 143–212. [31] A. Donazzi, A. Beretta, G. Groppi, P. Forzatti, J. Catal. 255 (2008) 259–268. [32] J.R. Rostrup-Nielsen, J.-H. Bak Hansen, J. Catal. (1993) 38–49. [33] K. Aasberg-Petersen, T.S. Christensen, I. Dybkjær, J. Sehested, M. Østberg, R.M. Coertzen, M.J. Keyser, Fischer–Tropsch Technology, Elsevier, Amsterdam, 2004, pp. 258–405. [34] S. Helveg, J. Sehested, J.R. Rostrup-Nielsen, Catal. Today 178 (2011) 42–46. [35] S. Helveg, P.L. Hansen, Catal. Today 111 (2006) 68–73. [36] J.R. Rostrup-Nielsen, L.J. Christiansen, Concepts in Syngas Manufacture, Imperial College Press, London, 2011, pp. 233–294. [37] T. Borowiecki, Appl. Catal. (1982) 223–231. [38] S. Tang, L. Ji, J. Lin, H.C. Zeng, K.L. Tan, K. Li, J. Catal. 194 (2000) 424–430. [39] A. Agüero, M. Gutiérrez, L. Korcakova, T.T.M. Nguyen, B. Hinnemann, S. Saadi, Oxid. Met. (2011) 1–20. [40] H.J. Grabke, Mater. Corros. 54 (2003) 736–746. [41] P.V.D.S. Gunawardana, T.T.M. Nguyen, J.C. Walmsley, H.J. Venvik, Ind. Eng. Chem. Res. 53 (2014) 1794–1803.

151

[42] T.S. Christensen, Appl. Catal. A: Gen. 138 (1996) 285–309. [43] J.R. Rostrup-Nielsen, J. Catal. 85 (1984) 31–43. [44] T.S. Christensen, J.-H. Bak Hansen, Fundaments in design of adiabatic preconverters for steam reforming of hydrocarbons, in: Second Nordic Symposium on Catalysis, Lyngby, Denmark, 1989. [45] J. Sehested, A. Carlsson, T.V.W. Janssens, P.L. Hansen, A.K. Datye, J. Catal. 197 (2001) 200–209. [46] K.O. Christensen, D. Chen, R. Lødeng, A. Holmen, Appl. Catal. 314 (2006) 9–22. [47] J. Sehested, J. Catal. 217 (2003) 417–426. [48] J. Sehested, J.A.P. Gelten, I.N. Remediakis, H. Bengaard, J.K. Nørskov, J. Catal. 223 (2004) 432–443. [49] J. Sehested, Catal. Today 111 (2006) 103–110. [50] U.S. Geological Survey, Mineral Commodity Summaries 2013, U.S. Geological Survey, 2013. [51] T. Bligaard, G.H. Jóhannesson, A.V. Ruban, H.L. Skriver, K.W. Jacobsen, J.K. Nørskov, Appl. Phys. Lett. 83 (2003) 4527–4529. [52] V. Pareto, Manual of Political Economy, Macmillan, New York, 1971. [53] C.E. Quincoces, S. Dicundo, A.M. Alvarez, M.G. González, Mater. Lett. 50 (2001) 21–27. [54] L. Guczi, G. Stefler, O. Geszti, I. Sajó, Z. Pászti, A. Tompos, Z. Schay, Appl. Catal. A: Gen. 375 (2010) 236–246. [55] F. Besenbacher, I. Chorkendorff, B.S. Clausen, B. Hammer, A.M. Molenbroek, J.K. Nørskov, I. Stensgaard, Science 279 (1998) 1913–1915. [56] F. Frusteri, L. Spadaro, F. Arena, A. Chuvilin, Carbon 40 (2002) 1063–1070. [57] T.S. Christensen, J.R. Rostrup-Nielsen, Deactivation and Testing of Hydrocarbon-Processing Catalysts, American Chemical Society, 1995, pp. 186–200. [58] H.C. Dibbern, P. Olesen, J.R. Rostrup-Nielsen, P.B. Tøttrup, N.R. Udengaard, Hydrocarb. Process. 65 (1986) 71–74. [59] J.R. Rostrup-Nielsen, Catal. Today 111 (2006) 4–11. [60] L. Storgaard, H.C. Nielsen, Digit. Refin. (1999) 1–8. [61] T.E. Rufford, S. Smart, G.C.Y. Watson, B.F. Graham, J. Boxall, J.C. Diniz da Costa, E.F. May, J. Pet. Sci. Eng. 94–95 (2012) 123–154. [62] N.R. Udengaard, J.-H. Bak Hansen, D.C. Hanson, J.A. Stal, Oil Gas J. 90 (1992) 62–67.