Ultra-deep Desulfurization of Liquid Hydrocarbon Fuels: Chemistry ...

INTERNATIONAL JOURNAL OF GREEN ENERGY Vol. 1, No. 2, pp. 167–191, 2004

Ultra-deep Desulfurization of Liquid Hydrocarbon Fuels: Chemistry and Process Chunshan Song, Ph.D.* and Xiaoliang Ma, Ph.D. Clean Fuels and Catalysis Program, The Energy Institute, The Pennsylvania State University, University Park, Pennsylvania, USA

ABSTRACT Recently, ultra-deep desulfurization of liquid hydrocarbon fuels is becoming very important worldwide not only because of the heightened interest for cleaner air and thus increasingly stringent environmental regulations for fuel sulfur content, but also because of the great need for making ultra-low-sulfur fuels used in hydrocarbon fuel process for fuel cell applications. This article is a selective review on chemistry and process concerning the ultra-deep desulfurization of liquid hydrocarbon fuels. The principles and problems for the existing hydrodesulfurization processes and the challenges, concepts, advantages, and disadvantages of various new approaches are discussed, including (1) sulfur compounds in liquid hydrocarbon fuels; (2) Reactivity and mechanistic aspect of various sulfur compounds; (3) Challenges in ultra-deep desulfurization processes; (4) Approaches to ultra-deep desulfurization process. Key Words: Desulfurization; Adsorption; Hydrodesulfurization; Gasoline; Diesel fuel; Fuels; Catalysis.

*Correspondence: Chunshan Song, Ph.D., Clean Fuels and Catalysis Program, The Energy Institute, The Pennsylvania State University, 209 Academic Projects Building, University Park, PA 16802, USA; Fax: (814) 865-3248; E-mail: [email protected]. 167 DOI: 10.1081/GE-120038751 Copyright & 2004 by Marcel Dekker, Inc.

1543-5075 (Print); 1543-5083 (Online) www.dekker.com

ORDER

REPRINTS

168

Song and Ma

1. INTRODUCTION Desulfurization process is one of the most important processes in the field of petroleum refining. Many hydrotreating processes in petroleum refining are commonly referred to as or include hydrodesulfurization (HDS) process. The major goal of desulfurization is to reduce the sulfur concentration in fuels, including gasoline, jet fuel, kerosene, diesel, and heating oil to meet the need of environmental protection and to reduce the sulfur concentration in feed-stocks to prevent sulfur poisoning of noble metal catalysts used in the following processes. Table 1 shows the recent US EPA regulations for diesel fuels (EPA-Diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003; News-EPA, 2001), gasoline (EPA-Gasoline RIA, 1999; EPARFG-II, 1999), respectively. Currently the fuel specifications for all highway diesel fuels in the US, Japan, and Western Europe limit the sulfur content of the diesel fuels to be less than 0.05 wt% or 500 parts per million by weight (ppmw). The current specification for gasoline in the US is 350 ppmw. The new government regulations in many countries will further lower the contents of sulfur and aromatics in the year 2005–2010 (EPA-diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003). In January 2001, US EPA announced new rules that will require a reduction in sulfur content of highway diesel fuel to 15 ppmw from current 500 ppmw, starting from June 2006 (News-EPA, 2001). New gasoline sulfur regulations will require most refiners to meet a 30 ppmw sulfur average with an 80 ppmw cap for both conventional and reformulated gasoline by January 1, 2006 (EPA-Gasoline RIA, 1999; EPA-RFG-II, 1999). The ultra-deep desulfurization is commonly referred to as a desulfurization process to reduce sulfur level to 15–30 ppmw or less. Recently, ultra-deep desulfurization of transportation fuels is becoming very important worldwide not only because of the heightened interest for cleaner air and thus increasingly stringent environmental regulations for fuel sulfur content, but also because the great need for making ultra-low-sulfur fuels used in hydrocarbon fuel processor for fuel cell applications, especially for a proton exchange membrane fuel cell (PEMFC). As the sulfur compounds in liquid hydrocarbon fuels and the H2S produced from these sulfur compounds in the hydrocarbon reforming process are poisonous to both the catalysts in hydrocarbon fuel processor and the electrode catalysts in fuel cell stacks, the sulfur content in the liquid hydrocarbon fuels have to be reduced to a very low level ( thiophenes > BTs > BNTs DBTs without any alkyl group at the 4- and/or 6-position > dibenzothiophenes with one or two alkyl group(s) at the 4and/or 6-position(s). For the sulfur compounds without a conjugation structure between the lone pairs on S atom and the p-electrons on aromatic ring, including disulfides, sulfides, thiols, and tetrahydrothiophene, HDS occurs directly through hydrogenolysis pathway: RSH þ H2 ! RH þ H2 S RSR0 þ 2H2 ! RH þ H2 S þ R0 H RSSR0 þ 3H2 ! RH þ 2H2 S þ R0 H These sulfur compounds exhibit higher HDS reactivity than that of thiophenes by an order of magnitude because they have higher electron density on the S atom and weaker C–S bond. For the thiophenic compounds, in which the lone pairs on the S atom conjugate with the p-electrons on the ring, including thiophenes, BTs, DBTs, PTs, and BNTs, HDS over the commercial catalysts proceeds usually through two pathways, the hydrogenation pathway (hydrogenation followed by hydrogenolysis) and the hydrogenolysis pathway (direct elimination of S atom via C–S bond cleavage) (Devanneaux and Maurin, 1981; Houalla et al., 1978; Kim et al., 2003b; Ma et al., 1995a, 1996b; Sapre et al., 1980; Virnat, 1983; Vanrysselberghe and Froment, 1996), as shown in Sch. 1. The HDS reactivity of the thiophenic compounds is dominantly dependent on both the electron structure and the steric hindrance of alkyl groups. The reactions of the thiopheneic compounds proceed usually through the two pathways. For thiophene and BT, the total HDS reactivity is greater than that of DBT by about an order of magnitude (Kilanowski et al., 1978), because there are higher p-electron density at C(2)–C(3) and C(4)–C(5) bonds in thiophene and at C(2)–C(3) bond in BT, leading to their greater HDS reactivity through the hydrogenation pathway. The p-electron distribution on dibenzothiophene is more uniform, like benzene ring, resulting in its lower hydrogenation reactivity. Thus, HDS of DBT over commercial CoMo catalysts proceeds dominantly through the hydrogenolysis pathway. However, if one or two alkyl groups are attached at the 4- or/and 6-positions of DBT, the hydrogenolysis pathway will be blocked strongly and the hydrogenation pathway becomes dominant (Isoda et al., 1994a). In this case, the HDS reactivity will decrease significantly. These sulfur compounds are called the refractory sulfur compounds. Some researchers attributed the refractoriness of 4,6-DMDBT to the steric hindrance towards the adsorption on the active sites (Ma and Schobert, 1997c; Shafi and Hutchings, 2000). On the basis of the heats of adsorption for DBT, 4-MDBT, and 4,6-DMDBT, which were obtained by measuring the parameters in a simplified Langmuir-Hinshelwood equation, Kabe and co-workers concluded that 4-MDBT or 4,6-DMDBT can be adsorbed more strongly on the catalyst through p-electrons in

ORDER

REPRINTS

172

Song and Ma Hydrogenation pathway

BT

Hydrogenolysis pathway

DBT

S

S

S

Hydrogenation pathway S

Hydrogenolysis pathway

Scheme 1. Hydrogenation pathway and hydrogenolysis pathway for thiophenic sulfur compounds.

the aromatic rings than DBT, and the retarding effect of the methyl substitution on HDS rates of DBTs was not attributed to the inhibition of the adsorption of DBTs on the catalyst but to the steric hindrance to the C–S bond scission of the adsorbed DBTs (Kabe et al., 1993; Zhang et al., 1996). Recently, based on the study in the promoter effect of Co and Ni on HDS of DBTs, Batail et al. proposed that the lower reactivity of 4,6-DMDBT than that of DBT over the promoted catalysts could not be attributed to difference in the adsorption strength of the reactants, but to a steric hindrance by the methyl groups to the adsorption of the dihydrointermediates and/ or others (Batail et al., 2000). Recently, Ma et al. compared the competitive adsorption of DBT, 4-MDBT, 4,6-DMDBT, and 2-methylnaphthalene over sulfided NiMo and CoMo catalysts (Ma et al., 2003a). They found that the catalysts exhibit much higher adsorption selectivity than 2-MNAPH, as shown in Fig. 2, indicating that the interaction between S atom in DBTs and the catalyst surface plays an important role in adsorption. In comparison of DBT, 4-MDBT, and 4,6-DMDBT, increases of adsorption selectivity in the order of 4,6-DMDBT < 4-MDBT < DBT implies that the methyl groups at the 4- and 6-positions inhibit the interaction between the S atom in DBTs and the active sites on the catalyst. This result supports that the steric hindrance of methyl groups at the 4- and/or 6-positions towards the adsorption of 4,6-DMDBT on the active sites is a key factor for the refractoriness of 4,6-DMDBT. Some coexisting compounds in petroleum fractions and in the products during the HDS process exhibit a strong effect on HDS. H2S was found to be one of the main inhibitors of the hydrogenolysis pathway (Broderick and

ORDER

REPRINTS

Ultra-deep Desulfurization of Liquid Hydrocarbon Fuels

173

Concent Conc ntra ration on aat out outlet,, mmol ol/l

6 5 2-M 2-MNAPH, , CoM CoMo DBT BT,, CoMo CoM 4-MDBT 4-M BT,, CoM CoMo 4,6-DMDBT 4,6-D BT,, Co CoMo 2-MNAPH,, NiMo 2-M DBT BT,, NiMo N 4-MDBT 4-M BT,, NiMo 4,6-DMDBT 4,6-D BT,, NiMo

4 3 2 1 0

0

1

2

3 4 5 6 rams of fuel fue per per gra gram of ccatalys yst,, g/ g/g Gra

7

Figure 2. The break-through curves of sulfur compounds and 2-methylnaphthalene over sulfided NiMo and CoMo at 50 C (Kim et al., 2003). (View this art in color at www.dekker.com.)

Gates, 1981; Farag et al., 1999; Isoda et al., 1995; Nagai and Kabe, 1983), while polyaromatic compounds were found to be the main inhibitors of the hydrogenation pathway (Farag et al., 1999; Girgis and Gates, 1991; Isoda et al., 1994b; Nagai and Kabe, 1983;). Basic nitrogen compounds affect both the hydrogenolysis pathway and the hydrogenation pathway. Some excellent reviews concerning this aspect are available in the literature (Girgis and Gates, 1991; Whitehurst et al., 1998). With respect to the characterization of catalyst surface structure, various models for the active site structures have been proposed on the basis of surface characterization using X-ray diffraction (XRD), scanning electron microscopy (SEM), transmission electron microscopy (TEM), and Mo¨ssbauer emission spectroscopy (MES). Especially within the past two decades, the detailed structural elucidation of molybdenum sulfide species has been achieved through the use of high-energy techniques such as X-ray photoelectron spectroscopy (XPS), extended X-ray absorption fine structure (EXAFS) spectroscopy, and X-ray absorption nearedge structure (XANES) spectroscopy. Recently, an atomically resolved scanning tunneling microscopy technology has been used to identify HDS active site on MoS2 nanoclusters in situ, which provides a fundamental insight into the HDS catalytic mechanism (Lauritsen et al., 2003; Topsøe, 2003). The coordination chemistry investigations have found that there are S-, Z2-, Z4-, 5 Z -, Z4-/S-m2-, and Z4-/S-m3-bound models in organometallic complexes for thiophene coordination with transition metal complexes (Angelici, 2001; Bianchini and Meli, 1998; Jones et al., 1997; Vecchi et al., 2003), as shown in Fig. 3. These models indicate likely adsorption configurations of thiophene on the active sites of the catalyst surface.

ORDER

REPRINTS

174

Song and Ma

S S

S

M

S

M

M ηS

M

S-µ3

1

M S

η

M

4

η5

M

M S

S

S M η 4, S-µ 2

M η 4, S-µ3

M η2

M η 1C

Figure 3. Known coordination geometries of thiophene in organometallic complexes, indicating likely adsorption configurations of thiophenic compounds on the surface of adsorbents.

Since it is still difficult experimentally to characterize the chemical properties of the catalyst surface and the adsorbates, especially at the actual reaction conditions, many attempts at the molecular simulation of HDS over molybdenum-sulfide-based catalysts have been made on the basis of various cluster models and different softwares. The molecular simulation provides an important insight into the HDS mechanism. On the basis of the fundamental studies, many models of HDS active site have been proposed. The typical models include the Rim-Edge model by Daage and Chianelli (1994), the Co-Mo-S phase model by Topsoe et al. (1996), the remote control model by Delmon et al. (Delmon and Grange, 1988; Grange and Vanhaeren, 1997) and 1 010 and 303 0 edge model by Ma and Schobert (1997a, 1997b, 1997c 2000).

4. CHALLENGES IN ULTRA-DEEP DESULFURIZATION PROCESSES Recently, investigations have demonstrated that sulfur compounds remaining in diesel fuels at sulfur level lower than 500 ppm are dominantly the DBTs with alkyl substituents at the 4- and/or 6-positions, and are lower in HDS reactivity (Gates and Topsoe, 1997; Kabe et al., 1992; Ma et al., 1995a, 1996a, 1996b, 2002b; Vanrysselberghe and Froment, 1996). These species are termed refractory sulfur compounds. Both steric hindrance and electronic factor are responsible for the observed low reactivity of 4- and 6-substituted DBTs (Ma et al., 1995b). Based on the HDS reactivity of a gas oil (Ma et al., 1995a, 1996b, 2002b), the sulfur compounds can be classified into four groups according to their HDS reactivities that were described by the pseudo-first-order rate constants. The first group consists of alkyl BTs; the second, DBT and alkyl DBTs without alkyl

ORDER

REPRINTS


175

10000 10000.0

Pre resent nt 500 ppm

Sulfu fur ccon ontent, t, p ppmw mw

1000. 1000.0

Total S Grou oup-1 -1 Grou oup-2 -2 Grou oup-3 -3 Grou oup-4 -4

2005 30 ppm 2006 15 ppm

100.0 100.

Fue uel cell c pplication on appl 0.1 ppm

10. 10.0

1.0 2.5 1. 1.0

3.2 ~7 0.1 0. 0

0

1

2

3

4

5

6

7

Relati ative ve vol volume me of catal catalys yst b bed Figure 4. Simulated HDS of diesel to meet 15 and 0.1 ppm level on the basis of a conventional single-stage reactor, assuming 1.0 wt% S in feed; HDS kinetic model: CS,total ¼ CS1o ek1t þ CS2o ek2t þ CS3o ek3t þ CS4o ek4t (Song and Ma, 2003). (View this art in color at www.dekker.com.)

substituents at the 4- and 6-positions; the third, alkyl DBTs with only one alkyl substituent at either the 4- or 6-position; the fourth, DBTs with alkyl substituents at the 4- and 6- positions. The sulfur distribution in the four groups in the gas oil is 39, 20, 26, and 15 wt%, respectively, and the relative HDS rate constant for each of the four groups is 36, 8, 3, and 1, respectively (Ma et al., 2002b). Figure 4 shows the relative volume of catalyst bed requirements for various levels of sulfur removal by conventional single-stage HDS of diesel fuels (Song and Ma, 2003). According to this result, when the total sulfur content is reduced to 500 ppmw, the sulfur compounds remaining in the hydrotreated oil are the third and fourth group sulfur compounds. When the total sulfur content is reduced to 30 ppmw, the sulfur compounds remaining in the hydrotreated oil are only the fourth group sulfur compounds, indicating that the less the sulfur concentration is, the lower the HDS reactivity of the remainder sulfur compounds. To put these problems into perspective based on conventional approaches for HDS of diesel fuels, if reducing the sulfur level from current 500 to 15 ppmw (the regulation in 2006) by a conventional single-stage HDS process, the volume of catalyst bed will have to be increased by 3.2 times as that of the current HDS catalyst bed. If reducing the sulfur level to 0.1 ppmw by the conventional HDS process for fuel cell application, the volume of catalyst bed will have to be increased by about seven times. As well known, increasing volume of high temperature and high pressure reactor is very expensive. In another scenario, with current commercial HDS processes without changing the reactor volume the catalyst activity will have to

ORDER

REPRINTS

176

Song and Ma

be increased by a factor of 3.2 and 7 to meet the new regulation and PEM fuel cell applications, respectively. It might be hard to meet such a demand by making small incremental improvements in the existing hydrotreating catalysts that have been developed during the last 50 years. For desulfurization of sulfur species in gasoline, it is not difficult to remove the sulfur compounds in naphtha range by current catalytic HDS processes. For the US refineries, most sulfur in gasoline pool is found in FCC naphtha. The challenge in deep desulfurization of FCC naphtha is selective conversion of sulfur compounds without saturation of olefinic compounds, which account for about 15–25 wt% in FCC naphtha and contribute to octane number enhancement. High hydrogen consumption in ultra-deep hydrodesulfurization of FCC naphtha is another issue that needs to be considered. For straight run kerosene that is used for making jet fuels, the sulfur removal by HDS is more difficult than that from naphtha but less difficult than that from gas oil. Therefore, the petroleum refining industry is facing a major challenge to meet the new stricter sulfur specifications and the need for fuel cell applications in the early 21st century when the quality of the crude oils continue to decline in terms of increased sulfur content and decreased API gravity.

5. APPROACHES TO ULTRA-DEEP DESULFURIZATION PROCESS Many approaches have been developed for ultra-deep desulfurization. These approaches include: (1) improving catalytic activity by new catalyst formulation for HDS of 4,6-DMDBT; (2) tailoring reaction and process conditions; (3) designing new reactor configurations; and (4) developing new processes. One or more approaches may be employed by a refinery to meet the challenges of producing ultra-clean fuels at affordable cost.

5.1. Improving Catalytic Activity by New Catalyst Formulation Design approaches for improving catalytic activity for ultra deep hydrodesulfurization of diesel fuel focus on how to remove 4,6-DMDBT more effectively, by modifying catalyst formulations to (1) enhance hydrogenation of aromatic ring in 4,6-DMDBT by increasing hydrogenating ability of the catalyst; (2) incorporate acidic feature in catalyst to induce isomerization of methyl groups away from the 4- and 6-positions; and (3) remove inhibiting substances (such as H2S) and tailoring the reaction conditions for specific catalytic functions. The catalytic materials formulations may be improved for better activity by using different supports (carbon, TiO2, TiO2-Al2O3, HY, MCM-41, etc.) for conventional aluminasupported CoMo, NiMo, and NiW catalysts; by increasing loading level of active metal (Mo, W, etc.); by adding one more base metal (e.g., Ni to CoMo or Co to NiMo); and by incorporating a noble metal (Pt, Pd, Ru, etc.). The catalyst development has been one of the focuses of industrial research and development for deep hydrodesulfurization. For example, new and improved

ORDER

REPRINTS


177

catalysts have been developed and marketed by Akzo Nobel, Criterion, Haldor Topsoe, IFP, United Catalyst/Sud-Chemie, Advanced Refining, ExxonMobil, Nippon Ketjen in Japan, and RIPP in China. Akzo Nobel has developed and commercialized various catalysts that can be used for HDS of diesel feed: KF 752, KF 756, and KF 757, and KF 848 (Akzo Nobel, 2002). KF 752 can be considered to be typical of an Akzo Nobel catalyst of the 1992–1993 timeframe. KF 756 is a Co-Mo catalyst with high HDS activity. KF 757 is Akzo Nobel’s latest Co-Mo with higher HDS activity and optimized pore structure; it was announced in 1998 (Akzo Nobel, 2002). KF 757 (Co-Mo) and KF 848 (Ni-Mo) were developed by using what Akzo Nobel calls STARS (Super Type II Active Reaction Sites) technology. Type II refers to a specific kind of catalyst site, which is more effective for removing sulfur from sterically hindered compounds. KF 848 was announced in 2000 (Gerritsen, 2000). KF 848 Ni-Mo STARS is 15–50% more active than KF 757 Co-Mo STARS under medium to high pressure. In terms of sulfur removal, Akzo Nobel projects, in which a desulfurization unit produces 500 ppmw sulfur with KF 752, would produce 405, 270, and 160 ppm sulfur with KF 756, KF 757, and KF 842, respectively (EPA-Diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003). More recently, NEBULA catalyst has been developed jointly by Exxon Mobil, Akzo Nobel, and Nippon Ketjen and commercialized in 2001 (Meijburg, 2001). NEBULA stands for New Bulk Activity and is bulk base metal catalyst without using a porous support. The NEBULA-1 catalyst is even more active than KF 848 STARS catalyst with respect to HDS and HDN and diesel hydrotreating; it has been successfully applied in several diesel hydrotreaters for months as of early 2002 (Meijburg, 2001). Figure 5 shows the relative activity of the new NEBULA and STARS catalysts compared to conventional CoMo/Al2O3 developed over the last 50 years (Plantenga, 2002). Haldor-Topsoe has also developed a more active catalyst. Its TK-554 catalyst is analogous to Akzo Nobel’s KF 756 catalyst, while its newer, more active catalyst is termed TK-574. For example, in pilot plant studies, under conditions where TK-554 produces 400 ppmw sulfur in SRGO, TK 574 will produce 280 ppmw (EPA-Diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003). Criterion Catalyst Company announced two new lines of catalysts. One is called Century, and the other is called Centinel. These two lines of catalysts are reported to be 45–70% and 80% more active, respectively, at desulfurizing petroleum fuel than conventional catalysts used in the mid-1990s (EPA-Diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003). For selective removal of sulfur from FCC naphtha by HDS with minimum hydrogenation of olefins, ExxonMobil and Akzo Nobel (ExxonMobil, 2001; Halbert et al., 2001; Kaufmann et al., 2000) jointly developed RT-225 catalyst.

5.2. Tailoring Reaction and Process Conditions Tailoring process conditions aims at achieving deeper hydrodesulfurization with a given catalyst in an existing reactor without changing the processing scheme, with no or minimum capital investment. The parameters include those that can be tuned

ORDER

178

REPRINTS

Song and Ma

Figure 5. Relative activity of the new NEBULA and STARS catalysts compared to conventional CoMo/Al2O3 developed over the last 50 years (Plantenga, 2002).

without any new capital investment (space velocity, temperature, pressure), and those that may involve some relatively minor change in processing scheme or some capital investment (expansion in catalyst volume or density, H2S scrubber from recycle gas, improved vapor–liquid distributor). First, space velocity can be decreased to increase the reactant-catalyst contact time. More refractory sulfur compounds would require lower space velocity for achieving deeper HDS. Second, temperature can be increased, which increases the rate of HDS. Higher temperature facilitates more of the high activation-energy reactions. Third, hydrogen pressure can be increased. Fourth, improvements can be made in vapor–liquid contact, which more effectively utilizes all surface area of the catalyst. Fifth, the concentration of hydrogen sulfide in the recycle stream can be removed by scrubbing. Since H2S is an inhibitor to HDS, its build-up in high-pressure reactions through continuous recycling can become significant. Finally, more volume of catalyst can be used, either through catalyst bed volume expansion or more dense packing. Some of these factors are elaborated further below. It should be noted that conventional approaches for fuel desulfurization in response to the 1993 diesel fuel sulfur regulation in the US were to increase process severity of HDS, increase catalysts to fuel ratio, increase residence time, and enhance hydrogenation, or to use additional low-sulfur blending stocks either from separate process streams or purchased. It is becoming more difficult to meet the new fuel specifications by fuel hydrodesulfurization using the conventional approaches. The decrease in the concentration of hydrogen sulfide in gas phase could reduce the inhibition of the desulfurization (Ma et al., 1994a, 1994b, 2002b; Zhang et al., 1996) and hydrogenation reactions. Role of H2S in deep HDS of gas oils has been discussed in detail by Sie (1999). H2S can be removed by chemical scrubbing. Haldor-Topsoe indicates that decreasing the concentration of hydrogen sulfide at the

ORDER

REPRINTS


179

Figure 6. ConocoPhillips’s S-Zorb sulfur removal process based on solid adsorbent and its continuous regeneration (Gislason, 2002; Phillips Petroleum, 2001).

inlet to a co-current reactor by 3–6 vol% can decrease the average temperature needed to achieve a specific sulfur reduction by 15–20 C, or reduce final sulfur levels by more than two-thirds (EPA-Diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003). UOP projects that scrubbing hydrogen sulfide from recycle hydrogen can reduce sulfur levels from roughly 285 to 180 ppmw in an existing hydrotreater (EPA-Diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003). The increase in hydrogen partial pressure and/or purity can improve hydrodesulfurization and hydrogenation. Haldor-Topsoe indicates that increasing hydrogen purity is preferable to a simple increase in the pressure of the hydrogen feed gas, since the latter will also increase the partial pressure of hydrogen sulfide later in the process, which inhibits both beneficial reactions (EPA-Diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003). Haldor-Topsoe projects that an increase in hydrogen purity of 30% would lower the temperature needed to achieve the same sulfur removal rate by 8–9 C. Or temperature could be maintained while increasing the amount of sulfur removed by roughly 40%.

ORDER

REPRINTS

180

Song and Ma

The improvement in vapor–liquid contact can enhance the performance of distillate hydrotreaters. Akzo Nobel estimates that an improved vapor–liquid distributor can reduce the temperature necessary to meet a 50 ppmw sulfur level by 10 C, which in turn would increase catalyst life and allow an increase in cycle length from 10 to 18 months (EPA-Diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003). Similarly, in testing of an improved vapor–liquid distributor in commercial use, Haldor-Topsoe found that the new distributor allowed a 30% higher sulfur feed to be processed at 25 C lower temperatures, while reducing the sulfur content of the product from 500 to 350 ppmw (EPA-Diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003).

5.3. Designing New Reactor Configurations Industrial reactor configuration for deep hydrodesulfurization of gas oils in terms of reaction order and effect of produced H2S has been discussed by Sie (1999). The reactor design and configuration involve both single-stage and two-stage desulfurization processes. Desulfurization processes in use today in the US generally use only one reactor, due to the need to only desulfurize diesel fuel to 500 ppmw or lower. Hydrogen sulfide strongly suppresses the activity of the catalyst for converting the refractory sulfur compounds, which should occur in the major downstream part of a cocurrent trickle-bed reactor during deep desulfurization. The normally applied cocurrent trickle-bed single reactor is therefore not the optimal technology for deep desulfurization (Sie, 1999). A second reactor can be used, particularly to meet lower sulfur levels. Both desulfurization and hydrogenation in the second reactor can be improved by removing H2S and NH3 from the exit gas of first reactor before entering the second reactor. This last technical change is to install a complete second stage to the existing, one-stage hydrotreater. This second stage would consist of a second reactor, and a high pressure, hydrogen sulfide scrubber between the first and second reactor. Assuming use of the most active catalysts available in both reactors, UOP projects that converts from a one-stage to a two-stage hydrotreater could produce 5 ppmw sulfur relative to a current level of 500 ppmw today (EPA-Diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003). A new way of reactor design is to have two or three catalyst beds, which are normally placed in separate reactors, within a single reactor shell and have both cocurrent and counter-current flows. This new design was pioneered by ABB Lummus and Criterion, as represented by their SynSat process (ABB Lummus Global Inc., 1998; Maxwell, 1997; Suchanek, 1996). Traditional reactors are cocurrent in nature. The hydrogen is mixed together with the distillate at the entrance to the reactor and flow through the reactor together. Because the reaction is exothermic, heat must be removed periodically. This is sometimes done through the introduction of fresh hydrogen and distillate at one or two points further down the reactor. The advantage of cocurrent design is practical, it eases the control of gas–liquid mixing and contact with the catalyst. The disadvantage is that the concentration of H2 is the highest at the front of the reactor and lowest at the outlet. The opposite is true for the concentration of H2S. This increases the difficulty of achieving extremely low sulfur levels due to the low H2 concentration and high H2S concentration at the end of the reactor. The normal solution to this

ORDER

REPRINTS


181

problem is to design a counter-current reactor, where the fresh H2 is introduced at one end of the reactor and the liquid distillate at the other end. Here, the hydrogen concentration is highest (and the hydrogen sulfide concentration is lowest) where the reactor is trying to desulfurize the most difficult (sterically hindered) compounds. The difficulty of counter-current designs in the case of distillate hydrotreating is vapor–liquid contact and the prevention of flooding and hot spots within the reactor. The SynAlliance (consisting of ABB Lummus, Criterion Catalyst Corp., and Shell Oil Co.) has patented a counter-current reactor design called SynTechnology. With this technology, in a single reactor design, the initial portion of the reactor will follow a co-current design, while the last portion of the reactor will be countercurrent (EPA-Diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003). In a two-reactor design, the first reactor will be co-current, while the second reactor will be counter-current. ABB Lummus estimates that the counter-current design can reduce the catalyst volume needed to achieve 97% desulfurization by 16% relative to a co-current design (EPA-Diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003). The impact of the counter-current design is even more significant when aromatics control (or cetane improvement) is desired in addition to sulfur control. However, operation of counter-current flow reactor might be not possible in packed beds of the usual catalyst particles because of the occurrence of flooding at industrially relevant fluid velocities. Some novel reactor concepts based on special structured packings or monoliths that allow such conter-current operation have been presented (Sie, 1999). 5.4. Developing New Processes Among the new process concepts, design approaches for ultra deep desulfurization focus on (1) adsorption and sulfur atom extraction—remove sulfur by using reduced metals to react with sulfur to form metal sulfides at elevated temperatures under H2 atmosphere without hydrogenation of aromatics; (2) selective adsorption of sulfur compounds—remove sulfur by selective interaction with sulfur compounds in the presence of aromatic hydrocarbons under ambient or mild conditions without hydrogen; (3) oxidation and extraction—oxidize sulfur compounds by liquid-phase oxidation reactions with or without ultrasonic radiation, followed by separation of the oxidized sulfur compounds; and (4) biodesulfurization—attack sulfur atoms by using bacteria via microbial desulfurization.

5.4.1. Sulfur Adsorption and S-Zorb Process Phillips Petroleum conducted an internal study of its refineries and concluded the use of hydrotreating technologies to reach ultra-low sulfur levels in gasoline to be a cost-prohibitive option (Phillips Petroleum, 2001). A prospective diesel desulfurization process, S-Zorb Diesel, was recently announced by Phillips Petroleum, which is an extension of their S-Zorb process for gasoline (at 377–502 C, 7.0–21.1 kg/cm2) (Phillips Petroleum, 2001). S-Zorb for diesel contacts highway diesel fuel (typically with about 350 ppmw sulfur) with a solid sorbent in a fluid bed reactor at relatively

ORDER

REPRINTS

182

Song and Ma

low pressures and temperature in the presence of hydrogen. The sulfur atom in the sulfur-containing compounds is adsorbed onto the sorbent and the reacts with the sorbent. Phillips Petroleum uses a proprietary sorbent that attracts sulfur-containing molecules and removes the sulfur atom from the molecule. The sulfur atom is retained on the sorbent while the hydrocarbon portion of the molecule is released back into the process stream. Hydrogen sulfide is not released into the product stream and therefore prevents recombination reactions of hydrogen sulfide and olefins to make mercaptans. Phillips Petroleum (now ConocoPhillips) also developed a new S-Zorb process for making low-sulfur gasoline (Phillips Petroleum, 2001). The first commercial S-Zorb-Gasoline unit began operations in a refinery in Texas, USA, in early 2001 (EPA-Diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003). Figure 6 illustrates the principle of S-Zorb process and shows the scheme of S-Zorb Gasoline process (Gislason, 2002; Phillips Petroleum, 2001).

5.4.2. Selective Adsorption for Removing Sulfur (PSU-SARS) A PSU-SARS process is being explored at Pennsylvania State University’s Energy Institute (Ma et al., 2001, 2002a). The process is shown in Fig. 7. This process can remove organic sulfur compounds from distillate fuels (diesel, gasoline, and jet fuels) at low temperatures and ambient pressure without the use of hydrogen. The process uses a transition-metal-based adsorbent that selectively adsorbs the sulfur compounds, but leaves aromatic compounds unaffected. The major potential advantage of the SARS process for ultra-deep desulfurization of the distillate fuels are that the adsorption process works at ambient temperature and ambient pressure without using hydrogen and the adsorbent has high selectivity toward sulfur and high adsorption capacity. The major challenge in this process is to develop the new adsorbent with high capacity, selectivity, and regenerability. Recently, PSU group is exploring two promising types of adsorbents, including the metal-sulfide-based and metal-oxide-based adsorbents, which can be regenerated facilely by the solvent washing or the air oxidation (Ma et al., 2003; Watanabe et al., 2003).

5.4.3. Adsorption Desulfurization Using Alumina Based or Zeolites Based Adsorbents The IRVAD process by Black & Veatch Pritchard Inc. and Alcoa Industrial Chemicals is claimed to be a low-cost process for low-sulfur gasoline (Hydrocarbon Processing, 1999; Irvine, 1998). The process uses an alumina-based selective adsorbent to counter-currently contact liquid hydrocarbon in a multistage adsorber. The adsorbent is regenerated in a continuous cross-flow reactivator using heated reactivation gas. The process operates at lower pressure, does not consume hydrogen or saturate olefins. Recently, Yang’s group reported that they developed a Cu(I)-Y

ORDER

REPRINTS


183

Figure 7. The proposed adsorption process integrated with HDS in a refinery (SARSHDSCS) for deep desulfurization (Ma et al., 2001; 2002a).

Zeolite adsorbent for adsorptive desulfurization of diesel fuel and gasoline via p-complexation at ambient conditions (Yang et al., 2003a, 2003b). Research Triangle Institute (RTI) is developing another reactive adsorption process called TReND (transport reactor for naphtha desulfurization) which is based on metal oxide sorbent. The feature of TReND process is that supported metal oxide-based regenerable sorbent is used to capture the sulfur in a transport reactor which is similar to FCC reactor (Gupta, 2003). RTI has conducted extensive studies on desulfurization of synthesis gas from coal gasification (Gupta et al., 2001), and recently applied their expertise in H2S removal using metal oxide to organic sulfur removal from liquid fuels at 800–1000 F (426–535 C) with or without the presence of H2 gas feed in the TReND process (Turk and Gupta, 2001). Figure 8 shows the scheme of the transport reactor for TReND process (Turk et al., 2002a, 2002b).

5.4.4. Oxidation and Extraction for Desulfurization Oxidation of sulfur atom in liquid phase, with or without radiation by ultrasound or UV light, followed by extraction of oxidized species can lead to desulfurization of diesel fuels. Otsuki et al. studied the oxidation of dibenzothiophene (DBT) using t-butyl hypochlorite (t-BuOCl) in the presence of several catalysts (Otsuki et al., 2001). In a flow reactor under ambient pressure at 30–70 C, more than 90% of DBT could be oxidized in the decahydronaphthalene (decalin)

ORDER

184

REPRINTS

Song and Ma

Figure 8. RTI’s transport reactor for naphtha desulfurization (TReND) process (Turk et al., 2002a, 2002b).

solution. The catalyst was necessary to oxidize dibenzothiophene with t-BuOCl, and gamma-Al2O3 supported catalysts have relatively high activities. Aida et al. discussed the oxidation processes for reducing the sulfur content in diesel fuel (Aida et al., 2000). The oxidation method has capabilities, not only to decrease the sulfur content in light oil below 0.1 ppm, but also to recover the sulfur component as organic sulfur compound such as sulfone of dibenzothiophene derivatives that has a potential industrial use (Aida et al., 2000). Shiraishi et al. reported photochemical desulfurization using hydrogen peroxide (H2O2) aqueous solution extraction system for high-sulfur-content straight-run light gas oil and aromatic-rich light cycle oil (LGO) (Shiraishi et al., 1999). In the case of LGO of high sulfur content, 77% of the sulfur is removed by 36 h of photoirradiation, and the quantity of sulfur removed from LGO is six-fold greater than that in the case of commercial light oil (CLO). GC-AED analysis shows that benzothiophenes in all the feedstocks are more easily desulfurized than dibenzothiophenes. Highly substituted dibenzothiophenes in LCO, especially those having substituted carbon number of 4–6, are hardly desulfurized by the proposed method (Shiraishi et al., 1999). With respect to commercial oxidation process development, Petrostar recently announced a desulfurization technology which removes sulfur from diesel fuels using chemical oxidation (Chapados, 2000). Desulfurization of diesel fuel is accomplished

ORDER

REPRINTS


185

by first forming a water emulsion with the diesel fuel. In the emulsion, the sulfur atom is oxidized to a sulfone using catalyzed peroxyacetic acid. With an oxygen atom attached to the sulfur atom, the sulfur-containing hydrocarbon molecules becomes polar and hydrophilic and then move into the aqueous phase. EPA has indicated that there is another chemical oxidation process which currently is in the patenting process (EPA-Diesel RIA, 2000; EPA Nonroad Diesel RIA, 2003). This process is similar to the Petrostar process, except instead of keeping the sulfone intact, this process separates the oxidized sulfur atom from the hydrocarbon immediately after the oxidation reaction.

5.4.5. Biodesulfurization Biodesulfurization is a process that removes sulfur from fossil fuels using a series of enzyme-catalyzed reactions (Monticello, 1998). Several recent reviews outlines the progress in the studies of the microbial desulfurization from the basic and practical point of view (McFarland et al., 1998; Monticello, 1998; Ohshiro and Izumi, 1999). Biodesulfurization is rising as one of the candidates for ultra-deep desulfurization process. Recently, utilization of byproduct from biodesulfurization has also been reported and the process system is being explored (Lange and Lin, 2000). Biocatalytic sulfur removal from fuels has applicability for producing low sulfur gasoline and diesel fuels. Certain microbial biocatalysts have been identified that can biotransform sulfur compounds found in fuels, including ones that selectively remove sulfur from dibenzothiophene heterocyclic compounds (McFarland et al., 1998). Energy Biosystems is developing a biodesulfurization process (Monticello, 1998). It involves the removal of sulfur-containing hydrocarbon compounds from distillate or naphtha streams using bacteria. As indicated by McFarland et al. (McFarland, 1999; McFarland et al., 1998), microbial sulfur-specific transformations have been identified that selectively desulfurize organic sulfur compounds in fossil fuels. Recent discoveries related to biodesulfurization mechanisms may lead to commercial applications of biodesulfurization through engineering recombinant strains for over-expression of biodesulfurization genes, removal of end product repression, and/or by combining relevant industrial and environmental traits with improvements in bioprocess design.

CONCLUSIONS Heightened concerns for cleaner air and increasingly more stringent regulations on sulfur contents in transportation fuels and nonroad fuels will make desulfurization more and more important. The sulfur problem is becoming more serious in general, because the regulated sulfur contents are getting an order of magnitude lower, while the sulfur contents of crude oils refined in the US are becoming higher. The challenge for gasoline deep desulfurization is the selective HDS of thiophenic compounds in FCC naphtha without a significant loss in octane number and in yield of naphtha. The challenge for deep desulfurization of diesel

ORDER

REPRINTS

186

Song and Ma

fuels is the difficulty of removing refractory sulfur compounds, particularly 4,6dimethyldibenzothiophene, with conventional hydrodesulfurization processes. The problem is exacerbated by the inhibiting effects of polyaromatics and nitrogen compounds as well as H2S on deep HDS. New catalysts that have been developed in the recent past focus on higher activity for hydrogenation to enhance 4,6-DMDBT conversion and higher activity for HDN (for removal of nitrogen compounds at low concentrations). New and improved catalysts with higher hydrogenation activity coupled with improved reactor operation under more severe conditions can improve deep hydrodesulfurization for meeting EPA 2006 regulations (15 ppmw sulfur in highway diesel by 2006 and in nonroad diesel by 2010). Some new processing approaches show advantages with further deeper desulfurization (