- Email: [email protected]
Effect of basic nitrogen compounds on gas oil hydrodesulfurization and deposit formed on the catalyst
Fuel 153 (2015) 455–463
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Effect of basic nitrogen compounds on gas oil hydrodesulfurization and deposit formed on the catalyst Ryutaro Koide a,b,⇑, Yoshimu Iwanami a, Shozaburo Konishi a, Nobuharu Kimura a, Shinya Takahashi a, Masato Kamata a, Toshihide Baba b a
Central Technical Research Laboratory, JX Nippon Oil & Energy Corporation, 8 Chidoricho, Naka-ku, Yokohama 231-0815, Japan Department of Environmental Chemistry and Engineering, Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, 4259 Nagatsuta, Midori-ku, Yokohama 226-8502, Japan b
h i g h l i g h t s Residue desulfurization gas oil contained basic nitrogen compounds like anilines. Light cycle oil contained polyaromatic hydrocarbons, indoles and carbazoles. A low deactivation rate was observed for feed with residue desulfurization gas oil. The deposit from feed with light cycle oil was clarified as aromatic.
a r t i c l e
i n f o
Article history: Received 6 September 2014 Received in revised form 9 January 2015 Accepted 25 February 2015 Available online 16 March 2015 Keywords: Residue desulfurization gas oil Hydrodesulfurization Light cycle oil Coke Catalyst Basic nitrogen compounds
a b s t r a c t The authors studied the effect of the composition of cracked gas oils on the hydrodesulfurization to obtain clean diesel fuel. The required reaction temperature for sulfur specification of diesel fuel using a feedstock mixed with residue desulfurization gas oil (RDS-GO) was equivalent to that using light cycle oil (LCO). RDS-GO has a high content of basic nitrogen compounds, whereas LCO has a high content of aromatic hydrocarbons that are known to inhibit the hydrodesulfurization of gas oil. Using two-dimensional gas chromatography, LCO was found to contain polyaromatic hydrocarbons and non-basic nitrogen compounds, such as indoles and carbazoles, whereas RDS-GO contained monoaromatic hydrocarbons and basic nitrogen compounds, such as anilines. In a deactivation test conducted over 2300 h, the feedstock mixed with RDS-GO showed a lower deactivation rate than LCO. Raman spectroscopy, 13C NMR, and electron energy-loss spectroscopy in transmission electron microscopy showed that the deposits formed on the used catalysts during the deactivation test with the feedstock mixed with LCO contained more aromatic compounds than those with RDS-GO. The coke in the deposits from the feed containing RDS-GO was expected to be more anisotropic than that from the LCO. The improved hydrodesulfurization activity observed with RDS-GO was attributed to its characteristic components, which are thought to inhibit the growth of coke on the catalyst. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Abbreviations: LCO, light cycle oil; FCC, fluid catalytic cracking; HDS, hydrodesulfurization; SRGO, straight run gas oil; RDS, residue hydrodesulfurization; GO, gas oil fraction; VDS, vacuum gas oil hydrogenation; TEM-EELS, electron energy-loss spectroscopy in transmission electron microscopy; 2D-GC, two-dimensional gas chromatography; XRD, X-ray diffraction; LHSV, liquid hourly space velocity; XRF, X-ray fluorescence; FID, flame ionization detector; NCD, nitrogen chemiluminescence detector ; HAADF, high-angle annular dark-field; 4,6-DMDBT, 4,6-dimethyldibenzothiophene. ⇑ Corresponding author at: Central Technical Research Laboratory, JX Nippon Oil & Energy Corporation, 8 Chidoricho, Naka-ku, Yokohama 231-0815, Japan. Tel.: +81 45 625 7301. E-mail address: [email protected] (R. Koide). http://dx.doi.org/10.1016/j.fuel.2015.02.112 0016-2361/Ó 2015 Elsevier Ltd. All rights reserved.
In recent years, refineries have had to utilize many intermediate fractions from various types of refining units [1]. ‘‘Cracked gas oils’’ are gas oil fractions obtained from hydrotreating or cracking units using heavy feedstocks. Light cycle oil (LCO) is a cracked gas oil that is obtained as a gas oil fraction from a fluid catalytic cracking (FCC) unit, and is widely used as a component of industrial heavy fuel oil or banker fuel oil because it has a lower viscosity and sulfur content than other potential components. The recent use of LCO as a feedstock for gas oil hydrodesulfurization (HDS) units has
456
R. Koide et al. / Fuel 153 (2015) 455–463
resulted in the upgrading of heavy industrial fuel oil into clean diesel fuel for transportation [2]. During gas oil HDS, LCO is hydrotreated by mixing it with a straight run gas oil (SRGO) fraction from crude oil [3]. However, LCO contains large amounts of aromatic hydrocarbons, which results in a highly exothermic reaction and high hydrogen consumption. Moreover, the HDS of feedstocks containing LCO results in rapid deactivation of the HDS catalyst caused by the accelerated deposition of coke on its surface [3,4]. This behavior limits the amount of LCO that can be added to a gas oil HDS unit. Meanwhile, other intermediate gas oil fractions from various units have mainly been used as components of industrial heavy fuel oil due to their having lower viscosities and lower sulfur contents than the other residual components in industrial fuel oil [5]. A residue hydrodesulfurization (RDS) unit produces a gas oil fraction (RDS-GO) as a by-product. RDS-GO is utilized as a diluting fraction in industrial heavy fuel oil, and is expected to become a feedstock for the gas oil HDS unit as well as LCO. A vacuum gas oil hydrogenation (VDS) unit also produces a gas oil fraction (VDS-GO) that is used as a diluting fraction. RDS-GO and VDS-GO are classified as ‘‘hydrotreated gas oils’’, and are considered to be good candidates for use as a feedstock for gas oil HDS units due to their having lower sulfur contents than SRGO. RDS-GO is particularly promising because it is produced in larger volumes in refineries than is VDS-GO. The use of RDS-GO in the gas oil HDS reaction and its deactivation behavior are not well understood. When using RDS-GO as a feedstock for gas oil HDS, it is important to know the composition of the gas oil feedstock and understand the effect the components will have on HDS activity and the deactivation of the HDS catalysts during long-term operation. It has been reported that the aromatic compounds in LCO can poison the catalyst during gas oil HDS [6,7], and that nitrogen compounds also affect the catalyst’s gas oil HDS activity [8]. Therefore, an investigation of the hydrocarbon and nitrogen-containing hydrocarbons in the feedstock is significant, as carbonaceous deposits on the catalyst are presumed to cause the deactivation of HDS activity in the gas oil HDS unit [9,10]. Several analytical methods have been used to characterize these carbonaceous compounds, including Raman spectroscopy [11,12], temperature programmed oxidation [12], and electron energy-loss spectroscopy in transmission electron microscopy (TEM-EELS) [13,14]. The purpose of this study is to investigate the composition of cracked gas oil and the effect of the feedstock composition on gas oil HDS. Two-dimensional gas chromatography (2D-GC) was used to clarify the types of hydrocarbons and nitrogen-containing compounds contained in the feedstock. Reactivity tests and deactivation tests were used to evaluate the effect of the feedstock composition on HDS activity and deactivation of gas oil. Furthermore, the deposits on the used catalysts were characterized using 13C NMR, Raman spectroscopy, TEM-EELS, and X-ray diffraction.
Table 1 Properties of the gas oil fractions used in the reactivity test and the deactivation test. Property
SRGO
LCO
VDSGO
RDSGO-A
RDSGO-B
RDSGO-C
Density (g/cm3) 90% boiling pointa (K) Sulfurb (mass ppm) 4,6-DMDBTc (mass ppm) Nitrogend (mass ppm) Basic nitrogene (mass ppm) Aromatic compoundsf (vol%)
0.8511 618.2 10,500 155 110 50
0.9493 624.7 4100 115 470 40
0.8769 662.7 870 105 250 100
0.8738 638.2 730 68 350 190
0.8609 604.2 180 80 83 56
0.8596 557.7 80 21 64 43
26
77
44
43
38
42
MIX-D 638.4
MIX-E 633
MIX-F 638.4
MIX-G MIX-H 627.4 629.6
SRGO Feed nameg 625.7 Required reaction temperature for sulfur content of 8 mass ppm in product oilh (K) a
Measured by ASTM D86. Determined by X-ray fluorescence (XRF). c 4,6-Dimethyldibenzothiophene determined by GC-sulfur chemiluminescence detector. d Measured by ASTM D3228. e Measured by UOP 269. f Measured by JPI-5S-49. g MIX-D, -E, -F, -G, and -H were prepared by mixing 85 vol% of SRGO and 15 vol% of LCO, VDS-GO, RDS-GO-A, RDS-GO-B, and RDS-GO-C, respectively. h The reaction conditions are: P = 5.5 MPaG, LHSV = 1.5 h 1, H2/Oil = 250 NL/L. b
2.2. Reactivity test The reactivity of the cracked gas oils was tested using a highpressure bench-scale fixed bed reactor. The feedstock properties are summarized in Table 1. A CoMo catalyst (100 mL; MoO3: 22.0 mass%, CoO: 3.0 mass%, Al2O3 support) for gas oil HDS was used for these tests after presulfiding with excess amount of sulfur reagent for sulfiding molybdenum on the catalyst with solvent. The following test conditions were used: hydrogen pressure: 5.5 MPaG, liquid hourly space velocity (LHSV): 1.0 h 1, ratio of hydrogen to oil (H2/Oil): 250 NL/L. The product oil from the reactivity test was bubbled with nitrogen gas at room temperature for 8 h to remove hydrogen sulfides. The sulfur content of the product oil was measured by X-ray fluorescence (XRF) using an Axios spectrometer (PANalytical). 2.3. Deactivation test The deactivation tests were carried out for MIX-D and MIX-F with the same catalyst and the test conditions as were used for the reactivity test after a period of about 500–2300 h. The reaction temperature was adjusted to obtain a sulfur content of nearly 8 mass ppm every 72 h. The normalized reaction temperature of hydrodesulfurization was calculated by the activation energy of 113 kJ/molK, which was calculated by the reactivity test.
2. Experimental
2.4. Two-dimensional gas chromatography
2.1. Feedstock
The components of SRGO, LCO, RDS-GO-A, and VDS-GO were determined using 2D-GC (KT2006, ZEOX) with two columns: BPX-5 (30 m 0.25 mm, i.d. 0.25 lm) and BPX-50 (2 m 0.1 mm, i.d. 0.1 lm) and a flame ionization detector (FID) or nitrogen chemiluminescence detector (NCD). Helium gas was used as the carrier gas at a flow rate of 2.58 mL/min.
SRGO was obtained by distillation of a typical Arabian crude. LCO was obtained from a FCC unit of hydrodesulfurized residue. RDS-GO-A, -B, and -C were obtained from three different RDS units for hydrodesulfurization of atmospheric residue. VDS-GO was obtained from a vacuum gas oil HDS unit for hydrodesulfurization of vacuum gas oil. The properties of the gas oils are summarized in Table 1. For the reactivity and deactivation tests, MIX-D, -E, -F, -G, and -H were prepared by mixing 85 vol% of SRGO and 15 vol% of LCO, VDS-GO, RDS-GO-A, RDS-GO-B, and RDS-GO-C, respectively.
2.5. Used catalysts from the reactivity tests and deactivation tests After the reactivity tests and deactivation tests, residual gas oil in the used catalysts was removed by Soxhlet extraction with
R. Koide et al. / Fuel 153 (2015) 455–463
toluene until the extracted liquid turned clear. The treated catalysts were dried at room temperature for 12 h and then evacuated at 353 K for 2 h. 2.6.
13
C NMR spectra
The 13C NMR spectra of the used catalysts were measured with an NMR System 500 (Varian) equipped with a CP/MAS probe with a resonance frequency of 125 MHz. Each sample was measured using a pulse at 47.6° for 4 ms with a pulse interval of 10 s for 3900 scans. 2.7. Raman spectra The Raman spectra of the deposits on the used catalysts were measured with an NRS-5100 (JASCO) equipped with an Ar+ laser (532 nm, 0.1 MW) as the excitation source. The diffraction slit size was 25 lm 1000 lm. 2.8. Electron energy-loss spectroscopy in transmission electron microscopy TEM-EELS images and spectra of the used catalysts were measured using a JEM-ARM200F (JEOL) operated at 200 kV. High-angle annular dark-field (HAADF) images were obtained by performing Fourier transformations. 2.9. X-ray diffraction XRD spectra of the used catalysts were measured with a RINT2500 (RIGAKU) equipped with a Cu Ka source using a scanning rate of 1°/min. 3. Results and discussion 3.1. Effect of sulfur, nitrogen, basic nitrogen and aromatic compounds on reaction temperature Gas oil is desulfurized in HDS units at refineries to satisfy the diesel specifications in Europe, North America and Japan, which require that sulfur content be no higher than 10 mass ppm. To meet this requirement, most refineries control the reaction temperature of the gas oil HDS units to obtain a sulfur content of less than 10 mass ppm; there is often around 8 mass ppm of sulfur in the product oil. Therefore, the reactivities of SRGO, LCO, VDS-GO, RDS-GO-A, RDS-GO-B, and RDS-GO-C were evaluated by determining the reaction temperature required to obtain a sulfur content of 8 mass ppm. The required reaction temperatures for SRGO and the other mixed feedstocks of the cracked gas oil (MIX-D, -E, -F, -G, and -H) are shown in Table 1. MIX-D, which contained 15 vol% of LCO, was desulfurized to a sulfur content of 8 mass ppm at 638 K, which was 13 K higher than the temperature required for SRGO. This result indicates that LCO has a lower reactivity than SRGO, as was previously reported [6]. The densities of the other cracked gas fractions decreased in the order: VDS-GO > RDS-GO-A > RDSGO-B > RDS-GO-C, whereas the required reaction temperatures of the corresponding mixed feedstocks (MIX-E, -F, -G, and -H, respectively) did not show the same order. MIX-G, which consisted of 15 vol% of RDS-GO-B, required the lowest temperature, and MIXF, which contained 15 vol% of RDS-GO-A, required the highest temperature, equal to that of MIX-D (638 K). Other properties of these gas oil fractions may contribute to their reactivity for HDS. It is widely known that the properties of a feedstock affect the HDS activity of gas oil [15]. Gas oils contain various types of organic sulfur compounds, and their reactivities for HDS are varied.
457
For example, refractory sulfur compounds like 4,6-dimethyldibenzothiophene (4,6-DMDBT) are more difficult to remove than other types of sulfur-containing compounds, such as thiols, sulfides, and thiophenes [2]. Moreover, contaminants in feedstocks, including nitrogen and aromatic compounds, poison the active sites on the catalyst, resulting in a decrease in HDS activity [15,16]. To evaluate the effects of sulfur, nitrogen, and aromatic compounds on HDS activity, the reciprocal of the reaction temperature required to obtain a sulfur content of 8 mass ppm in the product oil, which represents high HDS activity at low values, was plotted against the content of 4,6-DMDBT, nitrogen, basic nitrogen, aromatic and polyaromatic compounds in the feedstock (Fig. 1(a)–(e), respectively. See the determination of the contents of these values in Table 1). Low correlation coefficients were obtained for 4,6-DMDBT (Fig. 1(a)) and the polyaromatics (Fig. 1(e)), indicating weak correlation [17]. This result suggests that the content of sulfur and polyaromatic compounds in the feedstock do not strongly correlate with the HDS reactivity. Conversely, the reaction temperatures for MIX-E, -F, -G, and -H were strongly correlated with the nitrogen, basic nitrogen, and aromatic content (Fig. 1(b)–(d), respectively). However, MIX-D did not correlate with MIX-E, -F, -G, and –H in these figures, indicating that LCO affects the HDS activity of gas oil by a different mechanism than the cracked gas oils from the other HDS units, i.e., ‘‘hydrotreated gas oils’’. Moreover, the effects of the nitrogen, basic nitrogen and aromatic compounds on HDS activity are presumably different for the hydrotreated gas oil fractions and LCO. That the hydrotreated gas oils and LCO contain different nitrogen and aromatic compounds is likely what accounts for these observed differences. 3.2. Distribution of hydrocarbons and nitrogen compounds As the differences in HDS activity were thought to be related to the nitrogen and aromatic compounds in the feedstocks, the distributions of the nitrogen and aromatic compounds in the SRGO, LCO, VDS-GO, and RDS-GO-A were analyzed by 2D-GC. The distributions of the hydrocarbon types in the LCO and RDS-GO-A are illustrated in Fig. 2(a) and (b), and the determined values are summarized in Table 2. SRGO contained a high ratio of saturated hydrocarbons to aromatic compounds (80 mol%:20 mol%). LCO contained a low ratio of saturated hydrocarbons relative to the aromatics (20 mol%:80 mol%), as reported in previous studies [18], and the majority of the aromatic compounds in LCO were polyaromatic compounds. Meanwhile, the ratios of saturated to aromatic compounds for both VDS-GO and RDS-GO-A were about 60 mol%:40 mol%, and most of the aromatic compounds were one-ring aromatics, such as alkylbenzenes, and one- and two-ring naphthenobenzenes. The distributions of nitrogen compounds in LCO and RDS-GO-A are illustrated in Fig. 3(a) and (b), and the quantitative values are summarized in Table 3. SRGO contained four types of nitrogen compounds: anilines, quinolines, indoles, and carbazoles. Meanwhile, LCO contained only indoles and carbazoles. VDS-GO and RDS-GO-A contained only anilines and carbazoles. Pyridines, quinolines, and anilines, which have a lone pair of electrons localized over their nitrogen atom, are categorized as basic nitrogen compounds, whereas indoles and carbazoles, which have hydrogens bound to the nitrogen atom, are classified as non-basic nitrogen compounds. The effect of the type of nitrogen compounds on the HDS activity of gas oil was compared using the slopes from Fig. 1(b) and (c). The slope for basic nitrogen was greater than that for total nitrogen, indicating that the basic nitrogen content of the hydrotreated gas oil has a greater influence on HDS activity. This result agrees well with the fact that nitrogen compounds are known to inhibit the active sites of an HDS catalyst. Pérot et al. claimed that nitrogen impurities may be detrimental to the acid
458
R. Koide et al. / Fuel 153 (2015) 455–463
Fig. 1. Relationship between the reaction temperature for a sulfur content of 8 mass ppm in the product oil and the (a) sulfur, (b) nitrogen, (c) basic nitrogen, (d) aromatics, and (e) polyaromatics content of the feed. The dotted line in each figure is the linear correlation among SRGO, MIX-E, -F, -G, and -H. The slope (S) and correlation coefficient (R) are shown in the figures.
components present in the catalyst [19]. This was discussed for HDS catalysts containing zeolites, which have stronger acidity than those with alumina as the support. While the role of the type of nitrogen compounds was not discussed in that paper, the strong acidity from zeolite may result in the strong interaction with any types of nitrogen compounds. Song et al. claimed that inhibiting effect of nitrogen compounds in some diesel blend stocks on deep HDS of gas oil [22]. The inhibiting effect was discussed for noble metal catalysts. These catalysts were strongly inhibited by nitrogen or sulfur impurities due to their sensitivity to these substances. Meanwhile, Furimsky et al. discussed conventional hydrodesulfurization catalysts based on alumina supports [15]. Since their acidity is relatively weak, the type of nitrogen compounds may significantly affect inhibiting the HDS activity. Thus, it was found that basic nitrogen compounds strongly inhibit the active sites of an HDS catalyst [20–22].
The distributions of basic nitrogen compounds in LCO, VDS-GO, and RDS-GO-A were significantly different (Table 3). LCO contained mainly indoles and carbazoles, classified as non-basic nitrogen compounds, whereas both VDS-GO and RDS-GO-A contained anilines as basic nitrogen compounds and carbazoles as non-basic nitrogen compounds. LCO contained more carbazoles than both VDS-GO and RDS-GO-A. Considering the effect of non-basic nitrogen compounds on HDS activity, Koltai suggested that non-basic polynuclear nitrogen compounds, such as carbazoles, could also inhibit the gas oil HDS reaction by enhanced basicity through hydrogenation of carbazoles [23]. If carbazoles strongly affected the HDS activity in addition to the inhibiting effect produced by a large amount of polyaromatic compounds, a greater decrease in HDS activity should be observed for the feed containing LCO than for that containing RDS-GO-A; however, the HDS reactivity of LCO was almost equivalent to that of RDS-GO-A. Therefore, anilines
R. Koide et al. / Fuel 153 (2015) 455–463
459
Fig. 2. 2D-GC/FID maps of gas oil fractions: (a) LCO and (b) RDS-GO-A.
as basic nitrogen compounds presumably influenced the decrease of HDS activity for MIX-F (Table 1).
3.3. Deactivation behavior To investigate the effect of nitrogen compounds on deactivation of the HDS catalysts, deactivation tests were carried out using MIXD and MIX-F as the feedstocks. Fig. 4 shows the deactivation behavior of the reaction temperature using MIX-D and MIX-F as feedstocks to obtain a sulfur content of 8 mass ppm in the product oil. After a 500 h reactivity test period, the reaction temperature was controlled to maintain a sulfur content of about 8 mass ppm in the product oil. The reaction temperature required to maintain a sulfur content of 8 mass ppm using MIX-D as the feedstock increased steadily until the 2300 h mark, and the deactivation rate was 0.12 K/day. Meanwhile, MIX-F, which showed an equivalent reaction temperature during the reactivity test period, showed a smaller increase in reaction temperature than did MIX-D during the deactivation test. The deactivation rate of the reaction temperature for MIX-F was 0.04 K/day. The difference in the deactivation rates for MIX-D and MIX-F is thought to be due to a difference
in the extent of coke deposition on the catalyst surface. Coke deposition on the catalyst surface is widely recognized as a major factor in the deactivation of HDS catalysts in gas oil HDS processes [18,24]. Several deactivation mechanisms for coke deposition have been proposed [25–29]. Richardson et al. proposed a model in which coke deposits form on the support, away from the active metal sulfides [30]. The coke is deposited and stacked, and as a result it may be difficult for sulfur compounds to access the active molybdenum sulfides.
3.4. Characteristics of deposits on the catalyst To investigate the characteristics of the deposits on used HDS catalysts, 13C NMR spectra have been commonly used [31,32]. This technique allows us to distinguish between carbon types in the deposit. Spectra were obtained for the deposits on the used catalysts for both the reactivity and deactivation tests. With MIXD, the amount of carbon in the used catalysts increased from the end of the reactivity test to the end of the deactivation test (6.0 mass% to 7.7 mass%). Fa, which is the ratio of aromatic carbons (determined by 13C NMR spectra) to the total carbon content of the
460
R. Koide et al. / Fuel 153 (2015) 455–463
Table 2 Types of hydrocarbons in SRGO, LCO, VDS-GO, and RDS-GO-A, as determined by 2D-GC/FID. Hydrocarbon type
Distribution (mol%) SRGO
a b c d
Saturated
Normal-paraffins Iso-paraffins 1-Ring naphthenes 2-Ring naphthenes
Aromatics
Alkylbenzenes 1-Ring naphthenobenzenesa 2-Ring naphthenobenzenesb Naphthalenes Biphenyls Acenaphthenes 3-Ring aromaticsc 4-Ring aromaticsd
VDS-GO
RDS-GO-A
31.4 29.7 13.6 2.4
LCO 6.5 9.5 3.7 0.9
22.6 21.6 13.8 2.3
22.4 20.6 13.2 2.8
8.8 4.8 0.1 4.9 2.2 1.0 1.0 0.0
7.2 8.5 0.1 39.9 10.5 4.6 7.9 0.7
16.3 11.2 0.2 5.4 3.9 1.5 1.0 0.1
15.2 11.7 5.2 3.0 3.6 1.5 0.8 0.0
Naphthenobenzene compounds containing one naphthene ring. Naphthenobenzene compounds containing two naphthene rings. polyaromatic compounds containing three benzene rings. polyaromatic compounds containing four benzene rings.
used catalyst, also increased (from 0.4 to 0.6). Thus, the amount of aromatic carbons increased from 2.4 mass% to 4.6 mass% vs. the catalyst weight. Meanwhile, MIX-F deposited a smaller amount of carbon on the catalyst than did MIX-D during the reactivity test (4.1%), whereas the Fa value was the same (0.4). After the reactivity test period, the carbon deposits on the spent catalyst did not increase much with MIX-F over the duration of the deactivation test (4.1–4.8%), and the Fa value stayed at a low level (0.3–0.4). Consequently, the amount of aromatic carbons on the used catalyst after the deactivation test was 1.4 mass%, almost equal to that after the reactivity test (1.6 mass%). This difference is presumably due to the different types of coke on the catalysts. Koizumi et al. reported that amorphous coke changed into graphite-like coke with increasing time on stream [11]. The increase in the weight of the carbon deposit and the aromaticity of the coke seen for LCO contained in MIX-D in this work is consistent with these previously reported results. With MIX-F, however, there was no increase in either the carbon content or the aromatic ratio, implying that the characteristic components of RDS-GO-A contributed to the initial
Fig. 3. 2D-GC/NCD maps of gas oil fractions: (a) LCO and (b) RDS-GO-A.
R. Koide et al. / Fuel 153 (2015) 455–463
461
Table 3 Nitrogen-containing compounds in SRGO, LCO, RDS-GO-A, and VDS-GO, as determined by 2D-GC/NCD. Nitrogen compound type
Anilines Quinolines Indoles Carbazoles Total
Nitrogen content (mass ppm) SRGO
LCO
VDS-GO
RDS-GO-A
33 10 12 55
7 0 97 366
43 0 0 207
171 0 0 179
110
470
250
350
formation of deposits on the catalyst, but did not cause further growth of the coke. The carbon structure of the deposits on the catalysts may also be related to the deactivation behavior. The carbon structures on the used catalysts following both the reactivity tests and the deactivation tests were examined using Raman spectroscopy (Fig. 5). Two bands were observed in the Raman spectra; the sharp band observed around 1580 cm 1 indicates graphitization and is referred to as the G-band [11], and the broad band at around 1360 cm 1 indicates the defect structure and is referred to as the D-band [12]. The ratio of G-band to D-band intensities is used to evaluate the regularity of the carbon structure. A low ratio is considered to represent a defective structure. After the reactivity test, this ratio was lower for the catalyst tested with MIX-F (1.70) than that tested with MIX-D (2.03). Furthermore, after the deactivation test, the ratio for the used catalyst tested with MIX-F (1.78) was also lower than that tested with MIX-D (1.92). Therefore, the graphite structure of the deposits on the catalyst from MIX-F is thought to be more anisotropic than that of the deposits from MIX-D. The anilines and other characteristic components of RDSGO-A may inhibit the aromatic hydrocarbons from stacking or the graphite structures from growing during gas oil HDS. The carbon structure of the deposit was also examined by TEMEELS. With TEM we can observe the stacked structure of the graphite layers [13]. TEM images and electron energy-loss spectra for four areas of the used catalysts after the deactivation tests with MIX-D and MIX-F are shown in Fig. 6(a) and (b), respectively. No graphitic layer structure was observed in the TEM images for either catalyst. However, clear energy losses at 285 eV, corresponding to the p⁄ anti-bonding state [31], were observed for the four areas in the spectra in Fig. 6(a), whereas the energy losses at 285 eV in the spectra in Fig. 6(b) were not significant and had lower signalto-noise ratios than the patterns in Fig. 6(a). This result indicates that the deposits on the used catalyst from MIX-E had a more amorphous-like structure [13] than those from MIX-D.
Fig. 4. Deactivation behaviors of reaction temperature for 8 massppm of sulfur in the product oil for MIX-D and MIX-F. The reaction conditions are: P = 5.5 MPaG, LHSV = 1.5 h 1, H2/Oil = 250 NL/L.
Fig. 5. Raman spectra of the used catalysts (top to bottom: MIX-D, deactivation test: MIX-D, reactivity test; MIX-F, deactivation test; MIX-F, reactivity test. Each measurement was repeated three times).
XRD measurements were carried out to clarify the stacked carbon structure of the deposits on the spent catalysts. Apparent diffraction patterns around 25° of 2h for the (0 0 2) plane of the graphite structure [33] were not observed. This result indicates that the carbonaceous deposits on the used catalysts had amorphous structures or few stacked layers. As the 13C NMR results indicated that the deposits contained considerable amounts of aromatic carbons, presumably the aromatic carbons were deposited irregularly. This assumption is consistent with the results of Raman spectroscopy and TEM-EELS.
3.5. Coke formation on the catalyst As we know that coke forms on the catalysts, the types of hydrocarbons in the feedstocks are important. LCO contained many two- and three-ring aromatic compounds (Table 3), and therefore accelerated the deposition of aromatic compounds on the catalyst. Meanwhile, RDS-GO-A contained monoaromatic hydrocarbons, such as one- and two-ring naphthenobenzenes, which correspond to partially hydrogenated two- and three-ring aromatic compounds, respectively. These compounds could be precursors of the coke formed on the catalyst during HDS, similar to the twoand three-ring aromatic compounds in LCO. However, MIX-F, which contained 15 vol% RDS-GO-A, did not accelerate coke deposition and aromatization of the deposits. This may be due to a difference in the adsorption geometries of the inhibiting compounds. Sun et al. reported that basic nitrogen compounds preferably adsorb on molybdenum sulfides as end-on configuration with the nitrogen atom whereas aromatic compounds do as side-on configuration through p electron of the aromatic compounds [22]. The basic nitrogen compounds that are a characteristic component in RDS-GO-A; these compounds, especially anilines, are
462
R. Koide et al. / Fuel 153 (2015) 455–463
C-K
1
1s→π π*
2
3 4 Mo-M
(a)MIX-D, Deacvaon test
(a)MIX-D, Deacvaon test
Fig. 6. TEM images (left) and electron energy-loss spectra (right) observed for the used catalysts after the deactivation test: (a) MIX-D and (b) MIX-F.
considered to possibly suppress growth of coke and increase of Fa value from aromatic compounds in the feedstock. If model feedstock of denitrogenated gas oil with basic nitrogen compounds, similar decrease of the HDS activity and similar slow deactivation behavior will be expected. The mechanism of the coke formation will be the subject of a future study.
results suggest that the characteristic components in RDS-GO helped the catalyst retain HDS activity, possibly by preventing the aromatic hydrocarbons in the feedstock from contributing to the growth of the coke.
4. Conclusions
This study was supported by the Japan Petroleum Energy Center (JPEC) under the sponsorship of the Ministry of Economy, Trade and Industry (METI) of Japan.
The required temperatures in the reactivity tests were nearly equal for the feedstocks with RDS-GO and LCO, whereas a lower deactivation rate in the deactivation test was observed for the feedstock mixed with RDS-GO. LCO contained polyaromatic hydrocarbons and non-basic nitrogen compounds, whereas RDS-GO contained monoaromatic hydrocarbons and basic nitrogen compounds. These differences in composition affected the properties of the coke that formed on the catalyst. The coke on the catalyst from the feedstock containing LCO had a higher aromatic ratio and a large G-band/D-band ratio than the coke from the feedstock containing RDS-GO. Furthermore, a clear energy loss at 285 eV, assigned to p⁄, was only observed for the used catalyst from the feedstock containing LCO. The deposits on the catalysts from the feedstock containing LCO was more aromatic than those from RDS-GO, which were expected to be more anisotropic. These
Acknowledgements
References [1] Japan Petroleum Energy Center. Research and Development of Advanced Hydrotreating Technology for Cracked Gas Oil Fraction. JPEC Activity Report; 1999:C2.1.3. [2] Choi KH, Mochida I, Sano Y, Nakano K. In: 16th Saudi–Japanese catalyst Symposium; 2004. [3] Choi K-H, Korai Y, Mochida I. Possible ways to achieve deep HDS of light cycle oil. Prepr Pap – Am Chem Soc, Div Fuel Chem 2003;48:653–4. [4] Palmer E, Polcar S, Wong A. Clean diesel hydrotreating. Pet Technol Quart 2009;Q1:91–100. [5] Brackett A. Study predicts viscosity of gas oils, heavy blends. Oil Gas J 2008;106(34). [6] Azizi N, Ali SA, Alhooshani K, Kim T, Lee Y, Park J-I, et al. Hydrotreating of light cycle oil over NiMo and CoMo catalysts with different supports. Fuel Process Technol 2013;109:172–8.
R. Koide et al. / Fuel 153 (2015) 455–463 [7] Ancheyta-Juárez J, Aguilar-Rodríguez E, Salazar-Sotelo D, Betancourt-Rivera G, Leiva-Nuncio M. Hydrotreating of straight run gas oil–light cycle oil blends. Appl Catal A 1999;180:195–205. [8] Koizumi N, Urabe Y, Hata K, Shingu M, Inamura K, Sugimoto Y, et al. Characterization of carbonaceous compounds deposited on NiMo catalyst used for ultra deep hydrodesulfurization of gas oil by means of temperatureprogrammed oxidation method. J Jpn Pet Inst 2005;48:204–13. [9] Koh JH, Lee JJ, Kim H, Cho A, Moon SH. Correlation of the deactivation of CoMo/ Al2O3 in hydrodesulfurization with surface carbon species. Appl Catal B 2009;86:176–81. [10] Callejas MA, Martínez MT, Blasco T, Sastre E. Coke characterisation in aged residue hydrotreating catalysts by solid-state 13CNMR spectroscopy and temperature-programmed oxidation. Appl Catal A 2001;218:181–8. [11] Koizumi N, Urabe N, Inamura K, Itoh T, Yamada M. Investigation of carbonaceous compounds deposited on NiMo catalyst used for ultra-deep hydrodesulfurization of gas oil by means of temperature-programmed oxidation and Raman spectroscopy. Catal Today 2005;106:211–8. [12] Li C, Stair PC. Ultraviolet Raman spectroscopy characterization of coke formation in zeolites. Catal Today 1997;33:353–60. [13] Berhault G, Mehta A, Pavel AC, Yang J, Rendon L, Yacaman MJ, et al. The role of structural carbon in transition metal sulfides hydrotreating catalysts. J Catal 2001;198:9–19. [14] van Donk S, de Groot FMF, Stéphan O, Bitter JH, de Jong KP. Monitoring the location, amount, and nature of carbonaceous deposits on aged zeolite ferrierite crystals by using STEM-EELS. Chem Eur J 2003;9:3106–11. [15] Furimsky E, Massoth FE. Deactivation of hydroprocessing catalysts. Catal Today 1999;52:381–495. [16] Fujikawa T. Deactivation and catalyst life prediction of ultra-deep HDS catalyst for diesel fractions. J Jpn Catal Soc 2010;52:27–32. [17] Zhang HG, Han XJ, Yao BF, Li GX. Study on the effect of engine operation parameters on cyclic combustion variations and correlation coefficient between the pressure-related parameters of a CNG engine. Appl Energy 2013;104:992–1002. [18] Calemma V, Giardino R, Ferrari M. Upgrading of LCO by partial hydrogenation of aromatics and ring opening of naphthenes over bi-functional catalysts. Fuel Process Technol 2010;91:770–6.
463
[19] Pérot G. Hydrotreating catalysts containing zeolites and related materialsmechanistic aspects related to deep desulfurization. Catal Today 2003;86:111–28. [20] Temel B, Tuxen AK, Kibsgaard J, Topsøe N-Y, Hinneman B, Knudsen KG, et al. Atomic-scale insight into the origin of pyridine inhibition of MoS2-based hydrotreating catalysts. J Catal 2010;271:280–9. [21] Moses PG, Hinnemann B, Topsøe H, Nørskov JK. The hydrogenation and direct desulfurization reaction pathway in thiophene hydrodesulfurization over MoS2 catalysts at realistic conditions: a density functional study. J Catal 2007;248:188–203. [22] Sun M, Nelson AE, Adjaye J. First principles study of heavy oil organonitrogen adsorption on NiMoS hydrotreating catalysts. Catal Today 2005;109:49–53. [23] Koltai T, Macauda M, Guevara A, Schulz E, Lemaire M, Bacaud R, et al. Comparative inhibiting effect of polycondensed aromatics and nitrogen compounds on the hydrodesulfurization of alkyldibenzothiophenes. Appl Catal A 2002;231:253–61. [24] Iizuka C. Idemitsu Gihou 2010;53:63–6. [25] Koide R, Fukase S, Al-Barood A, Al-Dolama K, Stanislaus A, Absi-Halabi M. The effect of nitrogen in feed on coke formation in hydrotreating. In: Delmon B, Froment GF, editors. Catalyst deactivation. Elsevier Science; 1999. p. 19–22. [26] Iizuka C, Inamura K, Koizumi N. Prep Jpn Petrol Inst 2010;53:29. [27] Inamura K, Koshika H, Hirano T, Koizumi N, Urabe Y, Yamada M. Prep Jpn Petrol Petrochem 2005;35:54. [28] Urabe Y, Koizumi N, Inamura K, Itoh K, Yamada M. Prep Jpn Petrol Petrochem 2005;35:55–6. [29] Koizumi N, Urabe Y, Inamura K, Itoh K, Yamada M. Prep Jpn Petrol Inst 2005;43:43. [30] Richardson SM, Nagaishi H, Gray MR. Initial coke deposition on a NiMo/cAl2O3 bitumen hydroprocessing catalyst. Ind Eng Chem Res 1996;35:3940–50. [31] Matsushita K, Hauser A, Marafi A, Koide R, Stanislaus A. Initial coke deposition on hydrotreating catalysts. Part 1. Changes in coke properties as a function of time on stream. Fuel 2004;83:1031–8. [32] Hauser A, Marafi A, Stanislaus A, Al-Adwani A. Relation between feed quality and coke formation in a three-stage atmospheric residue desulfurization (ARDS) process. Energy Fuel 2005;19:544–53. [33] Channu VS, Bobba R, Holze R. Graphite and graphene oxide electrodes for lithium ion batteries. Colloids Surf A 2013;436:245–61.
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Effect of basic nitrogen compounds on gas oil hydrodesulfurization and deposit formed on the catalyst Ryutaro Koide a,b,⇑, Yoshimu Iwanami a, Shozaburo Konishi a, Nobuharu Kimura a, Shinya Takahashi a, Masato Kamata a, Toshihide Baba b a
Central Technical Research Laboratory, JX Nippon Oil & Energy Corporation, 8 Chidoricho, Naka-ku, Yokohama 231-0815, Japan Department of Environmental Chemistry and Engineering, Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, 4259 Nagatsuta, Midori-ku, Yokohama 226-8502, Japan b
h i g h l i g h t s Residue desulfurization gas oil contained basic nitrogen compounds like anilines. Light cycle oil contained polyaromatic hydrocarbons, indoles and carbazoles. A low deactivation rate was observed for feed with residue desulfurization gas oil. The deposit from feed with light cycle oil was clarified as aromatic.
a r t i c l e
i n f o
Article history: Received 6 September 2014 Received in revised form 9 January 2015 Accepted 25 February 2015 Available online 16 March 2015 Keywords: Residue desulfurization gas oil Hydrodesulfurization Light cycle oil Coke Catalyst Basic nitrogen compounds
a b s t r a c t The authors studied the effect of the composition of cracked gas oils on the hydrodesulfurization to obtain clean diesel fuel. The required reaction temperature for sulfur specification of diesel fuel using a feedstock mixed with residue desulfurization gas oil (RDS-GO) was equivalent to that using light cycle oil (LCO). RDS-GO has a high content of basic nitrogen compounds, whereas LCO has a high content of aromatic hydrocarbons that are known to inhibit the hydrodesulfurization of gas oil. Using two-dimensional gas chromatography, LCO was found to contain polyaromatic hydrocarbons and non-basic nitrogen compounds, such as indoles and carbazoles, whereas RDS-GO contained monoaromatic hydrocarbons and basic nitrogen compounds, such as anilines. In a deactivation test conducted over 2300 h, the feedstock mixed with RDS-GO showed a lower deactivation rate than LCO. Raman spectroscopy, 13C NMR, and electron energy-loss spectroscopy in transmission electron microscopy showed that the deposits formed on the used catalysts during the deactivation test with the feedstock mixed with LCO contained more aromatic compounds than those with RDS-GO. The coke in the deposits from the feed containing RDS-GO was expected to be more anisotropic than that from the LCO. The improved hydrodesulfurization activity observed with RDS-GO was attributed to its characteristic components, which are thought to inhibit the growth of coke on the catalyst. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction Abbreviations: LCO, light cycle oil; FCC, fluid catalytic cracking; HDS, hydrodesulfurization; SRGO, straight run gas oil; RDS, residue hydrodesulfurization; GO, gas oil fraction; VDS, vacuum gas oil hydrogenation; TEM-EELS, electron energy-loss spectroscopy in transmission electron microscopy; 2D-GC, two-dimensional gas chromatography; XRD, X-ray diffraction; LHSV, liquid hourly space velocity; XRF, X-ray fluorescence; FID, flame ionization detector; NCD, nitrogen chemiluminescence detector ; HAADF, high-angle annular dark-field; 4,6-DMDBT, 4,6-dimethyldibenzothiophene. ⇑ Corresponding author at: Central Technical Research Laboratory, JX Nippon Oil & Energy Corporation, 8 Chidoricho, Naka-ku, Yokohama 231-0815, Japan. Tel.: +81 45 625 7301. E-mail address: [email protected] (R. Koide). http://dx.doi.org/10.1016/j.fuel.2015.02.112 0016-2361/Ó 2015 Elsevier Ltd. All rights reserved.
In recent years, refineries have had to utilize many intermediate fractions from various types of refining units [1]. ‘‘Cracked gas oils’’ are gas oil fractions obtained from hydrotreating or cracking units using heavy feedstocks. Light cycle oil (LCO) is a cracked gas oil that is obtained as a gas oil fraction from a fluid catalytic cracking (FCC) unit, and is widely used as a component of industrial heavy fuel oil or banker fuel oil because it has a lower viscosity and sulfur content than other potential components. The recent use of LCO as a feedstock for gas oil hydrodesulfurization (HDS) units has
456
R. Koide et al. / Fuel 153 (2015) 455–463
resulted in the upgrading of heavy industrial fuel oil into clean diesel fuel for transportation [2]. During gas oil HDS, LCO is hydrotreated by mixing it with a straight run gas oil (SRGO) fraction from crude oil [3]. However, LCO contains large amounts of aromatic hydrocarbons, which results in a highly exothermic reaction and high hydrogen consumption. Moreover, the HDS of feedstocks containing LCO results in rapid deactivation of the HDS catalyst caused by the accelerated deposition of coke on its surface [3,4]. This behavior limits the amount of LCO that can be added to a gas oil HDS unit. Meanwhile, other intermediate gas oil fractions from various units have mainly been used as components of industrial heavy fuel oil due to their having lower viscosities and lower sulfur contents than the other residual components in industrial fuel oil [5]. A residue hydrodesulfurization (RDS) unit produces a gas oil fraction (RDS-GO) as a by-product. RDS-GO is utilized as a diluting fraction in industrial heavy fuel oil, and is expected to become a feedstock for the gas oil HDS unit as well as LCO. A vacuum gas oil hydrogenation (VDS) unit also produces a gas oil fraction (VDS-GO) that is used as a diluting fraction. RDS-GO and VDS-GO are classified as ‘‘hydrotreated gas oils’’, and are considered to be good candidates for use as a feedstock for gas oil HDS units due to their having lower sulfur contents than SRGO. RDS-GO is particularly promising because it is produced in larger volumes in refineries than is VDS-GO. The use of RDS-GO in the gas oil HDS reaction and its deactivation behavior are not well understood. When using RDS-GO as a feedstock for gas oil HDS, it is important to know the composition of the gas oil feedstock and understand the effect the components will have on HDS activity and the deactivation of the HDS catalysts during long-term operation. It has been reported that the aromatic compounds in LCO can poison the catalyst during gas oil HDS [6,7], and that nitrogen compounds also affect the catalyst’s gas oil HDS activity [8]. Therefore, an investigation of the hydrocarbon and nitrogen-containing hydrocarbons in the feedstock is significant, as carbonaceous deposits on the catalyst are presumed to cause the deactivation of HDS activity in the gas oil HDS unit [9,10]. Several analytical methods have been used to characterize these carbonaceous compounds, including Raman spectroscopy [11,12], temperature programmed oxidation [12], and electron energy-loss spectroscopy in transmission electron microscopy (TEM-EELS) [13,14]. The purpose of this study is to investigate the composition of cracked gas oil and the effect of the feedstock composition on gas oil HDS. Two-dimensional gas chromatography (2D-GC) was used to clarify the types of hydrocarbons and nitrogen-containing compounds contained in the feedstock. Reactivity tests and deactivation tests were used to evaluate the effect of the feedstock composition on HDS activity and deactivation of gas oil. Furthermore, the deposits on the used catalysts were characterized using 13C NMR, Raman spectroscopy, TEM-EELS, and X-ray diffraction.
Table 1 Properties of the gas oil fractions used in the reactivity test and the deactivation test. Property
SRGO
LCO
VDSGO
RDSGO-A
RDSGO-B
RDSGO-C
Density (g/cm3) 90% boiling pointa (K) Sulfurb (mass ppm) 4,6-DMDBTc (mass ppm) Nitrogend (mass ppm) Basic nitrogene (mass ppm) Aromatic compoundsf (vol%)
0.8511 618.2 10,500 155 110 50
0.9493 624.7 4100 115 470 40
0.8769 662.7 870 105 250 100
0.8738 638.2 730 68 350 190
0.8609 604.2 180 80 83 56
0.8596 557.7 80 21 64 43
26
77
44
43
38
42
MIX-D 638.4
MIX-E 633
MIX-F 638.4
MIX-G MIX-H 627.4 629.6
SRGO Feed nameg 625.7 Required reaction temperature for sulfur content of 8 mass ppm in product oilh (K) a
Measured by ASTM D86. Determined by X-ray fluorescence (XRF). c 4,6-Dimethyldibenzothiophene determined by GC-sulfur chemiluminescence detector. d Measured by ASTM D3228. e Measured by UOP 269. f Measured by JPI-5S-49. g MIX-D, -E, -F, -G, and -H were prepared by mixing 85 vol% of SRGO and 15 vol% of LCO, VDS-GO, RDS-GO-A, RDS-GO-B, and RDS-GO-C, respectively. h The reaction conditions are: P = 5.5 MPaG, LHSV = 1.5 h 1, H2/Oil = 250 NL/L. b
2.2. Reactivity test The reactivity of the cracked gas oils was tested using a highpressure bench-scale fixed bed reactor. The feedstock properties are summarized in Table 1. A CoMo catalyst (100 mL; MoO3: 22.0 mass%, CoO: 3.0 mass%, Al2O3 support) for gas oil HDS was used for these tests after presulfiding with excess amount of sulfur reagent for sulfiding molybdenum on the catalyst with solvent. The following test conditions were used: hydrogen pressure: 5.5 MPaG, liquid hourly space velocity (LHSV): 1.0 h 1, ratio of hydrogen to oil (H2/Oil): 250 NL/L. The product oil from the reactivity test was bubbled with nitrogen gas at room temperature for 8 h to remove hydrogen sulfides. The sulfur content of the product oil was measured by X-ray fluorescence (XRF) using an Axios spectrometer (PANalytical). 2.3. Deactivation test The deactivation tests were carried out for MIX-D and MIX-F with the same catalyst and the test conditions as were used for the reactivity test after a period of about 500–2300 h. The reaction temperature was adjusted to obtain a sulfur content of nearly 8 mass ppm every 72 h. The normalized reaction temperature of hydrodesulfurization was calculated by the activation energy of 113 kJ/molK, which was calculated by the reactivity test.
2. Experimental
2.4. Two-dimensional gas chromatography
2.1. Feedstock
The components of SRGO, LCO, RDS-GO-A, and VDS-GO were determined using 2D-GC (KT2006, ZEOX) with two columns: BPX-5 (30 m 0.25 mm, i.d. 0.25 lm) and BPX-50 (2 m 0.1 mm, i.d. 0.1 lm) and a flame ionization detector (FID) or nitrogen chemiluminescence detector (NCD). Helium gas was used as the carrier gas at a flow rate of 2.58 mL/min.
SRGO was obtained by distillation of a typical Arabian crude. LCO was obtained from a FCC unit of hydrodesulfurized residue. RDS-GO-A, -B, and -C were obtained from three different RDS units for hydrodesulfurization of atmospheric residue. VDS-GO was obtained from a vacuum gas oil HDS unit for hydrodesulfurization of vacuum gas oil. The properties of the gas oils are summarized in Table 1. For the reactivity and deactivation tests, MIX-D, -E, -F, -G, and -H were prepared by mixing 85 vol% of SRGO and 15 vol% of LCO, VDS-GO, RDS-GO-A, RDS-GO-B, and RDS-GO-C, respectively.
2.5. Used catalysts from the reactivity tests and deactivation tests After the reactivity tests and deactivation tests, residual gas oil in the used catalysts was removed by Soxhlet extraction with
R. Koide et al. / Fuel 153 (2015) 455–463
toluene until the extracted liquid turned clear. The treated catalysts were dried at room temperature for 12 h and then evacuated at 353 K for 2 h. 2.6.
13
C NMR spectra
The 13C NMR spectra of the used catalysts were measured with an NMR System 500 (Varian) equipped with a CP/MAS probe with a resonance frequency of 125 MHz. Each sample was measured using a pulse at 47.6° for 4 ms with a pulse interval of 10 s for 3900 scans. 2.7. Raman spectra The Raman spectra of the deposits on the used catalysts were measured with an NRS-5100 (JASCO) equipped with an Ar+ laser (532 nm, 0.1 MW) as the excitation source. The diffraction slit size was 25 lm 1000 lm. 2.8. Electron energy-loss spectroscopy in transmission electron microscopy TEM-EELS images and spectra of the used catalysts were measured using a JEM-ARM200F (JEOL) operated at 200 kV. High-angle annular dark-field (HAADF) images were obtained by performing Fourier transformations. 2.9. X-ray diffraction XRD spectra of the used catalysts were measured with a RINT2500 (RIGAKU) equipped with a Cu Ka source using a scanning rate of 1°/min. 3. Results and discussion 3.1. Effect of sulfur, nitrogen, basic nitrogen and aromatic compounds on reaction temperature Gas oil is desulfurized in HDS units at refineries to satisfy the diesel specifications in Europe, North America and Japan, which require that sulfur content be no higher than 10 mass ppm. To meet this requirement, most refineries control the reaction temperature of the gas oil HDS units to obtain a sulfur content of less than 10 mass ppm; there is often around 8 mass ppm of sulfur in the product oil. Therefore, the reactivities of SRGO, LCO, VDS-GO, RDS-GO-A, RDS-GO-B, and RDS-GO-C were evaluated by determining the reaction temperature required to obtain a sulfur content of 8 mass ppm. The required reaction temperatures for SRGO and the other mixed feedstocks of the cracked gas oil (MIX-D, -E, -F, -G, and -H) are shown in Table 1. MIX-D, which contained 15 vol% of LCO, was desulfurized to a sulfur content of 8 mass ppm at 638 K, which was 13 K higher than the temperature required for SRGO. This result indicates that LCO has a lower reactivity than SRGO, as was previously reported [6]. The densities of the other cracked gas fractions decreased in the order: VDS-GO > RDS-GO-A > RDSGO-B > RDS-GO-C, whereas the required reaction temperatures of the corresponding mixed feedstocks (MIX-E, -F, -G, and -H, respectively) did not show the same order. MIX-G, which consisted of 15 vol% of RDS-GO-B, required the lowest temperature, and MIXF, which contained 15 vol% of RDS-GO-A, required the highest temperature, equal to that of MIX-D (638 K). Other properties of these gas oil fractions may contribute to their reactivity for HDS. It is widely known that the properties of a feedstock affect the HDS activity of gas oil [15]. Gas oils contain various types of organic sulfur compounds, and their reactivities for HDS are varied.
457
For example, refractory sulfur compounds like 4,6-dimethyldibenzothiophene (4,6-DMDBT) are more difficult to remove than other types of sulfur-containing compounds, such as thiols, sulfides, and thiophenes [2]. Moreover, contaminants in feedstocks, including nitrogen and aromatic compounds, poison the active sites on the catalyst, resulting in a decrease in HDS activity [15,16]. To evaluate the effects of sulfur, nitrogen, and aromatic compounds on HDS activity, the reciprocal of the reaction temperature required to obtain a sulfur content of 8 mass ppm in the product oil, which represents high HDS activity at low values, was plotted against the content of 4,6-DMDBT, nitrogen, basic nitrogen, aromatic and polyaromatic compounds in the feedstock (Fig. 1(a)–(e), respectively. See the determination of the contents of these values in Table 1). Low correlation coefficients were obtained for 4,6-DMDBT (Fig. 1(a)) and the polyaromatics (Fig. 1(e)), indicating weak correlation [17]. This result suggests that the content of sulfur and polyaromatic compounds in the feedstock do not strongly correlate with the HDS reactivity. Conversely, the reaction temperatures for MIX-E, -F, -G, and -H were strongly correlated with the nitrogen, basic nitrogen, and aromatic content (Fig. 1(b)–(d), respectively). However, MIX-D did not correlate with MIX-E, -F, -G, and –H in these figures, indicating that LCO affects the HDS activity of gas oil by a different mechanism than the cracked gas oils from the other HDS units, i.e., ‘‘hydrotreated gas oils’’. Moreover, the effects of the nitrogen, basic nitrogen and aromatic compounds on HDS activity are presumably different for the hydrotreated gas oil fractions and LCO. That the hydrotreated gas oils and LCO contain different nitrogen and aromatic compounds is likely what accounts for these observed differences. 3.2. Distribution of hydrocarbons and nitrogen compounds As the differences in HDS activity were thought to be related to the nitrogen and aromatic compounds in the feedstocks, the distributions of the nitrogen and aromatic compounds in the SRGO, LCO, VDS-GO, and RDS-GO-A were analyzed by 2D-GC. The distributions of the hydrocarbon types in the LCO and RDS-GO-A are illustrated in Fig. 2(a) and (b), and the determined values are summarized in Table 2. SRGO contained a high ratio of saturated hydrocarbons to aromatic compounds (80 mol%:20 mol%). LCO contained a low ratio of saturated hydrocarbons relative to the aromatics (20 mol%:80 mol%), as reported in previous studies [18], and the majority of the aromatic compounds in LCO were polyaromatic compounds. Meanwhile, the ratios of saturated to aromatic compounds for both VDS-GO and RDS-GO-A were about 60 mol%:40 mol%, and most of the aromatic compounds were one-ring aromatics, such as alkylbenzenes, and one- and two-ring naphthenobenzenes. The distributions of nitrogen compounds in LCO and RDS-GO-A are illustrated in Fig. 3(a) and (b), and the quantitative values are summarized in Table 3. SRGO contained four types of nitrogen compounds: anilines, quinolines, indoles, and carbazoles. Meanwhile, LCO contained only indoles and carbazoles. VDS-GO and RDS-GO-A contained only anilines and carbazoles. Pyridines, quinolines, and anilines, which have a lone pair of electrons localized over their nitrogen atom, are categorized as basic nitrogen compounds, whereas indoles and carbazoles, which have hydrogens bound to the nitrogen atom, are classified as non-basic nitrogen compounds. The effect of the type of nitrogen compounds on the HDS activity of gas oil was compared using the slopes from Fig. 1(b) and (c). The slope for basic nitrogen was greater than that for total nitrogen, indicating that the basic nitrogen content of the hydrotreated gas oil has a greater influence on HDS activity. This result agrees well with the fact that nitrogen compounds are known to inhibit the active sites of an HDS catalyst. Pérot et al. claimed that nitrogen impurities may be detrimental to the acid
458
R. Koide et al. / Fuel 153 (2015) 455–463
Fig. 1. Relationship between the reaction temperature for a sulfur content of 8 mass ppm in the product oil and the (a) sulfur, (b) nitrogen, (c) basic nitrogen, (d) aromatics, and (e) polyaromatics content of the feed. The dotted line in each figure is the linear correlation among SRGO, MIX-E, -F, -G, and -H. The slope (S) and correlation coefficient (R) are shown in the figures.
components present in the catalyst [19]. This was discussed for HDS catalysts containing zeolites, which have stronger acidity than those with alumina as the support. While the role of the type of nitrogen compounds was not discussed in that paper, the strong acidity from zeolite may result in the strong interaction with any types of nitrogen compounds. Song et al. claimed that inhibiting effect of nitrogen compounds in some diesel blend stocks on deep HDS of gas oil [22]. The inhibiting effect was discussed for noble metal catalysts. These catalysts were strongly inhibited by nitrogen or sulfur impurities due to their sensitivity to these substances. Meanwhile, Furimsky et al. discussed conventional hydrodesulfurization catalysts based on alumina supports [15]. Since their acidity is relatively weak, the type of nitrogen compounds may significantly affect inhibiting the HDS activity. Thus, it was found that basic nitrogen compounds strongly inhibit the active sites of an HDS catalyst [20–22].
The distributions of basic nitrogen compounds in LCO, VDS-GO, and RDS-GO-A were significantly different (Table 3). LCO contained mainly indoles and carbazoles, classified as non-basic nitrogen compounds, whereas both VDS-GO and RDS-GO-A contained anilines as basic nitrogen compounds and carbazoles as non-basic nitrogen compounds. LCO contained more carbazoles than both VDS-GO and RDS-GO-A. Considering the effect of non-basic nitrogen compounds on HDS activity, Koltai suggested that non-basic polynuclear nitrogen compounds, such as carbazoles, could also inhibit the gas oil HDS reaction by enhanced basicity through hydrogenation of carbazoles [23]. If carbazoles strongly affected the HDS activity in addition to the inhibiting effect produced by a large amount of polyaromatic compounds, a greater decrease in HDS activity should be observed for the feed containing LCO than for that containing RDS-GO-A; however, the HDS reactivity of LCO was almost equivalent to that of RDS-GO-A. Therefore, anilines
R. Koide et al. / Fuel 153 (2015) 455–463
459
Fig. 2. 2D-GC/FID maps of gas oil fractions: (a) LCO and (b) RDS-GO-A.
as basic nitrogen compounds presumably influenced the decrease of HDS activity for MIX-F (Table 1).
3.3. Deactivation behavior To investigate the effect of nitrogen compounds on deactivation of the HDS catalysts, deactivation tests were carried out using MIXD and MIX-F as the feedstocks. Fig. 4 shows the deactivation behavior of the reaction temperature using MIX-D and MIX-F as feedstocks to obtain a sulfur content of 8 mass ppm in the product oil. After a 500 h reactivity test period, the reaction temperature was controlled to maintain a sulfur content of about 8 mass ppm in the product oil. The reaction temperature required to maintain a sulfur content of 8 mass ppm using MIX-D as the feedstock increased steadily until the 2300 h mark, and the deactivation rate was 0.12 K/day. Meanwhile, MIX-F, which showed an equivalent reaction temperature during the reactivity test period, showed a smaller increase in reaction temperature than did MIX-D during the deactivation test. The deactivation rate of the reaction temperature for MIX-F was 0.04 K/day. The difference in the deactivation rates for MIX-D and MIX-F is thought to be due to a difference
in the extent of coke deposition on the catalyst surface. Coke deposition on the catalyst surface is widely recognized as a major factor in the deactivation of HDS catalysts in gas oil HDS processes [18,24]. Several deactivation mechanisms for coke deposition have been proposed [25–29]. Richardson et al. proposed a model in which coke deposits form on the support, away from the active metal sulfides [30]. The coke is deposited and stacked, and as a result it may be difficult for sulfur compounds to access the active molybdenum sulfides.
3.4. Characteristics of deposits on the catalyst To investigate the characteristics of the deposits on used HDS catalysts, 13C NMR spectra have been commonly used [31,32]. This technique allows us to distinguish between carbon types in the deposit. Spectra were obtained for the deposits on the used catalysts for both the reactivity and deactivation tests. With MIXD, the amount of carbon in the used catalysts increased from the end of the reactivity test to the end of the deactivation test (6.0 mass% to 7.7 mass%). Fa, which is the ratio of aromatic carbons (determined by 13C NMR spectra) to the total carbon content of the
460
R. Koide et al. / Fuel 153 (2015) 455–463
Table 2 Types of hydrocarbons in SRGO, LCO, VDS-GO, and RDS-GO-A, as determined by 2D-GC/FID. Hydrocarbon type
Distribution (mol%) SRGO
a b c d
Saturated
Normal-paraffins Iso-paraffins 1-Ring naphthenes 2-Ring naphthenes
Aromatics
Alkylbenzenes 1-Ring naphthenobenzenesa 2-Ring naphthenobenzenesb Naphthalenes Biphenyls Acenaphthenes 3-Ring aromaticsc 4-Ring aromaticsd
VDS-GO
RDS-GO-A
31.4 29.7 13.6 2.4
LCO 6.5 9.5 3.7 0.9
22.6 21.6 13.8 2.3
22.4 20.6 13.2 2.8
8.8 4.8 0.1 4.9 2.2 1.0 1.0 0.0
7.2 8.5 0.1 39.9 10.5 4.6 7.9 0.7
16.3 11.2 0.2 5.4 3.9 1.5 1.0 0.1
15.2 11.7 5.2 3.0 3.6 1.5 0.8 0.0
Naphthenobenzene compounds containing one naphthene ring. Naphthenobenzene compounds containing two naphthene rings. polyaromatic compounds containing three benzene rings. polyaromatic compounds containing four benzene rings.
used catalyst, also increased (from 0.4 to 0.6). Thus, the amount of aromatic carbons increased from 2.4 mass% to 4.6 mass% vs. the catalyst weight. Meanwhile, MIX-F deposited a smaller amount of carbon on the catalyst than did MIX-D during the reactivity test (4.1%), whereas the Fa value was the same (0.4). After the reactivity test period, the carbon deposits on the spent catalyst did not increase much with MIX-F over the duration of the deactivation test (4.1–4.8%), and the Fa value stayed at a low level (0.3–0.4). Consequently, the amount of aromatic carbons on the used catalyst after the deactivation test was 1.4 mass%, almost equal to that after the reactivity test (1.6 mass%). This difference is presumably due to the different types of coke on the catalysts. Koizumi et al. reported that amorphous coke changed into graphite-like coke with increasing time on stream [11]. The increase in the weight of the carbon deposit and the aromaticity of the coke seen for LCO contained in MIX-D in this work is consistent with these previously reported results. With MIX-F, however, there was no increase in either the carbon content or the aromatic ratio, implying that the characteristic components of RDS-GO-A contributed to the initial
Fig. 3. 2D-GC/NCD maps of gas oil fractions: (a) LCO and (b) RDS-GO-A.
R. Koide et al. / Fuel 153 (2015) 455–463
461
Table 3 Nitrogen-containing compounds in SRGO, LCO, RDS-GO-A, and VDS-GO, as determined by 2D-GC/NCD. Nitrogen compound type
Anilines Quinolines Indoles Carbazoles Total
Nitrogen content (mass ppm) SRGO
LCO
VDS-GO
RDS-GO-A
33 10 12 55
7 0 97 366
43 0 0 207
171 0 0 179
110
470
250
350
formation of deposits on the catalyst, but did not cause further growth of the coke. The carbon structure of the deposits on the catalysts may also be related to the deactivation behavior. The carbon structures on the used catalysts following both the reactivity tests and the deactivation tests were examined using Raman spectroscopy (Fig. 5). Two bands were observed in the Raman spectra; the sharp band observed around 1580 cm 1 indicates graphitization and is referred to as the G-band [11], and the broad band at around 1360 cm 1 indicates the defect structure and is referred to as the D-band [12]. The ratio of G-band to D-band intensities is used to evaluate the regularity of the carbon structure. A low ratio is considered to represent a defective structure. After the reactivity test, this ratio was lower for the catalyst tested with MIX-F (1.70) than that tested with MIX-D (2.03). Furthermore, after the deactivation test, the ratio for the used catalyst tested with MIX-F (1.78) was also lower than that tested with MIX-D (1.92). Therefore, the graphite structure of the deposits on the catalyst from MIX-F is thought to be more anisotropic than that of the deposits from MIX-D. The anilines and other characteristic components of RDSGO-A may inhibit the aromatic hydrocarbons from stacking or the graphite structures from growing during gas oil HDS. The carbon structure of the deposit was also examined by TEMEELS. With TEM we can observe the stacked structure of the graphite layers [13]. TEM images and electron energy-loss spectra for four areas of the used catalysts after the deactivation tests with MIX-D and MIX-F are shown in Fig. 6(a) and (b), respectively. No graphitic layer structure was observed in the TEM images for either catalyst. However, clear energy losses at 285 eV, corresponding to the p⁄ anti-bonding state [31], were observed for the four areas in the spectra in Fig. 6(a), whereas the energy losses at 285 eV in the spectra in Fig. 6(b) were not significant and had lower signalto-noise ratios than the patterns in Fig. 6(a). This result indicates that the deposits on the used catalyst from MIX-E had a more amorphous-like structure [13] than those from MIX-D.
Fig. 4. Deactivation behaviors of reaction temperature for 8 massppm of sulfur in the product oil for MIX-D and MIX-F. The reaction conditions are: P = 5.5 MPaG, LHSV = 1.5 h 1, H2/Oil = 250 NL/L.
Fig. 5. Raman spectra of the used catalysts (top to bottom: MIX-D, deactivation test: MIX-D, reactivity test; MIX-F, deactivation test; MIX-F, reactivity test. Each measurement was repeated three times).
XRD measurements were carried out to clarify the stacked carbon structure of the deposits on the spent catalysts. Apparent diffraction patterns around 25° of 2h for the (0 0 2) plane of the graphite structure [33] were not observed. This result indicates that the carbonaceous deposits on the used catalysts had amorphous structures or few stacked layers. As the 13C NMR results indicated that the deposits contained considerable amounts of aromatic carbons, presumably the aromatic carbons were deposited irregularly. This assumption is consistent with the results of Raman spectroscopy and TEM-EELS.
3.5. Coke formation on the catalyst As we know that coke forms on the catalysts, the types of hydrocarbons in the feedstocks are important. LCO contained many two- and three-ring aromatic compounds (Table 3), and therefore accelerated the deposition of aromatic compounds on the catalyst. Meanwhile, RDS-GO-A contained monoaromatic hydrocarbons, such as one- and two-ring naphthenobenzenes, which correspond to partially hydrogenated two- and three-ring aromatic compounds, respectively. These compounds could be precursors of the coke formed on the catalyst during HDS, similar to the twoand three-ring aromatic compounds in LCO. However, MIX-F, which contained 15 vol% RDS-GO-A, did not accelerate coke deposition and aromatization of the deposits. This may be due to a difference in the adsorption geometries of the inhibiting compounds. Sun et al. reported that basic nitrogen compounds preferably adsorb on molybdenum sulfides as end-on configuration with the nitrogen atom whereas aromatic compounds do as side-on configuration through p electron of the aromatic compounds [22]. The basic nitrogen compounds that are a characteristic component in RDS-GO-A; these compounds, especially anilines, are
462
R. Koide et al. / Fuel 153 (2015) 455–463
C-K
1
1s→π π*
2
3 4 Mo-M
(a)MIX-D, Deacvaon test
(a)MIX-D, Deacvaon test
Fig. 6. TEM images (left) and electron energy-loss spectra (right) observed for the used catalysts after the deactivation test: (a) MIX-D and (b) MIX-F.
considered to possibly suppress growth of coke and increase of Fa value from aromatic compounds in the feedstock. If model feedstock of denitrogenated gas oil with basic nitrogen compounds, similar decrease of the HDS activity and similar slow deactivation behavior will be expected. The mechanism of the coke formation will be the subject of a future study.
results suggest that the characteristic components in RDS-GO helped the catalyst retain HDS activity, possibly by preventing the aromatic hydrocarbons in the feedstock from contributing to the growth of the coke.
4. Conclusions
This study was supported by the Japan Petroleum Energy Center (JPEC) under the sponsorship of the Ministry of Economy, Trade and Industry (METI) of Japan.
The required temperatures in the reactivity tests were nearly equal for the feedstocks with RDS-GO and LCO, whereas a lower deactivation rate in the deactivation test was observed for the feedstock mixed with RDS-GO. LCO contained polyaromatic hydrocarbons and non-basic nitrogen compounds, whereas RDS-GO contained monoaromatic hydrocarbons and basic nitrogen compounds. These differences in composition affected the properties of the coke that formed on the catalyst. The coke on the catalyst from the feedstock containing LCO had a higher aromatic ratio and a large G-band/D-band ratio than the coke from the feedstock containing RDS-GO. Furthermore, a clear energy loss at 285 eV, assigned to p⁄, was only observed for the used catalyst from the feedstock containing LCO. The deposits on the catalysts from the feedstock containing LCO was more aromatic than those from RDS-GO, which were expected to be more anisotropic. These
Acknowledgements
References [1] Japan Petroleum Energy Center. Research and Development of Advanced Hydrotreating Technology for Cracked Gas Oil Fraction. JPEC Activity Report; 1999:C2.1.3. [2] Choi KH, Mochida I, Sano Y, Nakano K. In: 16th Saudi–Japanese catalyst Symposium; 2004. [3] Choi K-H, Korai Y, Mochida I. Possible ways to achieve deep HDS of light cycle oil. Prepr Pap – Am Chem Soc, Div Fuel Chem 2003;48:653–4. [4] Palmer E, Polcar S, Wong A. Clean diesel hydrotreating. Pet Technol Quart 2009;Q1:91–100. [5] Brackett A. Study predicts viscosity of gas oils, heavy blends. Oil Gas J 2008;106(34). [6] Azizi N, Ali SA, Alhooshani K, Kim T, Lee Y, Park J-I, et al. Hydrotreating of light cycle oil over NiMo and CoMo catalysts with different supports. Fuel Process Technol 2013;109:172–8.
R. Koide et al. / Fuel 153 (2015) 455–463 [7] Ancheyta-Juárez J, Aguilar-Rodríguez E, Salazar-Sotelo D, Betancourt-Rivera G, Leiva-Nuncio M. Hydrotreating of straight run gas oil–light cycle oil blends. Appl Catal A 1999;180:195–205. [8] Koizumi N, Urabe Y, Hata K, Shingu M, Inamura K, Sugimoto Y, et al. Characterization of carbonaceous compounds deposited on NiMo catalyst used for ultra deep hydrodesulfurization of gas oil by means of temperatureprogrammed oxidation method. J Jpn Pet Inst 2005;48:204–13. [9] Koh JH, Lee JJ, Kim H, Cho A, Moon SH. Correlation of the deactivation of CoMo/ Al2O3 in hydrodesulfurization with surface carbon species. Appl Catal B 2009;86:176–81. [10] Callejas MA, Martínez MT, Blasco T, Sastre E. Coke characterisation in aged residue hydrotreating catalysts by solid-state 13CNMR spectroscopy and temperature-programmed oxidation. Appl Catal A 2001;218:181–8. [11] Koizumi N, Urabe N, Inamura K, Itoh T, Yamada M. Investigation of carbonaceous compounds deposited on NiMo catalyst used for ultra-deep hydrodesulfurization of gas oil by means of temperature-programmed oxidation and Raman spectroscopy. Catal Today 2005;106:211–8. [12] Li C, Stair PC. Ultraviolet Raman spectroscopy characterization of coke formation in zeolites. Catal Today 1997;33:353–60. [13] Berhault G, Mehta A, Pavel AC, Yang J, Rendon L, Yacaman MJ, et al. The role of structural carbon in transition metal sulfides hydrotreating catalysts. J Catal 2001;198:9–19. [14] van Donk S, de Groot FMF, Stéphan O, Bitter JH, de Jong KP. Monitoring the location, amount, and nature of carbonaceous deposits on aged zeolite ferrierite crystals by using STEM-EELS. Chem Eur J 2003;9:3106–11. [15] Furimsky E, Massoth FE. Deactivation of hydroprocessing catalysts. Catal Today 1999;52:381–495. [16] Fujikawa T. Deactivation and catalyst life prediction of ultra-deep HDS catalyst for diesel fractions. J Jpn Catal Soc 2010;52:27–32. [17] Zhang HG, Han XJ, Yao BF, Li GX. Study on the effect of engine operation parameters on cyclic combustion variations and correlation coefficient between the pressure-related parameters of a CNG engine. Appl Energy 2013;104:992–1002. [18] Calemma V, Giardino R, Ferrari M. Upgrading of LCO by partial hydrogenation of aromatics and ring opening of naphthenes over bi-functional catalysts. Fuel Process Technol 2010;91:770–6.
463
[19] Pérot G. Hydrotreating catalysts containing zeolites and related materialsmechanistic aspects related to deep desulfurization. Catal Today 2003;86:111–28. [20] Temel B, Tuxen AK, Kibsgaard J, Topsøe N-Y, Hinneman B, Knudsen KG, et al. Atomic-scale insight into the origin of pyridine inhibition of MoS2-based hydrotreating catalysts. J Catal 2010;271:280–9. [21] Moses PG, Hinnemann B, Topsøe H, Nørskov JK. The hydrogenation and direct desulfurization reaction pathway in thiophene hydrodesulfurization over MoS2 catalysts at realistic conditions: a density functional study. J Catal 2007;248:188–203. [22] Sun M, Nelson AE, Adjaye J. First principles study of heavy oil organonitrogen adsorption on NiMoS hydrotreating catalysts. Catal Today 2005;109:49–53. [23] Koltai T, Macauda M, Guevara A, Schulz E, Lemaire M, Bacaud R, et al. Comparative inhibiting effect of polycondensed aromatics and nitrogen compounds on the hydrodesulfurization of alkyldibenzothiophenes. Appl Catal A 2002;231:253–61. [24] Iizuka C. Idemitsu Gihou 2010;53:63–6. [25] Koide R, Fukase S, Al-Barood A, Al-Dolama K, Stanislaus A, Absi-Halabi M. The effect of nitrogen in feed on coke formation in hydrotreating. In: Delmon B, Froment GF, editors. Catalyst deactivation. Elsevier Science; 1999. p. 19–22. [26] Iizuka C, Inamura K, Koizumi N. Prep Jpn Petrol Inst 2010;53:29. [27] Inamura K, Koshika H, Hirano T, Koizumi N, Urabe Y, Yamada M. Prep Jpn Petrol Petrochem 2005;35:54. [28] Urabe Y, Koizumi N, Inamura K, Itoh K, Yamada M. Prep Jpn Petrol Petrochem 2005;35:55–6. [29] Koizumi N, Urabe Y, Inamura K, Itoh K, Yamada M. Prep Jpn Petrol Inst 2005;43:43. [30] Richardson SM, Nagaishi H, Gray MR. Initial coke deposition on a NiMo/cAl2O3 bitumen hydroprocessing catalyst. Ind Eng Chem Res 1996;35:3940–50. [31] Matsushita K, Hauser A, Marafi A, Koide R, Stanislaus A. Initial coke deposition on hydrotreating catalysts. Part 1. Changes in coke properties as a function of time on stream. Fuel 2004;83:1031–8. [32] Hauser A, Marafi A, Stanislaus A, Al-Adwani A. Relation between feed quality and coke formation in a three-stage atmospheric residue desulfurization (ARDS) process. Energy Fuel 2005;19:544–53. [33] Channu VS, Bobba R, Holze R. Graphite and graphene oxide electrodes for lithium ion batteries. Colloids Surf A 2013;436:245–61.