Al2O3 catalysts by nitrogen compounds during the hydrodesulfurization of dibenzothiophene compounds

Applied Catalysis B: Environmental 50 (2004) 17–24

Effect of fluorine addition on the poisoning of NiMo/Al2 O3 catalysts by nitrogen compounds during the hydrodesulfurization of dibenzothiophene compounds Heeyeon Kim, Jung Joon Lee, Jae Hyun Koh, Sang Heup Moon∗ School of Chemical Engineering, Institute of Chemical Processes, Seoul National University, San 56-1, Shillim-dong, Kwanak-ku, Seoul 151-744, South Korea Received 8 August 2003; received in revised form 17 October 2003; accepted 29 October 2003

Abstract The effect of fluorine added to NiMo/Al2 O3 catalysts on their poisoning by basic (quinoline) and non-basic (carbazole) nitrogen compounds during the hydrodesulfurization (HDS) of dibenzothiophene (DBT) and 4,6-dimethyldibenzothiophene (4,6-DMDBT) was investigated. Fluorinated NiMo catalysts had a higher activity than fluorine-free catalysts, and this superior activity of fluorinated catalysts was maintained even after poisoning by nitrogen compounds, as confirmed by NO chemisorption and FTIR spectroscopy. The HDS rate was retarded to a greater extent by quinoline than by carbazole. In the HDS of DBT, the difference between the activities of the two types of catalysts remained constant even when the poisoning was extensive. This is in contrast to the case of 4,6-DMDBT HDS, in which the difference in activity decreased when the catalysts were extensively poisoned. The product distribution changed with poisoning showing a characteristic trend that was dependant on the combination of reactants and nitrogen compounds. In DBT HDS, the hydrogenation (HYD) route was poisoned to a more significant extent than the direct desulfurization (DDS) route by both quinoline and carbazole, and was independent of catalyst fluorination. In the HDS of 4,6-DMDBT, quinoline retarded HYD more than DDS but carbazole retarded DDS more than HYD. © 2003 Published by Elsevier B.V. Keywords: Hydrodesulfurization; Fluorine; NiMo; Dibenzothiophene; 4,6-Dimethyldibenzothiophene; Quinoline; Carbazole

1. Introduction With air pollution regulations becoming increasingly stringent in recent years, interest in the deep hydrodesulfurization (HDS) of diesel fuel has increased. Numerous attempts have been made to improve the catalyst activity by modifying the support [1–3] or by using various additives [4–9]. However, much less information is available concerning the effect of nitrogen impurities contained in the fuel, which frequently cause severe poisoning of HDS catalysts even in the presence of small amounts [10]. Basic nitrogen compounds are known to strongly inhibit the rate of HDS even at concentrations as low as 30 ppm [10]. In general, gas oil contains 0.01–0.05 wt.% of nitrogen impurities, among which non-basic compounds, such as in-

∗

Corresponding author. Tel.: +82-2-880-7409; fax: +82-2-875-6697. E-mail address: [email protected] (S.H. Moon).

0926-3373/$ – see front matter © 2003 Published by Elsevier B.V. doi:10.1016/j.apcatb.2003.10.012

dole and carbazole, account for 70% and the remaining 30% are basic compounds, such as quinoline [11]. To date, most studies on the inhibiting effect of nitrogen impurities on HDS were made using relatively simple sulfur compounds as model reactants, e.g., thiophene [12], benzothiophene [13] and dibenzothiophene (DBT) [10,14]. Only a few studies have used heavy and complex compounds such as methyl-substituted DBTs, which are known to be refractory to HDS [15–19]. We previously reported that fluorinated NiMo catalysts showed an improved activity in deep HDS compared with fluorine-free catalysts [20]. The HDS of DBT was promoted when fluorine was added to the catalyst up to a nominal amount of 0.5 wt.%, and the activity was then lowered by further fluorine addition. In the HDS of 4,6-dimethyldibenzothiophene (4,6-DMDBT), the catalytic activity increased in proportion to the fluorine content up to 5.0 wt.%, as the result of the facilitated migration (MIG) of methyl groups on the acid sites of the catalysts [20].

18

H. Kim et al. / Applied Catalysis B: Environmental 50 (2004) 17–24

In this study, we report on an examination of the performance of the fluorinated catalysts in the presence of nitrogen impurities. We compared the deactivation behavior of NiMo catalysts, containing different amounts of fluorine, in the HDS of either DBT or 4,6-DMDBT when the reactant solution contained nitrogen impurities. Quinoline and carbazole were selected as model nitrogen compounds that represent basic and non-basic impurities, respectively. We also characterized the surface of the catalysts, either fresh or poisoned, based on infrared spectra of pyridine and the amounts of NO adsorbed to the catalysts.

2. Experimental 2.1. Materials NiMo/Al2 O3 catalysts were prepared by the sequential impregnation of ␥-Al2 O3 (214 m2 g−1 surface area and 0.7 cm3 g−1 pore volume) with aqueous solutions of NH4 F, (NH4 )6 Mo7 O24 ·4H2 O, and Ni(NO3 )2 ·6H2 O. Between the impregnation steps, the catalyst was dried in air at 383 K for 12 h and then calcined in air at 723 K for 4 h. The prepared catalysts contained 17 wt.% MoO3 , 4 wt.% NiO, and different amounts of fluorine. DBT (99% purity) and 4,6-DMDBT (95% purity), used as reactants, were obtained from Acros, and carbazole (98% purity) and quinoline (98% purity), used as nitrogen impurities, were obtained from Fluka and Aldrich, respectively. 2.2. Catalytic activity Before the activity tests, all catalysts in the oxide phase were pre-sulfided in a stream of 12.9% hydrogen sulfide/hydrogen at 673 K for 2 h, as previously described [21,22]. Reaction tests were conducted using a 100 cm3 stirred slurry tank reactor, charged with 0.03–0.5 g of reactant dissolved in 30 cm3 of solvent. The reactor was operated in a batch mode at 593 K under a hydrogen pressure of 4.0 MPa for 1/2–2 h. n-Pentadecane was used as a solvent for DBT and dodecane for 4,6-DMDBT. For reactions in the presence of nitrogen impurities, carbazole or quinoline was dissolved in tetralin and the resulting solution was added to the reaction mixture containing DBT or 4,6-DMDBT and the corresponding solvent. During the reaction, the solution was sampled through a 1/8 in. diameter tube and the composition analyzed by gas chromatography (GC) using a capillary column (HP1; 30 m × 0.53 mm) equipped with an FID detector. The reaction products were also analyzed by GC–MS to identify individual products more precisely. 2.3. Characterization Prior to NO chemisorption, the pre-sulfided catalyst was suspended in a solution (dodecane 29 cm3 , tetralin 1 cm3 )

containing different amounts of nitrogen impurities at 323 K under atmospheric pressure for 24 h. The catalyst was then filtered and dried at 473 K in a stream of He for 0.5 h. The amounts of chemisorbed NO were measured by a dynamic method [7,23]. That is, pulses of the adsorption mixture (5 vol.% NO/Ar) were injected into a He stream flowing through the reactor until three successive pulses gave outlet signals for which their intensity change was within 1%. The total uptake of NO in this case was obtained. The catalyst was then flushed with He at 373 K for 1 h to remove physically adsorbed NO from its surface and the NO injection procedure was repeated. The difference in the NO uptake between the first and the second adsorption cycles gave the amount of chemisorbed NO. Infrared (IR) spectra of the catalysts, poisoned by different amounts of nitrogen impurities, were obtained using the same procedure as described in our previous article [22]. The catalyst sample was pressed into a self-supporting thin wafer and placed at the center of an IR cell vertical to the IR beam. The cell was evacuated to pressures below 10−5 Torr at 673 K for 2 h, cooled to room temperature, and pyridine was then introduced into the cell at 2 Torr. After evacuating the cell at 423 K to remove physically adsorbed pyridine from the catalyst surface, IR spectra of the chemisorbed pyridine were obtained using a Midac IR spectrometer.

3. Results 3.1. Effect of tetralin Since tetralin was used as a solvent for the nitrogen compounds and was eventually added to the HDS reaction mixture, we examined the possible modification of HDS rates by added tetralin under reaction conditions of this study. The results are shown in Table 1. ‘X’ in ‘FNiMoX’ indicates the nominal amount of fluorine added to the catalyst in a 0.1 wt.% unit. ‘No addition’ denotes the case when no tetralin was added to the reaction mixture and ‘1 ml’ indicates that 1 ml of tetralin, containing no nitrogen compounds, was added to the mixture. The HDS activity, which is higher for FNiMo05 and FNiMo50 than for FNiMo00 in the cases of DBT and 4,6-DMDBT, is decreased by the presence of tetralin in the mixture. The activity is decreased by about 15% due to tetralin addition in DBT HDS, independent of the fluorine content, while it is decreased by about 50% in the case of 4,6-DMDBT. These results indicate that sulfur compounds and added tetralin are competitively adsorbed to the active sites of the catalysts. Tetralin retards the hydrogenation (HYD) step to a greater extent than the direct desulfurization (DDS) step, which is clearly demonstrated in the case of DBT HDS. In addition, tetralin retards the HDS of 4,6-DMDBT to a greater extent than the HDS of DBT because the former reaction proceeds largely via the HYD route compared with the case of the latter reaction. Based

H. Kim et al. / Applied Catalysis B: Environmental 50 (2004) 17–24

19

Table 1 The effect of tetralin on the HDS of DBT and 4,6-DMDBT (a) DBT HDS Catalyst

Amount of added tetralin

Conversion (relative value)

No addition 1 ml No addition 1 ml

0.61 0.51 0.65 0.55

Amount of added tetralin

Conversion (relative value)

Amount of product, 10−5 mol (relative amount) CHB + DCH

BP FNiMo00 FNiMo05

(1.00)a [1.00]b (0.84) [1.00]c (1.00) [1.07] (0.85) [1.08]

83.50 74.08 84.04 75.60

(1.00)a (0.89) (1.00) (0.90)

15.65 9.75 21.75 13.75

(1.00)a (0.62) (1.00) (0.63)

(b) 4,6-DMDBT HDS Catalyst

Amount of product, 10−5 mol (relative amount) MCHT + DMDCH

DMBP FNiMo00 FNiMo50 a b c

No addition 1 ml No addition 1 ml

(1.00)a

[1.00]b

(1.00)a

0.51 0.25 (0.47) [1.00]c 0.60 (1.00) [1.30] 0.31 (0.52) [1.24]

2.00 1.16 (0.58) 2.44 (1.00) 1.32 (0.54)

5.24 2.24 6.01 3.14

(1.00)a (0.43) (1.00) (0.52)

Normalized conversion referring to the case of no tetralin. Normalized conversion referring to the case of FNiMo00 when the reaction mixture contains no tetralin. Normalized conversion referring to the case of FNiMo00 when the reaction mixture contains 1 ml of tetralin.

on the above information, we used the result obtained with a reaction mixture containing ‘1 ml of tetralin’ as a reference and results obtained in the presence of nitrogen compounds are compared to it. 3.2. Effect of nitrogen compounds 3.2.1. HDS of DBT In Table 2, ‘Y’ in ‘QuiY’ or ‘CarY’ designates the amount of quinoline or carbazole, expressed in weight-based ppm units. Since the molecular weight of carbazole is 30% higher than that of quinoline, Qui50 and Car65 represent equivalent amounts of the compounds. The HDS rate is essentially unaffected by nitrogen compounds when their concentrations are low, e.g. Qui50 and Car65, but is decreased when

the concentrations are higher. Quinoline, a basic compound, inhibits the HDS rate to a greater extent than carbazole, a non-basic compound. It is noteworthy that the decrease in activity due to the presence of nitrogen compounds, i.e. relative changes in the conversion, are nearly the same for both FNiMo00 and FNiMo05, although the latter catalyst consistently shows higher conversions than the former. This trend is demonstrated more clearly when the conversion data are plotted, as shown in Fig. 1. Among the products of the HDS, biphenyl (BP) is obtained in larger amounts than the sum of cyclohexylbenzene (CHB) and dicyclohexyl (DCH) under all conditions of this study. This trend is characteristic of DBT HDS, when carried out using promoted Mo catalysts [16], in which case the rate of DDS is higher than that of ring HYD. The data

Table 2 HDS of DBT in the presence of nitrogen compoundsa Catalyst

Conc. of N-comp. (ppm)

Conversion (relative value)

Amount of product, 10−5 mol (relative amount) BP

FNiMo00

FNiMo05

a b

CHB + DCH

Pure (tet 1 ml)

0.51 (1.00)

74.08 (1.00)

9.75 (1.00) [11.63]b

Qui 50 Qui250 Qui500

0.51 (1.00) 0.42 (0.82) 0.38 (0.75)

73.50 (0.99) 62.50 (0.84) 57.39 (0.77)

9.38 (0.96) [11.32] 5.91 (0.61) [8.64] 3.92 (0.40) [6.39]

Car 65 Car650

0.51 (1.00) 0.47 (0.92)

72.87 (0.98) 67.84 (0.92)

9.45 (0.97) [11.48] 8.78 (0.90) [11.46]

Pure (tet 1 ml)

0.55 (1.00)

75.60 (1.00)

13.75 (1.00) [15.39]

Qui 50 Qui250 Qui500

0.55 (1.00) 0.45 (0.82) 0.41 (0.75)

75.51 (1.00) 66.74 (0.89) 62.19 (0.83)

13.54 (0.98) [15.20] 7.12 (0.72) [9.64] 4.77 (0.48) [7.12]

Car 65 Car650

0.55 (1.00) 0.51 (0.93)

74.92 (0.99) 71.07 (0.94)

13.91 (1.01) [15.66] 12.54 (0.91) [15.00]

Error range of conversions and product amounts is ±2%. Relative fractions (%) of HYD products among total products.

20

H. Kim et al. / Applied Catalysis B: Environmental 50 (2004) 17–24 0.30

0.55 0.25

0.50

Conversion

Conversion

Carbazole(FNiMo05)

Carbazole(FNiMo00)

0.45

Quinoline(FNiMo05)

0.40

0.20

Carbazole (FNiMo50)

0.15

Carbazole (FNiMo00)

0.10

Quinoline (FNiMo50)

0.05

Quinoline (FNiMo00)

Quinoline(FNiMo00)

0.35

0

100

200

300

400

500

600

0.00

700

0

Conc. of N-compounds (ppm)

100 200 300 400 500 600 700

Conc. of N-compounds (ppm)

Fig. 1. The effect of nitrogen compounds on the HDS of DBT.

Fig. 2. The effect of nitrogen compounds on the HDS of 4,6-DMDBT.

in Table 2 shows that the amounts of CHB and DCH are decreased to greater extents than those of BP in the presence of nitrogen compounds and that the trend is more evident in the case of quinoline. The results of DBT HDS can be summarized as follows: The activity is decreased to the same extent by nitrogen compounds independent of the extent of fluorination of the catalyst. However, the decrease in activity is more significant with quinoline than with carbazole and the HYD step is retarded to a greater extent than the DDS step.

is retarded even by small amounts of nitrogen compounds, which is in contrast to the case of DBT HDS. That is, the conversion obtained using FNiMo00 is lowered to 80% of the initial value, i.e. from 0.25 to 0.20, when 10 ppm quinoline, Qui10, is present in the reaction mixture. The conversion is further lowered to about 20% by the addition of Qui500. The same trend is observed with a fluorinated catalyst, FNiMo50. Again, quinoline is more effective in retarding the activity than carbazole. A closer observation of changes in the conversion between two catalysts, FNiMo00 and FNiMo50, reveals that the conversion is decreased to a greater extent for FNiMo00 than for FNiMo50, particularly at low concentrations of nitrogen compounds. The above trend is more distinctly demonstrated by a plot of the conversion data, as shown in Fig. 2. The conversion curves for two catalysts are not parallel with but

3.2.2. HDS of 4,6-DMDBT Table 3 shows that the rate of 4,6-DMDBT HDS is increased when fluorine is added to the catalyst, which is similar to the case of DBT HDS and was reported in our previous article [20]. However, the HDS of 4,6-DMDBT Table 3 HDS of 4,6-DMDBT in the presence of nitrogen compoundsa Catalyst

FNiMo00

FNiMo50

a b

Conc. of N-comp. (ppm)

Conversion (relative value)

Amount of product, 10−5 mol (relative amount) DMBP

MCHT + DMDCH

Pure (tet 1 ml)

0.25 (1.00)

1.16 (1.00)

2.24 (1.00) [65.88]b

Qui 10 Qui 50 Qui250 Qui500

0.20 0.15 0.08 0.05

(0.80) (0.60) (0.32) (0.20)

0.97 0.83 0.51 0.30

(0.84) (0.72) (0.44) (0.26)

1.75 1.26 0.72 0.41

(0.78) (0.56) (0.32) (0.18)

[64.34] [60.29] [58.54] [57.75]

Car13 Car 65 Car325 Car650

0.23 0.18 0.14 0.13

(0.92) (0.72) (0.56) (0.52)

1.06 0.86 0.69 0.62

(0.91) (0.74) (0.59) (0.53)

2.23 1.77 1.33 1.18

(1.00) (0.79) (0.59) (0.53)

[67.78] [67.30] [65.84] [65.56]

Pure (tet 1 ml)

0.31 (1.00)

1.32 (1.00)

3.14 (1.00) [70.40]

Qui 10 Qui 50 Qui250 Qui500

0.25 0.21 0.11 0.07

(0.81) (0.68) (0.35) (0.23)

1.14 0.91 0.59 0.30

(0.86) (0.69) (0.45) (0.23)

2.44 2.12 1.02 0.59

(0.78) (0.68) (0.32) (0.19)

[68.16] [69.97] [63.35] [66.29]

Car 13 Car 65 Car325 Car650

0.30 0.26 0.20 0.17

(0.97) (0.84) (0.65) (0.55)

1.14 1.00 0.77 0.63

(0.86) (0.76) (0.58) (0.48)

3.14 2.62 2.12 1.76

(1.00) (0.83) (0.68) (0.56)

[73.36] [72.38] [73.36] [73.64]

Error range of conversions and product amounts is ±2%. Relative fractions (%) of HYD products among total products.

H. Kim et al. / Applied Catalysis B: Environmental 50 (2004) 17–24

10

pure Car65 Qui50

9

05

05

00

00

Car650

-6

NO uptake (10 mol/g cat)

become closer to each other, for nitrogen levels above 250 (quinoline) or 325 ppm (carbazole). This trend is in contrast to that observed for DBT HDS (Fig. 1), which shows that two conversion curves are parallel, maintaining a constant difference between them, for all concentrations of nitrogen compounds. The reason for the contrasting trends between the two cases will be discussed in Section 4. The product distribution of 4,6-DMDBT HDS is different from that of DBT HDS, i.e. HYD products (MCHT + DMDCH) are produced in larger amounts than the DDS product (DMBP). The difference in product distribution from that for DBT HDS originates from steric hindrance originating from the methyl groups attached to the 4,6-DMDBT ring [15–18]. When carbazole is added to the reaction mixture, DDS products are retarded to a slightly greater extent than the HYD products, consistent with our previous study of CoMo/Al2 O3 catalysts [22]. On the other hand, when quinoline is added to the reaction mixture, the amounts of the HYD products are decreased to a greater extent than that for the DDS product. This result is similar to the case of DBT HDS, but is contrary to the result obtained with CoMo/Al2 O3 catalysts [22]. The difference in the retardation effect between quinoline and carbazole will be discussed in Section 4.

21

05

8 00

Qui500

7

00

05

6 0 (a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

Catalyst (Conc. of N-compound) Fig. 3. The amounts of NO chemisorbed on catalysts poisoned by different amounts of nitrogen compounds: (a) FNiMo00 (pure); (b) FNiMo00 (Qui50); (c) FNiMo05 (Qui50); (d) FNiMo00 (Qui500); (e) FNiMo05 (Qui500); (f) FNiMo00 (Car65); (g) FNiMo05 (Car65); (h) FNiMo00 (Car650); (i) FNiMo05 (Car650).

present in the reaction mixture. However, the amount of NO is consistently larger on FNiMo05 than on FNiMo00 even after poisoning by nitrogen impurities, indicating that the amount of active sites is larger on the former catalyst than on the latter. These trends are consistent with the reaction results.

3.3. NO chemisorption 3.4. Pyridine-IR NO is a probe molecule that is selectively adsorbed to the edge or the corner sites of MoS2 -like structures or to cobalt atoms at the edges of MoS2 in CoMoS catalysts [24], and can also be applied to NiMoS catalysts. Fig. 3 indicates that the amount of NO adsorbed on the catalyst is decreased in proportion to the level of the nitrogen impurity, which is more serious when quinoline is

Fig. 4 indicates that Lewis and Brönsted acid sites, corresponding to bands at 1450 and 1540 cm−1 , respectively [25], are present on the catalysts. The addition of fluorine to the catalysts increases the amounts of both acid sites but the Brönsted sites are increased to a greater extent.

FNiMo00

FNiMo50 L-acid site

L-acid site B-acid site

B-acid site

Car650

Car65

Qui500

Absorbance

Absorbance

Car650

Car65

Qui500 Qui50

Qui50

Pure

Pure 1540

1700

(a)

1600

1450

1500

1400 -1

Wavenumber (cm )

1700

(b)

1600

1500

1400 -1

Wavenumber (cm )

Fig. 4. FTIR spectra of pyridine adsorbed on (a) FNiMo00 and (b) FNiMo50 poisoned by different amounts of nitrogen compounds.

22

H. Kim et al. / Applied Catalysis B: Environmental 50 (2004) 17–24

When the catalysts are exposed to nitrogen impurities, both types of acid sites are poisoned as shown in Fig. 4a for the case of FNiMo00. Similar to the case of fluorine addition, the amounts of Brönsted sites are decreased to a greater extent than those of Lewis sites by the poisoning. This trend is more significant with quinoline than with carbazole. FNiMo50 is subject to the same poisoning effect as shown in Fig. 4b. However, relatively large amounts of acid sites are retained by the catalysts even after exposure to Qui500 or Car650. This result is in contrast to the case of FNiMo00, in which the intensity of the 1540 cm−1 peak for Brönsted acid sites, after poisoning by the same amounts of the impurities, is negligible.

4. Discussion 4.1. Mechanism of the HDS of DBT compounds The HDS of DBT compounds proceeds via two major routes, as schematically shown in Fig. 5 [26–28]. One is a DDS route, which involves the hydrogenolysis of C–S bonds with no hydrogenation of the aromatic rings, and the other is a HYD route, which requires ring saturation prior to the desulfurization step. The HDS of DBT proceeds mainly via the DDS route, while that of 4,6-DMDBT proceeds largely via the HYD route due to the steric hindrance of the methyl groups [15–19]. Fluorine, when added to NiMo/Al2 O3 , improves not only the dispersion of active species but the acidity of the catalyst as well. The increased acidity promotes the HYD of BP or DMBP to CHB or MCHT, which is further hydro-

Table 4 Product distribution in the hydrogenation and isomerization of 2,2 DMBPa Compounds

Amounts (mol cm−3 ) FNiMo00

2,2 -DMBP

(unreacted) MCHT DMDCH 3,3 - and 4,4 -DMBP 2,8-DMBP

10−5

4.53 × 1.82 × 10−6 1.53 × 10−6 2.23 × 10−7 Trace

FNiMo50 4.09 × 3.70 × 1.57 × 4.03 × Trace

10−5 10−6 10−6 10−7

a Reaction condition: P = 40 bar; temperature = 320 ◦ C; reaction peH2 riod = 2 h.

genated to DCH or DMDCH, respectively. In addition, the acidity facilitates the MIG of methyl substituents in the case of 4,6-DMDBT HDS [20,21]. As one of methods to verify the migration of methyl groups in the ring structure on fluorine-added catalysts, we made an independent reaction test using 2,2 -DMBP as a reactant [20]. In Table 4, DMBP isomers, with methyl groups attached to positions other than the 2,2 position, are obtained in smaller amounts than the HYD products, MCHT and DMDCH, by an order of magnitude. Nevertheless, it is apparent that DMBP isomers are obtained in larger amounts on fluorine-added catalyst than on fluorine-free one. However, the addition of excessive amounts of fluorine reduces the catalyst surface area and, as a result, the HDS activity changes showing a maximum with the fluorine content. In our previous study, the maximum activity was obtained with FNiMo05 in the case of DBT HDS and with FNiMo50 in the HDS of 4,6-DMDBT [20]. In fact, this is the reason for the selection of FNiMo05 and FNiMo50 as the two sample catalysts of this study.

Fig. 5. Reaction routes in the HDS of DBT and 4,6-DMDBT.

H. Kim et al. / Applied Catalysis B: Environmental 50 (2004) 17–24

It was reported for the case of CoMo/Al2 O3 that the HDS rate is retarded by the presence of quinoline and carbazole in the reaction mixture because the nitrogen impurities poison the acidic and the active sites of the catalyst [22]. This trend is somewhat opposite to that observed for fluorine addition. 4.2. Combined effects of fluorine and nitrogen compounds on HDS activity The results in Tables 2 and 3 indicate that the individual effects of fluorine and nitrogen compounds on HDS activity, which were independently investigated in previous studies [20,22], can be observed in a combined manner in this study. That is, HDS activity is increased by fluorine addition to the catalyst and decreased by the presence of nitrogen compounds in the reaction mixture. Quinoline, a basic compound, poisons the catalyst more significantly than carbazole, a non-basic compound [29]. This trend is observed in both cases of DBT and 4,6-DMDBT. However, the difference in the activity between fluorinated and fluorine-free catalysts is constant and independent of the impurity level in the case of DBT HDS, while the difference decreases with an increase in the impurity level in 4,6-DMDBT HDS. The reason for the above contrasting results can be explained as follows: The HDS of DBT is retarded because nitrogen compounds poison the catalytic sites that are responsible for HYD and DDS, which occurs to the same extent independent of the initial fluorine content of the catalyst. The HDS of 4,6-DMDBT, which is additionally affected by the MIG of methyl groups in the ring structure, is retarded to a greater extent than DBT HDS by the nitrogen impurities. In other words, the impurities poison the acidic sites of the catalyst, which are responsible for the MIG step, and therefore additionally retard the HDS rate. When the impurity level is relatively low, the rate of 4,6-DMDBT HDS is decreased due to the suppression of above three factors, HYD, DDS and MIG, and the extents of the poisoning are nearly the same for FNiMo00 and FNiMo50. On the other hand, when the impurity level is high, large amounts of acidic sites are poisoned by the impurities and therefore the contribution of MIG to the HDS rate decreases as well. FNiMo00, which contains relatively small amounts of acidic sites, is readily depleted of its initial acid sites by the poisoning and, as a result, the contribution of MIG to the HDS rate becomes negligible at high impurity levels. In this case, the deactivation of FNiMo00 proceeds largely due to the suppression of HYD and DDS. In fact, the slope of the deactivation curve obtained for FNiMo00 in 4,6-DMDBT HDS decreases with the concentration of the impurity and approaches that for DBT HDS, indicating that the catalyst is deactivated based on the same mechanism at high impurity levels in both cases. Unlike FNiMo00, FNiMo50, which contains many acid sites, retains a large fraction of the initial acidic sites even at high impurity levels. As a result, FNiMo50 is deactivated due to the suppression of three factors,

23

HYD, DDS and MIG, at all impurity levels and accordingly at a higher rate than in the case of FNiMo00. The above argument, which explains the non-parallel changes in the deactivation curves between FNiMo00 and FNiMo50 in the case of 4,6-DMDBT HDS, is supported by the following additional evidence. Fig. 3 shows that the amounts of adsorbed NO decrease in parallel with the impurity levels for both FNiMo00 and FNiMo05. However, the acidic sites are poisoned by the impurities and show different trends depending on the catalysts. The IR spectra of pyridine, shown in Fig. 4, indicate that, at high levels of nitrogen impurities (Qui500 or Car650), FNiMo00 is nearly depleted of Brönsted acid sites, which are responsible for MIG [30], while FNiMo50 still retains considerable amounts of these sites. 4.3. Combined effects of fluorine and nitrogen compounds on the product distribution When fluorine is added to NiMo/Al2 O3 , the amounts of HYD products are increased to a greater extent in DBT HDS, while DDS products are preferentially promoted in the HDS of 4,6-DMDBT. These results are consistent with our previous report [20,21]. When quinoline or carbazole is present in the reaction mixture during DBT HDS, the HYD step is retarded more significantly than the DDS step regardless of the fluorine contents of the catalysts. In the HDS of 4,6-DMDBT, the same trend is observed due to poisoning by quinoline. However, carbazole retards the DDS step to a slightly greater extent than the HYD step. Nitrogen compounds are adsorbed more effectively to the corner sites of MoS2 -like structures, which are responsible for the HYD step, than to the edge sites, which are responsible for the DDS step [11,31]. The acidic sites of the catalysts, which also promote the HYD step [32], are additionally poisoned by the nitrogen impurities. As a result, the HYD step is retarded to a greater extent than the DDS step in DBT HDS.The same explanation is valid for the case of 4,6-DMDBT HDS in the presence of quinoline, although the difference in the extents of poisoning between the HYD and DDS steps is not as large as in the case of DBT HDS. The small difference between the two steps in the case of 4,6-DMDBT is due to the fact that the DDS step, which is affected by MIG, is additionally suppressed by nitrogen impurities. The result in which carbazole retards DDS to a slightly greater extent than HYD, an opposite trend from the case of quinoline, can be explained as follows: Carbazole is not as effective as quinoline in poisoning the catalysts. As a result, the extents of suppression due to carbazole for the HYD and DDS steps, as well as the difference between the two extents, are smaller than in the case of quinoline. Nevertheless, the extent of suppression is greater for HYD than for DDS in the HDS of DBT, as shown in Table 2. In the HDS of 4,6-DMDBT, the trend changes depending on the nitrogen compounds because DDS is additionally affected by MIG, which is promoted on the

24

H. Kim et al. / Applied Catalysis B: Environmental 50 (2004) 17–24

acidic sites and therefore is retarded by the impurities. In the case of poisoning by quinoline, the effect of retarded MIG on DDS is dominated by the large suppression of HYD and, accordingly, the extent of HYD suppression remains greater than that of DDS suppression. In the case of poisoning by carbazole, which suppresses HYD and DDS to small and comparable extents, the retardation of MIG has a significant effect on DDS and, as a result, DDS is suppressed to a greater extent than HYD. The significant suppression of HYD by quinoline in the HDS of 4,6-DMDBT using NiMo/Al2 O3 is in contrast to the case of CoMo/Al2 O3 , which shows a larger suppression of DDS than HYD [22]. The reason for this is because NiMo/Al2 O3 has a higher HYD activity than CoMo/Al2 O3 and therefore, in the case of NiMo/Al2 O3 , the HYD rates are retarded to a greater extent than the DDS rates compared with the case of the CoMo catalyst.

5. Conclusion When fluorine-added NiMo/Al2 O3 catalysts are used for the HDS of DBT or 4,6-DMDBT in the presence of quinoline or carbazole, the promotional effect of fluorine and the poisoning effect of nitrogen compounds can be observed simultaneously. The HDS activity and the product distribution change characteristically depending on the combination of the reactants and the nitrogen impurities present. The HDS activity is decreased by the presence of the impurities in the reaction mixture and the extent of poisoning is greater for quinoline than for carbazole. However, fluorine-added catalysts maintain higher activities than fluorine-free ones at all impurity levels. The difference in the activities between the two types of catalysts is maintained at a constant level in DBT HDS but, in the HDS of 4,6DMDBT, the difference is decreased with an increase in the impurity level. The reason for this is because the HDS of 4,6DMDBT, particularly the DDS step, is additionally affected by MIG, which is promoted on acidic sites of the catalysts. The acidic sites on FNiMo00 are readily depleted by nitrogen impurities but FNiMo50, which contains relatively large amounts of acidic sites, retains a large fraction of the sites even after poisoning by the same amounts of the impurities. In DBT HDS, nitrogen impurities suppress HYD to a greater extent than DDS, and this effect is independent of the fluorine content of the catalysts. In the HDS of 4,6-DMDBT, the same trend as for DBT HDS is observed due to poisoning by quinoline but an opposite trend is observed in the case of carbazole. When the catalyst is poisoned by carbazole, which suppresses HYD and DDS to smaller extents than quinoline, the retardation of MIG contributes significantly to the suppression of DDS. The reason for why quinoline suppresses HYD to a greater extent than DDS on NiMo catalysts, which is opposite to the trend observed for CoMo catalysts, is because the former catalysts have a higher HYD activity than the latter.

Acknowledgements This work was supported by R&D Management Center for Energy and Resources, Brain Korea 21 project, and National Research Laboratory program. References [1] M.V. Landau, D. Berger, M. Herskowitz, J. Catal. 159 (1996) 236. [2] T. Isoda, S. Nagao, X. Ma, Y. Korai, I. Mochida, Energy Fuels 10 (1996) 1078. [3] P. Michaud, J.L. Lemberton, G. Perot, Appl. Catal. A 169 (1998) 343. [4] C. Kwak, M.Y. Kim, K. Choi, S.H. Moon, Appl. Catal. A 185 (1999) 19. [5] J.L.G. Fierro, R. Cuevas, J. Ramirez, A.L. Agudo, Bull. Soc. Chim. Belg. 100 (1991) 945. [6] C. Papadopoulou, C. Kordulis, A. Lycourghiotis, React. Kinet. Catal. Lett. 33 (1987) 259. [7] C. Papadopoulou, A. Lycourghiotis, P. Grange, B. Delmon, Appl. Catal. 38 (1988) 255. [8] Z. Sarbak, S.L.T. Andersson, Appl. Catal. 69 (1991) 235. [9] H.K. Matralis, A. Lycourghiotis, P. Grange, B. Delmon, Appl. Catal. 38 (1988) 273. [10] F. Looij, P. Laan, W.H.J. Stork, D.J. Dicamillo, J. Swain, Appl. Catal. A 170 (1998) 1. [11] H. Topsøe, B.S. Clausen, F.E. Massoth, in: J.R. Anderson, M. Boudart (Eds.), Catalysis Science and Technology, vol. 11, Springer, Berlin, 1996. [12] C.N. Satterfield, M. Modell, J.F. Mayer, AIChE J. 21 (1975) 1100. [13] S.W. Cowley, F.E. Massoth, Canad. J. Chem. 42 (1964) 843. [14] M. Nagai, T. Sato, A. Aiba, J. Catal. 97 (1986) 52. [15] X. Ma, K. Sakanishi, I. Mochida, Ind. Eng. Chem. Res. 33 (1994) 218. [16] V. Meille, E. Schulz, M. Lemaire, M. Vrinat, Appl. Catal. A 131 (1995) 143. [17] V. Meille, E. Schulz, M. Lemaire, M. Vrinat, J. Catal. 170 (1997) 29. [18] M. Houalla, D.H. Broderick, A.V. Sapre, N.K. Nag, V.H.J. de Beer, B.C. Gates, H. Kwart, J. Catal. 61 (1980) 523. [19] F. Bataille, J.L. Lemberton, P. Michaud, G. Pérot, M. Vrinat, M. Lemaire, E. Schulz, M. Breysse, S. Kasztelan, J. Catal. 191 (2000) 409. [20] H. Kim, J.J. Lee, S.H. Moon, Appl. Catal. B 44 (2003) 287. [21] C. Kwak, J.J. Lee, J.S. Bae, K. Choi, S.H. Moon, Appl. Catal. A 200 (2000) 233. [22] C. Kwak, J.J. Lee, J.S. Bae, S.H. Moon, Appl. Catal. B 35 (2001) 59. [23] H. Matralis, C. Papadopoulou, A. Lycourghiotis, Appl. Catal. A 116 (1994) 221. [24] N.Y. Topsøe, H. Topsøe, J. Catal. 84 (1983) 386. [25] Z. Sarbak, Appl. Catal. A 164 (1997) 13. [26] H. Farag, D.D. Whitehurst, K. Sakanishi, I. Mochida, Catal. Today 50 (1999) 49. [27] M.L. Vrinat, Appl. Catal. 6 (1983) 137. [28] M. Houalla, N.K. Nag, A.V. Sapre, D.H. Broderick, B.C. Gates, AIChE J. 24 (1978) 1015. [29] G.C. Laredo, J.A. Reyes, J.L. Cano, J. Castillo, Appl. Catal. A 207 (2001) 103. [30] B.C. Gates, J.R. Katzer, G.C.A. Schuit (Eds.), Chemistry of Catalytic Processes, McGraw-Hill, New York, 1979. [31] S. Kasztelan, H. Toulhoat, J. Grimblot, J.P. Bonnelle, Appl. Catal. 13 (1984) 127. [32] M. Breysse, M. Cattenot, V. Kougionas, J.C. Lavalley, F. Mauge, J.L. Portefax, J.L. Zotin, J. Catal. 168 (1997) 143.