Songdong
Yao
a,
Ying
Zheng
*a,
Lianhui
Ding
a,
Siauw
Ng
b and
Hong
Yang
b
aDepartment of Chemical Engineering, University of New Brunswick, P.O. Box 4400, Fredericton, NB E3B 5A3, Canada. E-mail: yzheng@unb.ca
bCanmetENERGY, Devon, AB T9G 1A8, Canada
First published on 7th May 2012
A hydrotreating catalyst, NiMo/Al2O3, was modified with different amounts of fluorine (F) and boron (B) through the pore-saturated impregnation method. The resulting catalysts were characterized by BET, pyridine-IR, XPS, and TEM techniques. Incorporation of fluorine and boron led to the variations in the catalyst acidity and the dispersion of active metals, causing direct impacts on the hydrotreating activities of the catalysts. The hydrodesulfurization (HDS), hydrodenitrogenation (HDN) and hydrodearomatization (HDA) activities of the catalysts were examined in an autoclave reactor using real light cycle oil as feed. The HDS activity decreased in the order NiMo/F,B-Al(5.0) > NiMo/F,B-Al(7.0) > NiMo/F,B-Al(3.5), indicating the existence of an optimum amount of F and B.
Fluorine and boron can be used as promoters to enhance the acidity of Al2O3-supported hydrotreating catalysts. It has been reported that promoter boron may change the type and strength of the acidity of alumina. With co-precipitation, boron may substitute some Al atoms to form B3+, and both Brønsted and Lewis acidic sites could be formed in the presence of H2O.12 The number of acid sites is increased with increasing boron incorporation.13 When boron was impregnated into Al2O3, H3BO3 may react with Al–OH to form B–OH or A1–O–B–O–A1.14,15 When the content of B2O3 is low (<0.8 wt%), it will form monolayer acid sites on the surface of Al2O3. This increases the strength of the acid sites but does not significantly change the total number of acid sites.14,16 When the content of B2O3 is higher than 0.8 wt%, the surface B–O–H groups will be condensed with the Al–OH groups to form B–O–Al species, increasing the strength of the acid sites.14
Introduction of fluorine can change the surface structure and acidic properties of alumina.17 Studies have shown that fluorine-doped alumina can enhance Brønsted and Lewis acid strength, and increase the number of Lewis acid sites.18–22 Meanwhile, fluorine can also improve the metal dispersion and enhance the formation of active CoMoS leading to augmented activity of the catalyst.23 However, excessive Brønsted acids induced by high content of fluorine can encourage the formation of coke, which accelerates the catalyst deactivation.23
In our previous study, the NiMo/γ-Al2O3 catalyst with simultaneous B and F modification presented a better hydrogenation activity than the individual fluorine or boron-modified catalyst.24 Synergic interactions between boron and fluorine species enhanced the performance of the catalyst. The present work focuses on the variations in the catalyst acidity with incorporation of both F and B promoters. The performances in hydrodesulfurization (HDS) and hydrodenitrogenation (HDN) of the catalysts were studied in an autoclave using light cycle oil (LCO) as feed.
MoNi/F,B-Al(3.5) | MoNi/F,B-Al(5.0) | MoNi/F,B-Al(7.0) | |
---|---|---|---|
Composition, wt% | |||
MoO3 | 15 | 15 | 15 |
NiO | 5 | 5 | 5 |
Al2O3 | 55.9 | 54.2 | 51.9 |
Binder Al2O3 | 20 | 20 | 20 |
F | 3.5 | 5.0 | 7.0 |
B | 0.56 | 0.8 | 1.12 |
Surface area, m2 g−1 | 171 | 122 | 99 |
Pore volume, ml g−1 | 0.48 | 0.46 | 0.45 |
Average MoS2 slab length, nm | 6.68 | 5.95 | 5.99 |
Average MoS2 slab layer | 2.52 | 2.31 | 2.23 |
Survey scans were collected for binding energy from 1100 eV to 0 with analyzer pass energy of 160 eV and a step of 0.35 eV. For the high-resolution spectra the pass-energy was 20 eV and the step was 0.1 eV. The number of scans varied depending on the signal/noise ratio, from 12 for S2p to 16 for Mo3d and Mo3p and 30 for Ni2p. Note that the major molybdenum peak Mo3d overlaps with the minor sulphur peak S2s. The high resolution C1s spectra of the adventitious hydrocarbon (at 284.6 eV) on the sample surfaces were recorded before and after each measurement and used as a reference for charge correction.
Fig. 1 presents the pyridine IR spectra of NiMo/F,B-Al(3.5), NiMo/F,B-Al(5.0) and NiMo/F,B-Al(7.0) desorbed at 423 K, 523 K, 623 K and 723 K, respectively. There are at least 4 peaks which can be found in the pyridine IR spectra in the range of 1400–1700 cm−1. The bands at 1450 cm−1 and 1540 cm−1 are attributable to the adsorption of pyridine on Lewis and Brønsted acid sites, respectively. The peak centering at 1492 cm−1 indicates co-adsorption of pyridine on both Lewis and Brønsted acid sites. The vibration intensity of the IR bands at 1450, 1492 and 1540 cm−1 becomes weaker as desorption temperature increases. When temperature reached 723 K, the bands at 1540 cm−1 representing the Brønsted acid site nearly disappeared. The bands at 1608 cm−1 can be assigned to the adsorption of pyridine on the tetrahedral aluminum.22,24–26 The quantities of the Lewis and Brønsted acid sites of the catalysts can be calculated from the areas under the peaks centered at 1450 cm−1 and 1540 cm−1 at the desorption temperature of 423 K.
Fig. 1 IR spectra of pyridine adsorption for catalysts. (a) MoNi/F,B-Al(3.5); (b) MoNi/F,B-Al(5.0); and (c) MoNi/F,B-Al(7.0) at 423 K, 523 K, 623 K and 723 K. |
Fig. 2 depicts the acidity values of Brønsted and Lewis acid sites for the three catalysts. It is noted that Lewis acidity predominated for all three catalysts. When boron content in the NiMo catalysts increased from 0.56 to 0.8, the total acidity of Brønsted and Lewis acid sites almost doubled. Higher contents of F and B, especially B, can encourage the formation of acid sites on the catalyst. This is due to the fact that boron acid tends to react with hydroxyl groups of the alumina surface to form monolayer B–O–H species.14 However, further increase in the content of B from 0.8 to 1.12 resulted in a significant drop in the total or individual acidity. Three possibilities might be responsible for this observation: (1) overloading of boron resulted in the formation of multilayer boron species which triggered a weaker interaction between the surface boron and alumina; (2) excessive addition of F resulted in the formation of a significant amount of AlF63− species, which covered the Lewis acid sites and thus reduced the acidity of the catalyst; and (3) overdosage of F replaced OH of Brønsted acid sites leading to a lower catalyst acidity.18
Fig. 2 Acidities of MoNi/F,B-Al(3.5), MoNi/F,B-Al(5.0) and MoNi/F,B-Al(7.0). |
Increasing temperature could cause desorption of pyridine from acid sites. Strongly absorbed pyridine molecules require more energy or higher temperature to be desorbed from Lewis and Brønsted acid sites. Therefore, the absorbed pyridine at different desorption temperatures (723, 623, 523 and 423 K) was arbitrarily used as the indicator to reflect the adsorption power of the acid sites (i.e., ultra-strong, strong, medium and weak, respectively). Since Brønsted acidity was minor for all the three catalysts, detailed analysis was focused on Lewis acid sites. Fig. 3 compares the relative Lewis acidity of different strength levels for all the three NiMo/F,B-Al catalysts. It is seen that the weak Lewis acid sites decreased while ultra-strong Lewis acid sites increased as the contents of F and B are increased. F may be responsible for the increase in the strength of the Lewis acid sites.18,24
Fig. 3 The weak, medium, strong and very strong L acid of NiMo/F,B-Al catalysts. |
Fig. 4 XPS spectra of the sulfided catalysts. (i) F1S; (ii) B1S; (iii) Ni2p; (iv) Mo3d; (a) NiMo/F,B-Al(3.5), (b) NiMo/F,B-Al(5.0) and (c) NiMo/F,B-Al(7.0). |
MoNi/F,B-Al(3.5) | MoNi/F,B-Al(5.0) | MoNi/F,B-Al(7.0) | ||
---|---|---|---|---|
F1S | AlF63− | 685.5 | 685.6 | 685.5 |
AlF3 | 687.8 | 687.9 | 687.8 | |
B1S | B2O3 | 192.0 | 191.9 | 191.9 |
B/Al2O3 | 192.6 | 192.6 | 192.6 | |
Ni2p3/2 | Disulfide | 853.5 | 853.7 | 853.5 |
— | Ni, sat | 861.4 | 862.9 | 862.6 |
NiAl2O4/Ni2O3 | 856.0 | 856.2 | 856.1 | |
Mo3d5/2 | MoS2 | 228.9 | 228.7 | 228.7 |
MoOx | 232.2 | 231.8 | 232.0 | |
S2s | S | 226.0 | 226.0 | 226.0 |
Atomic concentration ratio, % | ||||
F/Al | Bulk | 11.7 | 16.8 | 23.5 |
surface | 8.7 | 20.0 | 22.6 | |
B/Al | Bulk | 3.3 | 4.7 | 6.6 |
surface | Trace | Trace | Trace | |
Ni/Al | Bulk | 4.25 | 4.25 | 4.25 |
surface | 4.79 | 4.20 | 4.19 | |
Mo/Al | Bulk | 6.64 | 6.64 | 6.64 |
surface | 10.60 | 9.74 | 8.30 |
All the catalysts have similar patterns of F1s, B1s, Ni2p and Mo3d spectra. The binding energy of F1s (685.5–685.6 eV) observed in the XPS is close to that of Na3AlF6 (685.50 eV) but different from those of the NiF2 (685.1 eV) and NaBF4 (687.0 eV).27 Boorman et al. (1985) assigned this peak to the F ions replacing hydroxyl groups on the alumina surface, indicating that fluorine ions may react with the Al2O3 supports.28 A small shoulder peak of F1s (687.8–687.9 eV) may be assigned to the AlF3 (687.5 eV). The binding energy of B1s (192.0–192.6 eV) is close to that of B2O3 (192.4 eV) and (Al2O3)9(B2O3)2 (192.70 eV) as well as B/Al2O3 (192.70 eV), but slightly lower than that of H3BO3 (193.0 eV). This indicates that a small amount of boron may exist in the form of B2O3 and be bonded to the surface Al2O3. The binding energy of Ni2p3/2 indicates that Ni species are mainly in their sulfide status (853.5–853.7 eV). The Mo3d spectra can be split into some doublet peaks, corresponding to the Mo6+, Mo5+ and Mo4+ species as well as MoS2 species. This suggests that Mo species mainly exist in the form of MoOxSy, MoS2 or NiMoS phase and are not fully sulfided in the initial sulfiding step.10,11
Compared to catalyst NiMo/F,B-Al(3.5), a significant amount of F has been retained on the surface of catalyst NiMo/F,B-Al(5.0). More F species are present on the surface than in the bulk. The distribution pattern of B species remains the same, that is, only a trace amount of B can be identified on the surface of catalyst NiMo/F,B-Al(5.0). Opposite to the dispersion pattern of F species, approximately half of the active metal Ni has migrated to the bulk of catalyst NiMo/F,B-Al(5.0).
Further increase in incorporation of F and B seems to encourage the migration of F into the bulk. In catalyst NiMo/F,B-Al(7.0), F uniformly distributes on the surface and the bulk of catalyst. Similar to catalyst NiMo/F,B-Al(5.0), B and two active metals Mo and Ni exhibit similar distribution patterns in catalyst NiMo/F,B-Al(7.0).
Fig. 5 shows the representative TEM images of the three sulfided catalysts. The black lines in the image were identified as the layers of MoS2 in the verticle planes (Fig. 5). The number of the parallel black lines denotes the number of the layers. Arithmetic averages of the number of layers of MoS2 slabs and layer length are calculated from more than 300 crystallites measured from the TEM images.30–32
Fig. 5 TEM images of NiMo/F,B-Al(3.5), NiMo/F,B-Al(5.0) and NiMo/F,B-Al(7.0). |
Fig. 6 and 7 show the distribution of the length and the number of stack layers of MoS2 slabs. The average length and layers of the MoS2 slabs are listed in Table 1. More than 20.6% MoS2 slabs on catalyst NiMo/F,B-Al(3.5) have length longer than 7 nm. Catalysts NiMo/F,B-Al(5.0) and NiMo/F,B-Al(7.0) have lengths about 16% longer than 7 nm. Concerning the thickness of the MoS2 slabs, about 21.2% MoS2 slabs on the NiMo/F,B-Al(3.5) catalyst are thicker than 3 layers while only 12.7% and 13.4% MoS2 slabs are thicker than 3 layers for catalyst NiMo/F,B-Al(5.0) and NiMo/F,B-Al(7.0), respectively. The results indicate that the active phase of catalyst NiMo/F,B-Al(3.5) is more aggregated than that of the other two catalysts. This is consistent with the XPS results where the surface NiMo species decreased in the same order NiMo/F,B-Al(3.5) > NiMo/F,B-Al(5.0) > NiMo/F,B-Al(7.0). Increased addition of F and B can reduce the amounts of surface Ni and Mo species and therefore lower the aggregation of the active NiMoS slabs.
Fig. 6 The distribution of slab length of the sulfided NiMo/F,B-Al2O3 catalysts. |
Fig. 7 The distribution of slab layer of the sulfided NiMo/F,B-Al2O3 catalysts. |
(1) |
The amount of Mo atoms at top edges ntop(i) = 6nMo(i)− 6 | (2) |
The amount of Mo atoms at slabs nslab(i) = (3n2Mo(i) − 3nMo(i) − 1)N(i) | (3) |
The ratio of Mo atoms on top rim edges RatioMo = ∑ntop(i)/∑nslab(i) | (4) |
Table 3 shows the total sulfur and nitrogen contents of the feed and the hydrotreated LCO products. The HDS activities of the three catalysts decrease in the order NiMo/F,B-Al(5.0) > NiMo/F,B-Al(7.0) > NiMo/F,B-Al(3.5). HDS is generally considered taking place in two parallel reaction pathways, direct desulfurization (DDS) and hydrogenation (HYD). The former is carried out by hydrogenolysis of the C–S bond and the latter by the hydrogenation of the aromatic ring followed by hydrogenolysis of the C–S bond.1,7 The acidity of a catalyst can enhance its ability in the cleavage of the C–S bond and therefore increase the HDS activity. NiMo/F,B-Al(5.0) has the best HDS performance among the three catalysts because of its high acidity and higher hydrogenation activity.
Feed | MoNi/F,B-Al(3.5) | MoNi/F,B-Al(5.0) | MoNi/F,B-Al(7.0) | |
---|---|---|---|---|
The cracking rate indicates the amounts in the products whose distillation temperature is lower than the initial boiling point of feed (133.5 °C). | ||||
N, ppmw | 495.6 | 14.57 | 19.07 | 23.72 |
S, ppmw | 13000 | 985 | 654 | 786 |
Cracking rate, wt% | — | 1.36 | 0.90 | 0.44 |
In addition to the hydrogenation ability and acidity, the synergetic effect between Ni and MoS2 may also play an important role in the NiMo/Al2O3 catalysts hydrotreating.34 It is seen that the three catalysts are effective in removing nitrogen-containing compounds. Catalyst NiMo/F,B-Al(3.5) demonstrates a slightly better HDN performance than the other two catalysts with 14.57 ppm N left in the hydrotreated product. The surface distribution of NiMo species might be responsible for the HDN ability of catalyst NiMo/F,B-Al(3.5) which has a relatively higher surface Ni/Al ratio and Mo/Al ratio than the other two catalysts (Table 2). The nickel atom on the catalyst surface can not only position at the edge of the active Ni–Mo–S phase but also develop the donor phase, Ni2S3, to the acceptor MoS2. The spillover hydrogen has been reported to demonstrate mobility between the separated Ni and Mo clusters.35,36 The mobility of spillover hydrogen among the active NiMoS phase and NiSx clusters can enhance the hydrogenation and HDN ability of catalyst NiMo/F,B-Al(3.5).
Fig. 8 presents the distributions of hydrocarbon types in the final products being treated with the catalysts. After hydrogenation, more than 80% of polyaromatics and over 50% of diaromatics in the feedstock were converted—2/3 partially hydrogenated to monoaromatics and 1/3 completely hydrogenated into saturated hydrocarbons. The hydrodearomatization activity decreased in the order NiMo/F,B-Al(7.0) > NiMo/F,B-Al(5.0) > NiMo/F,B-Al(3.5). The ring opening activity (saturates from the hydrogenation of the aromatics) of the three catalysts is 14.4%, 13.1% and 11.2% respectively, which is in proportion to the ratio of top rim Mo atoms in the MoS2 slabs. Addition of F, B to the alumina support leads to variations in the distribution of NiMo species on the catalyst surface which can further affect the morphology of the active NiMoS slabs, the Mo ratio at the top rims in the NiMoS slabs and hydrogenation activity of the catalysts.
Fig. 8 Aromatics and saturates hydrocarbon distribution on the NiMo/F,B-Al catalysts. |
Fig. 9 shows the distillation curves for the LCO feed and the products after the feed was hydrotreated with the F, B promoted NiMo catalysts. The cracking rate in the product can be obtained from the distillation curve, that is, the content at temperature < 133.5 °C (the initial boiling point of feed) in the product. For the NiMo/F,B-Al(3.5), NiMo/F,B-Al(5.0) and NiMo/F,B-Al(7.0) catalysts, the cracking rate is only 1.36%, 0.90% and 0.44% (listed in Table 3). The low cracking rate means that less gas products are produced and more middle distillate products can be obtained after LCO was hydrotreated. It is because F, B promoted NiMo/Al2O3 catalysts mainly formed Lewis acid sites, which contribute less to the cracking reaction.
Fig. 9 Distillation curve for the feed and the products after hydrotreating with the F, B promoted NiMo catalysts. |
In summary, for HDA, high hydrogenation ability of the catalyst is required. For HDN, high hydrogenation ability and suitable acidity are important for the hydrogenation of N-containing aromatic rings followed by the C–N bond cleavage. For HDS, high acidity combined with suitable hydrogenation activity is preferred, because HDS can be carried out either by hydrogenation of the aromatic ring followed by hydrogenolysis of the C–S bond or by directly hydrogenolysis of the C–S bond.
The overall effects of the addition of F and B are seen in the catalyst activity. The evaluation results reveal that the HDA activities of the three catalysts studied in this work decreased in the order NiMo/F,B-Al(7.0) > NiMo/F,B-Al(5.0) > NiMo/F,B-Al(3.5). The HDN activities decreased in the order NiMo/F,B-Al(3.5) > NiMo/F,B-Al(5.0) ≈ NiMo/F,B-Al(7.0) while HDS activity decreased in the order NiMo/F,B-Al(5.0) > NiMo/F,B-Al(7.0) > NiMo/F,B-Al(3.5).
n Mo(i) | The amount of Mo atoms on one side of the MoS2 slab |
n slab(i) | The amount of Mo atoms at slabs |
n top(i) | The amount of Mo atoms at top edges of the MoS2 slab |
DDS | Direct desulfurization |
DMDS | Dimethyl disulfide |
HDA | Hydrodearomatics |
HDN | Hydrodenitrogenation |
HDS | Hydrodesulfurization |
HYD | Hydrogenation |
LCO | Light cycle oil |
This journal is © The Royal Society of Chemistry 2012 |