JFUE 10110 No.

of Pages 10, Model 5G

20 February 2016

Fuel xxx (2016) xxx–xxx

1

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

5
6

3 Deactivation of hydrotreating catalyst by metals in resin and asphaltene

4 parts of heavy oil and residues
7 K. Kohli a, R. Prajapati a, S.K. Maity a,⇑, M. Sau b, M.O. Garg a
8 a
Indian Institute of Petroleum, Mohkampur, Dehradun, Uttarakhand 248005, India
9 b
Research & Development Centre Indian Oil Corporation, Faridabad 121001, India

11
10
12

1 4
h i g h l i g h t s g r a p h i c a l a b s t r a c t
215
4
25
16
Deactivation by metals in resin is less Comparison of HDS activity of spent (SC) and regenerated (RC) catalysts using LCO as feed.
17
27 adverse than that in asphaltene.
18
Nickel deactivates HDS active sites to
28
19 a lesser extent than that done by
20 vanadium.
21
The feed having high concentration of
22 wax materials can deactivate a
23 catalyst rapidly.

30
a r t i c l e i n f o a b s t r a c t
3
4 2
5
33 Article history: The degree of deactivation was studied on the hydrotreating catalyst by using four different feed stocks. 46
34 Received 29 September 2015 This work demonstrated that the deactivation by metals in resin and asphaltene parts of the heavy crude 47
35 Received in revised form 4 February 2016 oil and residues was different. The degree of deactivation by metals does not always depend on the con- 48
36 Accepted 9 February 2016
centration of these, but also the nature of the metals. Firstly, the metals in resin part may not have as 49
37 Available online xxxx
adverse effect on deactivation as the metals in asphaltene part have. It may possible that these metals 50
in resin may deposit on the bare alumina surface, not on the active sites. Secondly, the nickel containing 51
38 Keywords:
compounds in the feed deactivate the HDS active sites in lesser extent than that done by vanadium com- 52
39 Deactivation
40 Heavy oil and vacuum residue
pounds. The deactivation by heavy crude having high concentration of wax materials suggests that long 53
41 HDM paraffinic or naphthene/aromatic substituted paraffinic chains may take part in coke formation during 54
42 HDS hydroprocessing reaction conditions. This coke mostly covers the HDS active sites of the catalyst; but 55
43 LCO does not precipitate as solid coke. The formation of solid coke is observed for the feeds having high per- 56
44 centage of asphaltene content. 57
Ó 2016 Elsevier Ltd. All rights reserved. 58
59

60
62
61 1. Introduction goes for zero fuel oil configuration. Hydroprocessing of residues or 65
heavy oils is used to upgrade residues or heavy oils to the more 66
63 Residue conversion is becoming one of the important process in valuable products. However, the catalytic process i.e. hydropro- 67
64 the refinery. It becomes more essential when most of the refinery cessing of residue is not a simple process like hydroprocessing of 68
lighter fractions. Residues and heavy oils contain high concentra- 69

⇑ Corresponding author. tion of contaminants and these contaminants have an adverse 70

http://dx.doi.org/10.1016/j.fuel.2016.02.036
0016-2361/Ó 2016 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Kohli K et al. Deactivation of hydrotreating catalyst by metals in resin and asphaltene parts of heavy oil and residues.
Fuel (2016), http://dx.doi.org/10.1016/j.fuel.2016.02.036
 JFUE 10110 No. of Pages 10, Model 5G
20 February 2016

2 K. Kohli et al. / Fuel xxx (2016) xxx–xxx

71 effect on catalyst. These heavy ends have a high percentage of relative HDS activity of a regenerated catalyst was very low even 137
72 asphaltene. During hydroprocessing reactions at high temperature, when metals deposit on the catalyst below 7 wt%. 138
73 mostly these asphaltene molecules are converted into coke [1–3]. The activity of a spent catalyst can also regain its activity to 139
74 Moreover, residue and heavy oils also consist of metals, mainly some extent by leaching out the deposited metals from the spent 140
75 vanadium and nickel. These metals also deactivate the catalyst in catalyst. Oxalic acid in the presence of H2O2 was found to be the 141
76 due course of the reaction. Furthermore, the deactivation by metals best leaching agent where the spent catalyst (after removal of met- 142
77 is irreversible. It is reported in literature that the deactivation of als by leaching) showed almost the same HDS activity as a fresh 143
78 catalyst depends on the metal content in the feeds. The catalyst catalyst did [16]. 144
79 is more rapidly deactivated by the feed containing more metals The activity studies on the spent catalyst can give better idea 145
80 compared to the feed containing less metal [4–6]. how the catalyst is deactivated during hydroprocessing of heavy 146
81 In the recent past, the study on deactivation by coke and metals oil and residue. Though there are several studies on the deactiva- 147
82 during hydroprocessing of heavy crude and residue became as tion, however activity studies on the spent catalyst is very limited 148
83 important topic for researchers. In this context, the catalyst deacti- in the literature. Therefore in this investigation different feeds are 149
84 vation during hydrotreating of Maya heavy crude was extensively used to obtain the spent catalysts. The hydrodesulfurization activ- 150
85 studied by Maity et al. [7–10]. The results of activities of spent and ity of light cyclic oil (LCO) on these spent catalysts was investi- 151
86 regenerated catalysts were compared and it was established that gated. The same HDS activity was also compared with the 152
87 both coke and metals contributed to the loss of HDM activity, regenerated and fresh catalysts. The basic idea of this work is to 153
88 whereas the loss of HDS activity only occurred by metals deposi- investigate the effect of metals as a whole and the nature of metals 154
89 tion. It was also said that the mechanism for metal and sulfur on the catalyst deactivation. 155
90 removal did not follow the same course. It is well understood from
91 the literature that the principle cause for initial deactivation of the
2. Experimental 156
92 hydrotreating catalyst is rapid coke deposition and deactivation by
93 metals becomes prominent at the later stage of the reaction. It was
2.1. Preparation of catalyst 157
94 reported by Zeuthen et al. [11] that the HDS activity of benzothio-
95 phene was unchanged up to 4 wt% of coke deposition. The same
The NiMo catalyst supported on commercial alumina was pre- 158
96 HDS activity was then drastically decreasing with further increase
pared by the sequential incipient wetness impregnation method. 159
97 of coke deposits. Gualda and Katsztelan [5] reported that deactiva-
An appropriate amount of ammonium heptamolybdate (Alfa 160
98 tion by metals was prominent even at the initial stage of operation.
Aesar) and nickel nitrate (Alfa Aesar) salts were used to prepare 161
99 The degree of deactivation depends on the properties of feed, oper-
aqueous solutions. The prepared solutions were impregnated into 162
100 ating conditions and the type of catalyst used for this reaction.
the dry support. The molybdenum/nickel promoted catalysts were 163
101 Maity et al. [9] reported that alumina supported catalyst was deac-
dried in the presence of air at 110 °C for 6 h and calcined at 450 °C 164
102 tivated faster by coke deposition whereas both metal sulfides and
for 4 h. This catalyst contains 10 wt% of MoO3 and 3 wt% of NiO on 165
103 coke deposits were responsible for deactivation of alumina–titania
the catalyst basis. 166
104 and alumina–silica supported CoMo catalysts.
105 To better understand the deactivation process, the activity of a
106 spent catalyst was also studied by investigators. The spent catalyst 2.2. Characterisation of fresh and spent catalysts 167

107 deactivated by coke deposit can be reused by decoking of the spent

108 catalyst. However, the metal deposition on a spent catalyst is an BET specific surface area, pore volume and pore size distribu- 168

109 irreversible process and the spent catalyst in general is not used, tion of support and supported catalyst were measured by nitrogen 169

110 particularly after processing of a feed having high concentration adsorption–desorption analysis at 77K (BEL MAXSORP). Prior to 170

111 of metals. But the usefulness of the spent catalyst having high con- the analysis, the samples were degassed for 2 h at 300 °C. 171

112 taminants from the feed was reported by some researchers [12]. If TPR analysis (Micromeritics 2920) was carried out using 10% H2 172

113 the spent catalyst was decoked and mixed with the fresh boehmite, in argon. Reducing gas mixture was purified by molecular sieves. 173

114 this mixed catalyst showed not only high HDS activity, but also Before each run, the baseline was stabilized at room temperature 174

115 showed good HDM activity [12]. Some researcher also assumed by passing gas mixture at a flow of 25 mL/min. After stabilization 175

116 that at high concentration of V in spent catalyst, V–Mo–S phase the sample was heated at 10 °C/min from room temperature to 176

117 was formed and it governed the catalyst activity [13]. 1000 °C. 177

118 The effect of vanadium poisoning on thiophene HDS activity X-ray diffractogram of the fresh catalyst NiMo/Al2O3 were 178

119 was studied by Ledoux and Hantzer [14]. A series of catalysts were recorded on a Bruker AXZ GMBH using a Cu Ka radiation 179

120 prepared by wetness impregnation of a water solution of ammo- (k = 0.15418 nm). 180

121 nium vanadate. The oxide form of NiMo catalysts was used for Weight loss due to combustion of the spent catalysts has been 181

122 the impregnation of vanadate. In this method authors only impreg- measured by thermogravimetric (Perkin Elmer TGA4000) analysis. 182

123 nated around maximum 10 wt% of vanadium. It was reported that For this analysis, 10 mg of the sample was placed in a platinum 183

124 HDS activity decreased with increasing concentration of vanadium crucible and heated from ambient temperature to 900 °C at a rate 184

125 on the catalyst. The fresh catalyst (0 wt% V) had very high HDS of 10 °C/min in the presence of air (flow rate 50 mL/min). The sur- 185

126 activity. In impregnation of only 0.87 wt% of V, HDS activity dras- face morphology and composition of the spent catalysts were also 186

127 tically decreased to almost less than a half. Different percentage studied by SEM-EDX (Quanta200F). 187

128 of vanadium was impregnated into the hydrotreating catalysts

129 and their HDM and HDS activities were studied by using Maya 2.3. Catalyst activation and activity studies 188
130 crude as feed [8]. It was noticed that the rate of HDS deactivation
131 was faster than that of HDM activity. It was concluded that the fas- To evaluate the catalytic behaviour with different industrial 189
132 ter deactivation of HDS activity was due to the coverage of active feedstocks, the catalyst was tested in the high pressure batch reac- 190
133 sites by impregnated vanadium. Koyama et al. [15] also reported tor. Five grams of a fresh catalyst was sulfided ex-situ for each 191
134 the adverse effect of metal deposit on the hydrotreating catalyst. experiment. An atmospheric unit was used for this sulfidation. In 192
135 The relative HDS activities of dibenzothiophene were investigated this unit, hydrogen was passed through a container having CS2, 193
136 on the spent and regenerated catalysts. It was found that the and then the saturated mixture of CS2 with hydrogen was passed 194

Please cite this article in press as: Kohli K et al. Deactivation of hydrotreating catalyst by metals in resin and asphaltene parts of heavy oil and residues.
Fuel (2016), http://dx.doi.org/10.1016/j.fuel.2016.02.036
 JFUE 10110 No. of Pages 10, Model 5G
20 February 2016

K. Kohli et al. / Fuel xxx (2016) xxx–xxx 3

195 through the reactor. The sulfiding conditions are as follows: tem- [17]. For this calibration, the TGA evaporation temperature of pure 249
196 perature – 400 °C, pressure – atmospheric and duration of sulfida- compounds (Naphthalene: 218 °C, Fluorene: 294 °C, Phenan- 250
197 tion – 4 h. threne: 340 °C, Pyrene: 390 °C and Dibenzo(a)anthracene: 251
198 For the activity study, 200 g of feed was taken into the batch 524 °C) are plotted against their normal boiling temperatures. 252
199 reactor (1L capacity). The sulfided catalyst was transferred into For an actual case, the TGA evaporation temperatures of the feed 253
200 the reactor under N2 atmosphere very quickly so that the catalyst and products are converted to their corresponding normal boiling 254
201 would not remain in contact with air for a long time. The reactor temperature from the calibration curve. 255
202 was sealed properly and checked for leakage. An appropriate
203 amount of hydrogen was added into the reactor so that the final
204 pressure can reached to 100 kg/cm2 at reaction temperature. The 3. Results & discussion 256
205 heating was started from room temperature to the required tem-
206 perature at a rate of 3 °C/min. Stirring was started when the tem- 3.1. Properties of fresh catalyst 257
207 perature reached at the required reaction temperature (410 °C) and
208 the time at this point was noted as the beginning of the reaction. The physical properties of the supported NiMo catalyst are: sur- 258
209 Several tests (without catalyst) were performed by using feed face area of 239 m2/g, pore volume 0.8 mL/g and average pore 259
210 VR-I at different reaction temperatures (400, 410 and 420 °C). It diameter 13.2 nm. The commercial alumina used as support is 260
211 was noted that with increasing temperature (at moderate H2 pres- cylindrical in shape having 1.9 mm dia. Adsorption and desorption 261
212 sure of 100 kg/cm2), the formation of carbonaceous material (solid) isotherms of alumina supported NiMo catalyst are presented in 262
213 was high. Therefore, moderate temperature (410 °C) was chosen Fig. 1. The textural properties ascertain that the catalyst used in 263
214 for this study. this study has mesoporous structure. 264
215 Solid and liquid products were separated from the catalyst after XRD diffraction pattern (Fig. 2) of the catalyst indicates that 265
216 the reaction. The metals, sulfur and asphaltene contents of the liq- active metal molybdenum and promoter are well dispersed into 266
217 uid products were analysed. The spent catalysts were washed with the alumina support. The absence of peaks at 2h = 25–27° showed 267
218 hot toluene by the Soxhlet process and dried at 110 °C for 4 h amorphous phase of molybdenum oxide. Only characteristics 268
219 before analysis. peaks of gamma alumina appear in the diffractogram. 269
220 To study the hydrodesulfurization of light cyclic oil (LCO), two
221 grams of spent catalyst was sulfided ex-situ for each experiment
222 according to the sulfidation procedure as described above. Two
223 hundred grams of LCO were taken into the reactor vessel (1 L
224 capacity). The sulfided catalyst was then transferred into the reac-
225 tor in a nitrogen atmosphere quickly. Hydrotreated products of
226 LCO were collected every hour and its sulfur content was analyzed.
227 The experimental conditions for both hydroprocessing of residues
228 and HDS of LCO are given in Table 1. The properties of LCO used
229 in this work are: API gravity of 29.38 and sulfur of 0.895 wt%.
230 The HDS activity of LCO was also studied with the regenerated
231 catalysts. Two grams of the spent catalyst was regenerated in a
232 flow of air at 550 °C for 4 h in atmospheric unit. After regeneration,
233 the catalyst was again sulfided and its activity was tested.

234 2.4. Analysis of feed and products

235 ASTM-D4124-09 was used for SARA analysis of the feeds.

236 Metals (Ni, V) of the feeds and products were analysed by using
237 ICP-AES (Leeman Model DRE, PS-3000UV). The total sulfur content
238 was analysed with an Oxford sulfur analyzer (Lab-X 3500) using
239 X-ray fluorescence. The elements like carbon, nitrogen, and hydro-
240 gen were also analysed using an Elemental Vario Micro CHNS ana- Fig. 1. N2-adsorption/desorption isotherms and pore size distribution (inset) of
241 lyzer. The density and kinematic viscosity of the feeds as well as NiMo/Al2O3.
242 liquid products were measured by Viscometer SVM 3000 (Anton
243 Paar). Asphaltene was measured by ASTM-D3279 method. Sulfur
244 distribution of LCO and hydrotreated products were measured by
245 GC using PFPD as sulphur specific detector.
246 Boiling point distribution of the feed and the products were
247 determined by thermogravimetric analysis (TGA). For these exper-
248 iments, a calibration curve was established as shown elsewhere

Table 1
Reaction conditions used for vacuum residue (VR) and light cyclic oil (LCO).

Conditions VR LCO
Catalyst weight (g) 5 2
Feed weight (g) 200 200
Temperature (°C) 410 360
Pressure (kg/cm2) 100 60
Duration (h) 6 6
Fig. 2. XRD diffractogram of NiMo-catalyst.

Please cite this article in press as: Kohli K et al. Deactivation of hydrotreating catalyst by metals in resin and asphaltene parts of heavy oil and residues.
Fuel (2016), http://dx.doi.org/10.1016/j.fuel.2016.02.036
 JFUE 10110 No. of Pages 10, Model 5G
20 February 2016

4 K. Kohli et al. / Fuel xxx (2016) xxx–xxx

270 Temperature programmed reduction (TPR) experiment was also only 0.27 wt% in HC. Since this asphaltene is known as coke precur- 302
271 carried out on the fresh catalyst to investigate metal-support inter- sor, the upgradation of these feeds is challenging. The idea behind 303
272 action and the nature of the active phases. The reduction profile of taking different kind of feeds is to investigate the effect of physical 304
273 NiMo/Al2O3 catalyst with temperature is presented in Fig. 3. Two properties of the feeds on the catalyst’s deactivation. Moreover, in 305
274 well separated reduction peaks appeared. The first Tm (reduction this study, three different vacuum residues are taken having differ- 306
275 at maximum temperature) appears at 400 °C and second one at ent percentage of contaminants. From the experimental results for 307
276 800 °C. The nature of the second reduction peak is very broad com- determination of molecular weight by HPLC, it was noticed that the 308
277 pared to the first one. This broadness indicates the complex beha- nature of asphaltene present is these feed are not similar type. 309
278 viour of this reduction peak. It is well discussed in the literature Therefore investigation is needed to see the deactivation trend 310
279 that there are two types of active molybdenum species present for the feed having not only different concentration of asphaltene 311
280 in the catalyst. The tetrahedral Mo-species are well dispersed but also its nature. The nature of asphaltene on deactivation rate 312
281 and are strongly bonded with alumina support. Therefore, the is also reported in the literature [19,20]. 313
282 higher temperature reduction peak may attribute to the reduction
283 of the strongly bonded tetrahedral molybdenum. On the other 3.3. Hydrotreating activities of fresh catalyst 314
284 hand, the active species of octahedral molybdenum are also pre-
285 sent and it appears at lower temperature. Due to the various layers The hydrotreating activities of the catalyst were studied for dif- 315
286 of these species, these are weakly bonded and hence easily reduced ferent feeds. In Fig. 4, the hydrodemetallization (HDM), 316
287 (or sulfided) at lower temperature. Other possible explanation for hydrodesulfurization (HDS) and hydrodeasphaltene (HDAs) activi- 317
288 appearance of two peaks is that the 1st peak is due to the reduction ties of NiMo/Al2O3 catalyst are presented. These results were 318
289 of MoO3 to MoO2 and 2nd one is due to the reduction of MoO2 to obtained in the batch reactor after 6 h of reaction. It is also worth 319
290 Mo [18]. to mention that total metals, sulfur and asphaltene content are 320
measured on the liquid products only. The HDM activity of this 321
291 3.2. Properties of feeds particular catalyst varies from 84% to 97%. It is observed that the 322
HDM activity is higher than that of HDS activity. During hydropro- 323
292 The feedstocks used for the present study were obtained from cessing reaction in batch reactor, H2S is produced and it is accumu- 324
293 different Indian refineries. The physico-chemical properties of the lated inside the reactor. The concentration of H2S increases with 325
294 feeds are given in Table 2. The origin of the heavy crude oil (HC) reaction time at the expense of reactive hydrogen. The higher par- 326
295 is from Mangala, Rajasthan (India). The vacuum residues used for tial pressure of H2S has a beneficial effect on HDM but it has 327
296 this study are obtained from vacuum distillation of different mixed adverse effect on HDS. Moreover higher reaction temperature is 328
297 crudes. The vacuum residues (VR) have large heteroatom content more favourable for HDM reaction than HDS. 329
298 compared to heavy crude (HC). VR-I contains 240 wppm of V The sulfur distributions in the liquid products were measured 330
299 whereas it is only 2.5 wppm in HC. Vanadium content is more than by GC-PFPD. The hydrotreating product was fractionated into three 331
300 nickel in all VR feeds as compared to HC. Asphaltene content in VR categories i.e. product below 150 °C, product in the range of 332
301 (s) is very high and it is around 22 wt% in VR-I feed whereas it is 150–350 °C and product above 350 °C. The sulfur distributions of 333
these three different fractions are presented in Fig. 5. The same sul- 334
fur distribution of the feed was intended to measure but it was not 335
possible due to the presence of high percentage of asphaltene. 336
Therefore, the hydrotreated product was taken for this analysis. 337
The results presented in figure show that the sulfur compounds 338
in the products are distributed throughout the range and mainly 339
these are concentrated in lower and higher boiling ranges com- 340
pared to that of 150–350 °C range. 341
Removal of asphaltene has also been measured and the results 342
are given in Fig. 4. In general the conversion of asphaltene follows 343
according to the asphaltene present in the feed. It is also noticed 344

Fig. 3. H2-TPR profile of NiMo-catalyst.

Table 2
Properties of four different feeds used for this study.

Properties Heavy crude VR-I VR-II VR-III

Density (g/mL) at 15.6 °C 0.8663 1.055 1.012 1.033
Kvis (cSt) at 100 °C 13.72 25,794 1872 2840
Sulfur (wt%) 0.26 5.45 4.75 5.45
Ni (wppm) 99.2 65 34.2 30.5
V (wppm) 2.5 240 84.5 117.5
Ni + V (wppm) 101.7 305 118.7 148.0
H/C (atomic) 1.59 1.07 1.05 1.64
Saturates (wt%) 32.65 8.89 14.18 28.00
Aromatics (wt%) 45.65 25.88 49.02 45.30
Resins (wt%) 21.43 43.25 23.50 16.47
Asphaltene (wt%) 0.27 21.98 13.3 9.40
Ni + V (wppm, in DAOa) 91 149 73.1 117.2
a
DAO = deasphalted oil. Fig. 4. Hydrotreating activities of NiMo/Al2O3 catalyst for different feeds.

Please cite this article in press as: Kohli K et al. Deactivation of hydrotreating catalyst by metals in resin and asphaltene parts of heavy oil and residues.
Fuel (2016), http://dx.doi.org/10.1016/j.fuel.2016.02.036
 JFUE 10110 No. of Pages 10, Model 5G
20 February 2016

K. Kohli et al. / Fuel xxx (2016) xxx–xxx 5

<150OC

150-350OC

>350OC

RT, min

Fig. 5. GC-PFPD chromatography of sulfur distribution in hydrotreated product of VR-II.

Table 3
Properties of hydrotreating products and amount of solid coke.

Properties Heavy Crude VR-I VR-II VR-III

Density (g/mL) at 15.6 °C 0.862 0.9166 0.9182 0.9049
Kvis (cSt) at 100 °C 4.9149 2.1849 2.3819 1.6560
H/C of HDT liquid (atomic) 1.96 2.38 1.79 2.22
Cokea (wt%) 0 19.1 12.19 14.42
H/C of spent Cat. (atomic)b 1.87 0.80 1.01 1.02
a
Solid coke separated from liquid product after reaction.
b
H/C ratio measured on used catalysts after reaction and wash.

oil most of the 550 °C+ materials are converted to diesel and VGO. 362
Therefore, it is concluded from these results that most of the resi- 363
due part in the VR(s) is preferably converted into the diesel 364
Fig. 6. Hydrocracking activities of different feeds (temperature – 410 °C, pressure – whereas the same part in HC is converted into both diesel and 365
100 kg/cm2, and duration – 6 h). (F – Feed, HDT – Hydrotreated liquid product) VGO. 366
Density and viscosities of the liquid products are measured and 367
the results are reported in Table 3. The Kvis is substantially 368
345 that the asphaltene conversion is lower than that of HDM and the
improved by hydrotreating reaction, irrespective of the amount 369
346 same was also observed by other researchers [17]. It indicates that
of sulfur and asphaltene content in the feeds. Hydrogen to carbon 370
347 the cracking of asphaltene is relatively tougher than that of
atomic ratio given in Tables 2 and 3 indicates that the feed VR-I is 371
348 hydrodemetallization. It also proved that metals are not only asso-
hydrogenated more during the reaction and that may be the reason 372
349 ciated with asphaltene moiety but also in resin part. The removal
for showing higher hydrotreating activities. 373
350 of metal from these resins is easier than that of asphaltene.
351 The hydrocracking activity of the catalyst was also studied. The
352 activity results are reported in Fig. 6. In this figure residue material 3.4. Characterization of spent catalysts 374
353 (>550 °C) is given in negative y-axis so that it is easily understood
354 how much materials of this range is converted into the lighter frac- TGA analysis of the spent catalysts is carried to know the nature 375
355 tions (given in positive y axis). The 550 °C+ material content in the of coke deposited on the catalyst during hydrotreating of heavy 376
356 VR is very high and it is as high as 93 wt% in VR-I. Most of this crude oil and residues. The weight loss and its derivatives with 377
357 material is converted into the lighter fractions in the range of die- temperature are reported in Fig. 7. All the derivative curves show 378
358 sel and VGO, particularly in diesel range (150–350 °C). Same trend two principal weight losses around 100 °C and 480–520 °C. The 379
359 is observed for all vacuum residues studied in this case. The max- second peak around 450 °C is very sharp and prominent. The 380
360 imum increment of diesel product is obtained in the VR-III hydro- weight loss at first peak is for the loss of water or lighter hydrocar- 381
361 treated product which is around 45 wt%. In the case of heavy crude bon trap inside the pore cavity of the spent catalyst. And the 382

Please cite this article in press as: Kohli K et al. Deactivation of hydrotreating catalyst by metals in resin and asphaltene parts of heavy oil and residues.
Fuel (2016), http://dx.doi.org/10.1016/j.fuel.2016.02.036
 JFUE 10110 No. of Pages 10, Model 5G
20 February 2016

6 K. Kohli et al. / Fuel xxx (2016) xxx–xxx

HC

VR-I

VR-II
VR-III

Fig. 7. Weight loss (solid line) and its derivative (dotted line) of the spent catalysts processed with different feeds.

383 second peak is identified as the weight loss due to coke burning. the catalyst surface and in the due course of the reaction; these 402
384 The peak position for the spent catalyst SC-HC is very near to metals enter into the pore cavity and block some of the active sites. 403
385 450 °C whereas this peak has been shifted to higher temperature It deactivates the catalyst and the degree of deactivation depends 404
386 for the spent catalysts treated with VR(s). It is 509, 513 and on the concentration of metals in the feed. The pore structure of 405
387 509 °C for the SC-VR-I, SC-VR-II, & SC-VR-III respectively. Begon the catalyst is also very important factor on how metals are depos- 406
388 et al. [21] observed three different regions during combustion of ited on the surface [22]. If the pore diameter is not sufficiently high 407
389 spent hydrotreating catalyst. The region I is <180 °C, region II is enough to enter the metal bearing compound then, the metals are 408
390 in between 180 and 330 °C and region III is in between 330 and deposited mainly on the periphery of the surface pore and gave M 409
391 750 °C. The first region is due to loss of water, whereas the region type of shape. 410
392 II is due to the weight loss of mobile carbonaceous residues which
393 is defined as soft coke. The bulky hard coke is combusted at higher 3.5. Hydrodesulfurization activity of the used catalysts 411
394 temperature and that appears at region III.
395 Profiles of nickel and vanadium deposits on the spent catalysts It was discussed above that one NiMo/Al2O3 catalyst was used 412
396 were also studied by SEM and presented in Fig. 8. Slightly higher for hydroprocessing of heavy crude oil and three different residues. 413
397 concentration of metals is deposited in the centre of the catalyst After the reaction, the used catalysts were separated from the liq- 414
398 particle. The profile of metal deposits indicates that the catalyst uid product and were washed by Soxhlet process. These spent cat- 415
399 pore mouth is sufficiently large for vanadium or nickel bearing alysts were used to study hydrodesulfurization of light cyclic oil 416
400 compounds in feeds to enter, and hence pore mouth plugging is (LCO) to investigate the nature of deactivation depending on the 417
401 not observed. The metals coming from the feed are deposited on feed stock properties. The HDS conversion of this spent catalyst 418

Please cite this article in press as: Kohli K et al. Deactivation of hydrotreating catalyst by metals in resin and asphaltene parts of heavy oil and residues.
Fuel (2016), http://dx.doi.org/10.1016/j.fuel.2016.02.036
 JFUE 10110 No. of Pages 10, Model 5G
20 February 2016

K. Kohli et al. / Fuel xxx (2016) xxx–xxx 7

Fig. 10. Light cycle oil HDS of fresh and regenerated catalysts (T = 60 °C, P = 0 kg/
cm2, and RC = regenerated catalyst).

relative HDS activity (HDS of reg. or spent cat/HDS of fresh cat.) is 437
given. The relative HDS activity of RC-HC is almost one, indicating 438
that the regenerated catalyst almost regains its original activity by 439
burning of deposited coke. However, the same relative activity of 440
SC-VR(I) and RC-VR(I) are same. It means that the deactivation of 441
this catalyst mainly happens due to the metals deposit. On the 442
other hand, both coke and metals deposits are the causes for deac- 443
tivation when the catalyst is treated with VR(II) and VR(III) feeds. 444
Therefore the above results indicate that metals either block the 445
Fig. 8. Radial distribution of nickel and vanadium deposits on the spent catalysts. pore mouth or can cover the active sites of spent catalysts. In this 446
case, the blockage of pore mouth can be discarded. SEM results 447
suggest that metals are not concentrated on the periphery of the 448
419 is plotted against contact time and it is presented in Fig. 9. The con- catalyst particle but enter deeply into the catalyst particle. There- 449
420 tact time for this experiment is calculated according to the follow- fore, the main reason for the deactivation by metals is the coverage 450
421 ing equation [5]. of HDS-active sites. Sometime higher activity of regenerated cata- 451
422
ðCatalyst weightÞðtotal reaction timeÞ lyst was also observed by researcher [23]. The cumene hydrocrack- 452
Contact time ¼ ing activity of regenerated catalyst was higher than that of fresh 453
424 Feed weight
catalyst and it was reported that it was due to the creation of 454
425 The HDS activity increases with contact time and which is nor- Brønsted acidity during regeneration of spent catalyst at elevated 455
426 mal in general. In this figure the HDS activity of the fresh catalyst temperature. 456
427 has also been compared with that of the spent catalyst. The HDS of It is noticed from Fig. 11, that the HDS activity of SC-HC catalyst 457
428 LCO was also studied with the regenerated catalysts and the results is almost in similar magnitude as SC-VR(s); though, the asphaltene 458
429 are shown in Fig. 10. HDS activity of LCO shows that the spent cat- content (coke precursor) in HC is very low compared that of in VR 459
430 alysts have lost activity during hydrotreating of residues. The cat- (s). The coke formation due to the presence of high asphaltene can 460
431 alyst processed with the vacuum residues is deactivated more and explain the deactivation of spent catalysts of VR feeds. But the 461
432 hence it loses more activity. However, when the catalyst is regen- same explanation is not convincing in case of HC feed having only 462
433 erated activity regains. This indicates that when deposited coke is 0.27 wt% of asphaltene. The nature of the heavy crude can explain 463
434 burnt off, the loss of activity due to the coke deposition is recov- this deactivation behaviour. From our detail analysis of HC feed, it 464
435 ered. The HDS activities of the spent and the regenerated catalysts is found that this feed contains very high percentage of wax 465
436 at 0.06 h of contact time are compared in Fig. 11. In this figure, the

Fig. 9. Light cycle oil HDS of fresh and spent catalysts (T = 60 °C, P = 0 kg/cm2, Fig. 11. Comparison of HDS activity of spent (SC) and regenerated (RC) catalysts
FC = fresh catalyst, SC = spent catalyst, HC = heavy crude, and VR = vacuum residue). using LCO as feed.

Please cite this article in press as: Kohli K et al. Deactivation of hydrotreating catalyst by metals in resin and asphaltene parts of heavy oil and residues.
Fuel (2016), http://dx.doi.org/10.1016/j.fuel.2016.02.036
 JFUE 10110 No. of Pages 10, Model 5G
20 February 2016

8 K. Kohli et al. / Fuel xxx (2016) xxx–xxx

466 material and it is around 30 wt% of the total feed. The boiling point fresh, spent and regenerated catalysts. Almost all benzothiophenic 488
467 of this wax material is in VGO range. Fig. 6 also supports that a sub- sulfur compounds are removed by hydrotreatment of LCO with the 489
468 stantial amount of materials are in VGO range. This wax material fresh catalyst. It suggests that the hydrodesulfurization active sites 490
469 may consist of saturated (normal and iso-alkanes) naphthenes presents in the NiMo/Al2O3 catalyst can effectively removed ben- 491
470 and alkyl or naphthene substituted aromatic compounds (Table 2). zothiophenic sulfur compounds. However, some of these sites are 492
471 During hydrotreating conditions, mainly at high temperature, covered by coke deposit and that’s why, sulfur compounds in this 493
472 these wax materials are converted to coke by polymerization of region once again appear when this LCO is hydrotreated by the 494
473 carbon free radicals. Mostly this coke either covers pore mouth spent catalyst (SC-HC). Those sulfur compounds once again disap- 495
474 or blocks the HDS active sites; but not precipitates as solid material pear when LCO is hydrotreated by the regenerated catalyst. It 496
475 (Table 3). means that the active sites (HDS) covered by coke are once again 497
476 To look more deeply inside the deactivation during hydrotreat- uncovered by burning of coke during regeneration process. It also 498
477 ing of heavy oil and residues, the sulfur distribution of LCO and suggest that in this case (feed HC), metals bearing compounds do 499
478 hydrotreated products of LCO by spent and regenerated catalysts not affect much on the catalytic active sites. 500
479 have been studied by GC analysis and results are given in In Fig. 12b, the similar comparison is performed for the spent 501
480 Figs. 12a and 12b. For comparison, the sulfur distributions of and regenerated catalysts obtained after hydrotreatment of VR-I 502
481 untreated LCO and hydrotreated product of LCO by fresh catalyst feed. The sulfur distributions in benzothiophenic range are almost 503
482 are also given in this figure. The sulfur distributions of two batches same for both hydrotreated products by SC-VR-I and RC-VR-I cata- 504
483 i.e. HC and VR-I are only given in Figs. 12a and 12b respectively. lysts. When sulphur distributions of HDT product of LCO with 505
484 Most of the sulfur compounds present in LCO are benzothiophene regenerated Cat-HC and HDT product with regenerated Cat-VR-I 506
485 (BT) and its derivatives and dibenzothiophene (DBT) and its deriva- is compared side by side, the differences are clearly observed. It 507
486 tives. Comparing the sulfur distribution of LCO feed, the benzothio- indicates that after burning of deposited coke, the catalyst RC- 508
487 phenic sulfur compounds are mostly affected by hydrotreating by VR-I does not regain its activity as shown in Fig. 11. Here in this 509

HDT Product of LCO

with regenerated Cat
–HC

HDT Product of
LCO with Spent
Cat–HC

HDT Product of
LCO withFresh
Catalyst

LCO-FEED

RT, min

Fig. 12a. GC-PFPD chromatography of sulfur distributions for LCO feed and hydrotreated products of LCO by fresh, spent (SC-HC) and regenerated catalyst (RC-HC).

Please cite this article in press as: Kohli K et al. Deactivation of hydrotreating catalyst by metals in resin and asphaltene parts of heavy oil and residues.
Fuel (2016), http://dx.doi.org/10.1016/j.fuel.2016.02.036
 JFUE 10110 No. of Pages 10, Model 5G
20 February 2016

K. Kohli et al. / Fuel xxx (2016) xxx–xxx 9

HDT Product of LCO

with regenerated Cat
–VR-I

HDT Product of
LCO with Spent
Cat–VR-I

HDT Product of
LCO with Fresh
Catalyst

LCO-FEED

RT, min

Fig. 12b. GC-PFPD chromatography of sulfur distributions for LCO feed and hydrotreated products of LCO by fresh, spent (SC-VR-I) and regenerated catalyst (RC-VR-I).

510 point, it is important to mention that sulfur distribution measured only 0.27 wt% of asphaltene, but it has 99.2 wppm of nickel and 531
511 by GC is qualitative. Total sulfur measured by XRF is slightly differ- 2.5 wppm of vanadium. The deasphalted oil contains 91 wppm of 532
512 ent from GC measurement. The results obtained for other two metals (Table 2). It suggests that most of the metals in HC feed 533
513 feeds (VR-II and VR-III) are in between the results of HC and VR- are in phorphyrin structure and are associated in resin part of 534
514 I, i.e. the total sulfur concentration in benzothiophenic region is the crude. On the contrary almost half of the total metals in vac- 535
515 lesser for regenerated catalysts compared that of the correspond- uum residues (VR-I and VR-II) are in asphaltene part. 536
516 ing spent catalyst (results are not given). It suggests that after Knozinger et al. [25,26] reported interesting results regarding 537
517 burning of coke the spent catalysts gain HDS activity to some interaction of metals (V and Ni) porphyrins with sulfided 538
518 extent for other two cases i.e. VR-II and VR-III. Fig. 11 shows that hydrotreating catalyst. Vanadium porphyrin is bound with catalyst 539
519 gaining of HDS activity is slightly higher for the RC-VR-II compared surface by donor–acceptor interaction where V4+ acts as donor 540
520 to that of RC-VR-III. (Lewis base) sites and bare surface of support acts as acceptor 541
521 During hydroprocessing reactions, the catalyst deactivation (Lewis acid) sites. It suggests that vanadium porphyrin adsorbs 542
522 occurs by both coke and metals deposits. At initial stage, coke on the surface, not on the active sites. On the other hand, vanadium 543
523 forms rapidly whereas the metals are continuously deposited on or nickel associated with asphaltene is deposited on the catalyst as 544
524 the catalyst surface. Metals in the crude are in the form of por- metal-sulfides and cover the active sites or plug the pore mouth. In 545
525 phyrin and non porphyrins structures [1,4,24]. The metal bearing this study, two types of feed are used. One is heavy crude where 546
526 porphyrin structures are well characterised whereas non por- most of the metals are associated with resin part and other is vac- 547
527 phyrin structures are less characterised which may be associated uum residues where these metals are linked with asphaltene. From 548
528 with nitrogen, oxygen and sulfur in the defect centres of asphal- our GC-PFPD results, it is found that the deactivation of the catalyst 549
529 tene moiety. It is proposed that this structure is more stable than (for HC feed) mostly occurs by coke deposit whereas the deactiva- 550
530 porphyrins structure. The heavy crude used in this study, contains tion by metals is clearly observed for VR-I feed. SEM results 551

Please cite this article in press as: Kohli K et al. Deactivation of hydrotreating catalyst by metals in resin and asphaltene parts of heavy oil and residues.
Fuel (2016), http://dx.doi.org/10.1016/j.fuel.2016.02.036
 JFUE 10110 No. of Pages 10, Model 5G
20 February 2016

10 K. Kohli et al. / Fuel xxx (2016) xxx–xxx

552 support that the metals deeply enter into the catalyst particle and References 599
553 therefore it is possible that the metals coming from asphaltene in
554 VR(s) are deposited as metals sulfides on the catalyst and cover [1] Furimsky E, Massoth FE. Deactivation of hydroprocessing catalysts. Catal Today 600
1999;52:381–495. 601
555 the active sites (HDS). Vanadium sulfide can decorate the edge of [2] Wiwel P, Zeuthen P, Jacobsen AC. In: Bartholomew CH, Butt JB, editors. Catalyst 602
556 a MoS2 slab just as nickel or cobalt does. But vanadium sulfide dis- deactivation, vol. 68. Amsterdam: Elsevier; 1991. p. 257. 603
557 places Ni from the edge site of MoS2 slab and becomes a source of [3] Absi-Halabi M, Stanislaus A, Trimm DL. Coke formation on catalysts during the 604
hydroprocessing of heavy oils. Appl Catal 1991;72:193–215. 605
558 deactivation, since the promotional effect of vanadium is less than [4] Mitchel PCH. Hydrodemetallisation of crude petroleum: fundamental studies. 606
559 that of nickel. Whereas the nickel in the resin of HC feed may Catal Today 1990;7:439–45. 607
560 deposit on the bare alumina surface and hence the initial deactiva- [5] Gualda G, Kastelan S. Initial deactivation of residue hydrodemetallization 608
catalysts. J Catal 1996;161:319–37. 609
561 tion only occurs by coke formation. It is also supported the results 610
[6] Oyekunle L, Hughes R. Catalysts deactivation during hydrodemetallization. Ind
562 found in the NiMo/Al2O3 catalyst treated with VR-III feed. Fig. 11 Eng Chem Res 1987;26:1945–50. 611
563 shows that this catalyst is also deactivated by metals though a con- [7] Maity SK, Pérez VH, Ancheyta J, Rana MS, Centeno G. Effect of asphaltene 612
contained in feed on deactivation of Maya crude hydrotreating catalyst. Petr 613
564 siderable amount of metals (79 wt%) in the feed VR-III are in resin 614
Sci Tech 2007;25:241–9.
565 part. But the vanadium content in this feed is higher than the [8] Maity SK, Ancheyta J, Alonso F, Vazquez JA. Study of accelerated deactivation of 615
566 nickel. However, the deactivation by coke deposit is also very high hydrotreating catalysts by vanadium impregnation method. Catal Today 616
2008;130:405–10. 617
567 though this feed contains very lower percentage of asphaltene. It
[9] Maity SK, Blanco E, Ancheyta J, Alonso F, Fukuyama H. Early stage deactivation 618
568 leaves a space for further detail investigation for this kind of feed of heavy crude oil hydroprocessing catalysts. Fuel 2012;100:17–23. 619
569 which has very high concentration of wax material. [10] Maity SK, Ancheyta J, Alonso F, Rayo P. Hydrodesulfurization activity of used 620
hydrotreating catalysts. Fuel Process Technol 2013;106:453–9. 621
[11] Zeuthen P, Bartholdy J, Wiwel P, Cooper BH. In: Delmon B, Froment GF, editors. 622
570 4. Conclusions Catalyst deactivation. Amsterdam: Elsevier; 1994. p. 199. 623
[12] Marafi M, Al-Sheeha H, Al-Omani S, Al-Barood A. Activity of hydroprocessing 624
catalysts prepared by reprocessing spent catalysts. Fuel Process Technol 625
571 The deactivation of NiMo/Al2O3 catalyst is investigated by using 626
2009;90:264–9.
572 different types of crudes having different concentration of asphal- [13] Dejonghe S, Hubaut R, Grimblot J, Bonelle JP, de Courieres T, Faure D. 627
573 tene and metals. The spent catalysts obtained after hydropropro- Hydrodemetallation of a vanadylporphyrin over sulfided NiMo/cAl2O3, Mo/ 628
574 cessing of these feeds are used for HDS of LCO. This HDS activity cAl2O3, and cAl2O3 catalysts-effect of the vanadium deposit on the toluene 629
hydrogenation. Catal Today 1990;7:569–85. 630
575 is also compared with the corresponding regenerated and fresh [14] Ledoux MJ, Hantzer S. Hydrotreatment catalyst poisoning by vanadium and 631
576 catalysts. This investigation demonstrated that the deactivation nickel porphyrin: ESR and NMR. Catal Today 1990;7:479–96. 632
[15] Koyama H, Nagai E, Kumagai H. Am Chem Soc Div Petrol Chem (Preprints) 633
577 by metals in resin and asphaltene parts of the heavy crude oil
1995;40:446–9. 634
578 and residues was different. The degree of deactivation by metals [16] Menoufy MF, Ahmed HS. Treatment and reuse of spent hydrotreating catalysts. 635
579 does not always depend on the concentration of these, but also Energy Sour, Part A 2008;30:1213–22. 636
580 the nature of the metals. Firstly, the metals having in resin part [17] Millan M, Adell C, Hinojosa C, Herod AA, Dugwell D, Kandiyoti R. Comparison 637
of fractionation methods for the structural characterization of petroleum 638
581 may not have as adverse effect on deactivation as the metals in residues. Energy Fuels 2007;21:1370–8. 639
582 asphaltene part have. It may possible that these metals (Ni and [18] Maity SK, Ancheyta J, Soberanis L, Alonso F, Llanos ME. Alumina–titania binary 640
583 V) in resin may deposit on the bare alumina surface not on the mixed oxide used as support of catalysts for hydrotreating of Maya heavy 641
crude. Appl Catal A Gen 2003;244:141–53. 642
584 active sites of the catalyst. Secondly, the nickel containing com- 643
[19] Vogelaar BM, Eijsbouts S, Bergwerff JA, Heiszwolf JJ. Hydroprocessing catalyst
585 pounds in the feed deactivate the HDS active sites in lesser extent deactivation in commercial practice. Catal Today 2010;154:256–63. 644
586 than that does by vanadium compounds. The deactivation by [20] Centeno G, Ancheyta J, Alvarez A, Marroquín G, Alonso F, Castillo A. Effect of 645
different heavy feedstock on the deactivation of a commercial hydrotreating 646
587 heavy crude having high concentration of wax materials suggests 647
catalyst. Fuel 2012;100:73–9.
588 that long paraffinic or naphthene or aromatic substituted paraffinic [21] Begon V, Warrington SB, Megaritis A, Charsley EL, Kandiyoti R. Composition of 648
589 chains may take part in coke formation during hydroprocessing carbonaceous deposits and catalyst deactivation in the early stages of the 649
hydrocracking of a coal extract. Fuel 1999;78:681–8. 650
590 reaction conditions by polymerization of free radical mechanism.
[22] Rana MS, Ancheyta J, Sahoo SK, Rayo P. Carbon and metal deposition during 651
591 The coke coming from these long paraffinic or substituted paraf- the hydroprocessing of Maya crude oil. Catal Today 2014;220–222:97–105. 652
592 finic compounds mostly covers the active sites; but does not pre- [23] Mancera PT, Rayo P, Ancheyta J, Marroquín G, Centeno G, Alonso F. 653
Characterization of spent and regenerated catalysts recovered from a residue 654
593 cipitate as solid coke. The formation of solid coke during
hydrotreating bench-scale reactor. Fuel 2015;149:143–8. 655
594 hydroprocessing reaction is observed when the catalyst treated [24] Beaton WI, Bertolacini RJ. Resid hydroprocessing at Amoco. Catal Rev Sci Eng 656
595 with a feed having high concentration of asphaltene. 1991;33:281–317. 657
[25] Vielhaber B, Knozinger H. In situ preresonance Raman and optical spectra of 658
vanadyl tetraphenylporphyrin adsorbed on Mo/Al2O3 catalysts. Appl Catal 659
596 Acknowledgments 1986;26:375–90. 660
[26] Knozinger H, Cordischi D, Vielhaber B. Adsorption of oxovanadium(IV) 661
597 K. Kohli thanks to DST-INSPIRE and IOCL-Faridabad for their porphyrins on oxides and molybdenum-containing catalysts. Catal Today 662
1990;7:447–66. 663
598 financial support for this work.
664

Please cite this article in press as: Kohli K et al. Deactivation of hydrotreating catalyst by metals in resin and asphaltene parts of heavy oil and residues.
Fuel (2016), http://dx.doi.org/10.1016/j.fuel.2016.02.036