Hydrotreating Reaction Optimisation Using Carbon Nanotubes Supported NiMo And Mo Based Catalysts In Refining Crude Oil

Aman A. Dhanani, Pandit Deendayal Petroleum University, Gandhinagar, India

1. Introduction:

Sulfide transition metal catalysts have been widely used for petroleum refining & hydroprocessing such as hydrodesulfurization (HDS) and hydrodenitrogenation (HDN). The traditional hydrotreating catalysts usually contain cobalt or nickel and molybdenum or tungsten with the support commonly used is alumina.

This type of hydrotreating catalysts have been used successfully and commercially used, there are difficulties in these traditional catalysts like difficulty in sulfurization and in reduction of Co-Mo and lower valence active phases. When alumina supported (single-layer) molybdenum oxide was heated in oxygen and H2S/H2 mixture upto 825K, a strong oxide-support interaction was observed, leading to highly dispersed molybdenum phases on alumina support as was studied by Hayden21.

Moreover, for the Hydrodenitrogenation, adsorbed nitrogen and metal heteroatoms are most often the contaminants responsible for downstream catalyst deactivation. The basic nature of these compounds causes them to adsorb onto Lewis acid sites on the catalyst surface, inhibiting the availability of the sites. This poisoning may be reversible or irreversible, depending on the hydroprocessing conditions (Furimsky and Massoth, 1999). High concentrations of organic nitrogen compounds can cause significant deactivation for reforming, cracking, hydrotreating, or any other type of hydroprocessing catalysts.

In the present area of nano-science and nano-technology the CNT supported catalysts have attracted much more attention in this field because of certain reasons like the CNT supports of the hydrotreating catalysts show a high activity, excel in lower coking deposition, easily recoverable from the waste catalysts by burning of the carbonaceous supports. Being a novel material, CNTs have high mechanical and unique electronic properties similar to hollow graphite fibres with seamless tube like graphitic walls with a perfect structure, sp2 carbon-carbon atoms, surface properties(e.g. high specific surface areas maintaining the large pore diameters(>50nm)) which can be easily modified and excellent electron-transporting capability. Moreover the CNTs are available cheaply with a good quality.

In the first study, some oxide state Mo, Co–Mo and sulphide state Mo supported on CNT catalysts were prepared, while the corresponding Al2O3 supported catalysts were prepared as comparison. Laser Raman Spectroscopy (LRS) and X-Ray Diffraction (XRD) techniques were adopted to characterize these CNTs and γ-Al2O3 supported catalysts. In the case of oxide state catalysts, the reducibility of active species was monitored by Temperature Programmed Reduction (TPR) technique. The catalytic performances of CNTs or γ-Al2O3 supported Co–Mo HDS catalysts were evaluated with dibenzothiophene (DBT) as model compound. For comparison γ-Al2O3 supported Co–Mo catalysts were also tested under the same conditions.

In the second study Multi-Walled Carbon Nanotubes (MWCNTs) were synthesised with specific pore diameter by chemical vapour deposition (CVD) of a carbon source on anodic aluminium oxide (AAO) template. AAO templates consist of pore channels in a uniform hexagonal arrangement that run parallel to surface of the film. AAO pore sizes are altered by creating an electrolysis cell (40V potential difference) with passive aluminium anode with (0.4M) oxalic acid as an electrolyte, thus enabling us to regulate the MWCNT pore size. Several grades of NiMo/MWCNT sulphide catalysts were prepared to find optimum pore size and their characterisation was done by Transmission Electron Microscopy (TEM), CO chemisorption, N2 adsorption and H2 Temperature Programmed Reduction (TPR).

2. Results and Discussion:

The Fig.1 shows XRD pattern of the CNT support with oxidised Mo catalysts with different MoO3 loadings with a strong peak at 26.05º showing the graphitic basal plane (002) in CNTs in all samples with or without catalysts loading. A relatively weak and broader peak at 43º due to diffraction of (100) plane showing graphitisation degree of CNTs. XRD pattern of 4%m MoO3/CNT catalyst are similar to that without catalyst loading implying well dispersion of Mo oxides but XRD pattern of 6%m MoO3/CNT and 8%m MoO3/CNT catalyst depict the broad and weak peaks of diffraction due to amorphous MoO2 active species in the catalyst. For 8%m MoO3/CNT catalyst diffraction peaks at 36.9º, 53.49º, 60.52º of MoO2 are obtained 8%m is the maximum loading beyond which bulk phase is lower.

The reason for this hydrophobic nature of CNT support results due to weak interaction between support and active phase. Mo oxide is reduced by CNT support during heat treatment process at 500 ºC and active phase of HDS activity is MoS2 so oxides needs to be transformed to sulphide before use.

The Fig.2 shows XRD patterns of S-MoO3/CNT and S-Co–Mo/CNT catalysts. For of S-MoO3/CNT(10%m MoO3) two broad diffraction peaks are observed characteristic of Mo3S4 with low valence. For S-Co–Mo/CNT new weak and very broad peaks at 14.4º and 32.71º got due to Mo3S4 crystallite and at 29º due to Co-MoS phase at low valence like Co-Mo3.13 and Co-MoS2.17. The active species are well dispersed on CNT support shows that introduction of Co species and sulfiding process help in active phase dispersion on CNT. It should be noted that a small amount of MoO2 still remains unsulfided even in the sulfide state catalysts, as evidenced by the XRD patterns.

Fig.1 XRD patterns of oxide MoO3/CNT Fig.2 XRD patterns of sulfide state catalysts[22] catalysts with different MoO3 loadings.[22]

The Raman Spectra of MoO3/CNT catalyst and the CNT are shown in Fig.3 accords the XRD pattern showing low graphitisation of CNT by peaks at 1346.64cm-1 and at 1585.36cm-1. Rest peaks are due to C-H vibration. In case of the catalyst two new weak bands in low frequency region show vibrations of Mo=O bond in MoO2 proving the XRD result.

The Raman Spectra of the sulfided Mo-based bimetallic catalysts with Co as promoter depict a peak at 353cm-1 due to Mo=S bond vibration. As no peak at 696cm-1 and at 820cm-1 found it concludes no bulk Co oxides or MoO3 in catalysts.

For the Co–Mo-0.7/CNT catalyst, it is believed that there exist three kinds of active species because there are three hydrogen consumption peaks in its TPR curve. In the TPR curve of Co–Mo-0.5/CNT catalyst, two reduction peaks appear. And taking the TPR property of Co–Mo/γ-Al2O3 into account, the reduction of well-dispersed MoO3 species takes place at about 900 °C while the reduction of bulk MoO3 starts at about 500 °C. This results show that active species in Co-Mo/CNT are easily are reduced than those in Co–Mo/γ-Al2O3, indicating CNT supports facilitates reduction of active phases and Co/Mo atomic ratio has great influence on number of active species that existed in catalysts and also the reducibility of active phases. In addition, the electron donor effect and the semiconductor property of CNT may also play a role in the reduction of active species. Moreover it can also be concluded that at least 2 kinds of active species are reduced by Co-Mo-0.7/CNT more easily that those with Co/Mo atomic ratio of 0.2,0.35 and 0.5 resp. Moreover it is interpreted that the all the Co-Mo/CNT catalysts show high selectivity and activity than the γ-Al2O3 supported Co-Mo catalyst. Fig 5,6,7,8 also show that the highest activity is obtained for the catalyst with Co/Mo ratio 0.7 confirm the Bouwens report that the maximum in HDS activity of a kind of activated carbon supported Co–Mo catalyst is situated at a Co/Mo atomic ratio of about 0.73.

Fig.3 Laser Raman spectra of oxide catalysts.5 Fig. 4: Laser Raman spectra of sulfide catalysts5

Now in the second study discussion the results of the following factors in detail as follows:

1. Variation of multi-walled carbon nanotube inner diameters 2. The effect of MWCNT pore diameter variation on catalyst performance 3. Variation of CVD parameters

Table 1. Selectivity and HDS conversion ratio for different catalysts shown below:

Catalysts Selectivity HDS ratio/%m CoMo-0.35/γ-Al2O3-500 4.35 83 CoMo-0.7/CNT-500 10.26 99.6 CoMo-1.0/CNT-500 11.1 91.24 CoMo-0.2/CNT-500 12 94.22 CoMo-0.5/CNT-500 12.4 95.99 CoMo0.35/CNT-500 16.48 96.71

Fig.5: HDS conversion vs. Catalyst Selectivity Fig.6 Catalyst selectivity affected by Co/Mo ratio

Fig.7 Variation of selectivity & HDS ratio/%m Fig.8 Variation of HDS conversion ratio with for different CNT and Alumina catalysts with Co/Mo atomic ratio for varied CNT loadings

0.35, γ-Al2O3

0.7,CNT

1.0,CNT

0.2,CNT

0.5,CNT 0.35,CNT

828486889092949698

100102

0 5 10 15 20

HDS

Ratio

/%m

Selectivity

HDS ratio/%m

HDS ratio/%m

02468

1012141618

0 0.5 1 1.5

Sele

ctiv

ity

Co/Mo Ratio Selectivity

0

20

40

60

80

100

120

HDSratio/%m

90919293949596979899

100101

0 0.2 0.4 0.6 0.8 1 1.2

HDS

ratio

/%m

Co/Mo Atomic RatioHDS ratio/%m

2.1 Variation in MWCNT pore diameter:

2.2 TEM Image of Functionalised MWCNTs:

The significant CNT thickness indicated a mutli-walled morphology and majority of the CNTs were quite linear. As the pore channel diameters increased within the AAO templates the Y-branched CNTs appeared to increase in number but CNT got by 0.5 mol/L oxalic acid concentration and Maximum annodizing voltage of 60V (fig 10).

2.3 Surface Characterization by N2 Adsorption/Desorption:

The Fig.shows the isotherm exhibited at 77K was consistent with the one defined by BET classification. There is marked increase in the adsorption at high relative pressures between an adsorbate and macroporous materials (>50nm). The surface area (A) was found by implementing the BET method, while the total pore volume (V) was determined by the single point method at a relative pressure of ~0.98 and for pore diameters less than ~110 nm. The average pore diameter (d) was then determined by assuming perfectly cylindrical pore spaces (i.e. d = 4V/A), a reasonable assumption for the nanotubes produced, with high reproducibility for each run. The pore diameters estimated after this test were far less than TEM images as CNTs remained close-ended after HNO3 treatment also affecting the pore volume due to Y-branching of the CNTs. Thus from fig.11, optimum value of 3 surface parameters. Also, Smaller AAO pore channels would then result in smaller CNT pore diameters, while larger AAO pore channels would produce more detrimental Y-branched CNTs.

2.4 Raman Spectroscopy:

The vibration characteristics for the different samples of the MWCNTs is shown in Fig.12, where 2 prime intense bands are common in each spectrum characteristic peaks of graphitic materials.

Fig.9 N2 adsorption/desorption isotherm Fig.10 TEM analysis shows pore diameters remain exhibited for condition 4 CNTs & similarly constant at different voltages and oxalic acid for all HNO3 functionalized CNTs.1 concentrations (mol/L)

The band that occurs between wavelengths of 1580 to 1600 cm-1 corresponds to the stretching mode of sp2 graphite bonds and is commonly referred to as the G-band (Eklund et al., 1995; Lefrant, 2002; Dresselhaus et al., 2005; Delhaes et al., 2006). The band appearing at wavelengths between 1330 and 1350 cm-1 is referred to as the D-band and is interpreted as the extent of imperfections and disorder in the graphite sheets (Eklund et al., 1995; Lefrant, 2002; Dresselhaus et al., 2005; Delhaes et al., 2006). The ratio of the intensity of these bands (RI = ID/IG) is often used to represent the quality of the MWCNT morphology.

RI values of the MWNCTs was relatively low(average of 0.97) while for commercial grade of MWCNTs with smaller pore diameters had RI value of 1.57. Pore diameter variation don’t affect the RI values of the graphite layers

2.5 Thermogravimetric Analysis:

The TGA results for all the samples was similar as in as highest rate of CNT oxidation was found to occur at 630°C.Variation in the CNT pore diameter by variation of the AAO synthesis parameters has no effect on the thermal stability of the CNTs.

2.6 Effect of the MWCNT Pore Diameter on Catalyst Performance:

Specific grades of functionalized MWCNTs were chosen for screening as catalyst supports for HDS and HDN of coker light gas oil derived from Athabasca bitumen. The catalysts prepared by incipient wetness co-impregnation method, were denoted according to their approximate inner diameters: cat-60, cat-65, cat-70, and cat-75.

Fig.11 BET surface Area Comparison Fig.12 Results by Raman Spectroscopy1

2.7 HAADF-STEM Images of Functionalised MWCNTs:

High angle annular dark field scanning transmission electron microscopy (HAADF-STEM) images shows that the catalyst particles were chemisorbed on inner and outer walls of the MWCNTs and

160

180

200

220

240

BET Surface Area BET Surface Area BET Surface Area

BET Surface Area Comparison

Oxalic Acid Concentration 0.3(mol\L) at 40,50,60 V Maximum Annodizing Potential

Oxalic Acid Concentration 0.4(m2ol\L) at 40,50,60 V Maximum Annodizing Potential

Oxalic Acid Concentration 0.5(mol\L) at 40,50,60 V Maximum Annodizing Potential

their sizes varied from 1 to 5nm in diameter. Catalyst sample image after the hydrotreating operation show that coke filaments branched of the metal catalysts (one of the reason for catalyst deactivation).

Fig.13 BET surface Area for different samples Fig.14 Total pore volume for different samples

2.8 Characterization by N2 Adsorption/Desorption:

The analysis performed on the pelletized catalysts show give the result that for cat-60 surface characteristics decrease significantly due pore blockage and support compression. But cat-65 has most desirable surface characteristics as per fig. Structural characteristics of the prepared NiMo/MWCNT catalysts (~2.5 wt.% Ni, ~13 wt.% Mo), wherein the thickness of each of the catalyst was found to be same1.

2.9 CO Chemisorption for Varied NiMo/CNT Pore Diameters:

This analysis gave the results as follows with cat-65 had the highest amount of CO uptake; 1.43 cc/g for a system pressure of 400 mmHg. The analysis as per fig. 16(red column: estimated metal exposure & blue column: CO uptake) showed that as surface area increased large amount of CO was chemisorbed and catalyst performance increased. Low exposure of the metals show large number of catalyst atoms were left inaccessible (here ~1.5%).

Fig. 15 Average Pore Diameter variation Fig. 16 CO chemisorption analysis

106

169

141154

0

50

100

150

200

cat-60 cat-65 cat-70 cat-75

BET Surface Area (sq.m/g)

0

0.1

0.2

0.3

0.4

0.5

0 1 2 3 4 5

Total Pore Volume (cc/g)

0

2

4

6

8

10

12

cat-60 cat-65 cat-70 cat-75

Aver

age

Pore

Dia

met

er

(nm

)

Average Pore Diameter (nm)

0

1

2

3

4

cat-60 cat-65 cat-70 cat-75

CO Chemisorption Analysis

2.10 Temperature Programmed Reduction with H2:

Fig.17 TPR using H2 of NiMo/MWCNT catalysts Fig.18 Carbon number composition of the steady- with varied pore diameters1 state coker light gas oil product1

The reduction peaks for the catalysts with varied pore diameters can be found in Fig.17. The hydrogen consumption peaks, which occurred within a domain of 445-456°C for the MWCNT-supported catalysts and at 506°C for the commercial γ–alumina catalyst, are the result of Mo reduction from the amorphous Mo6+ (octahedral) oxide phase to the ordered Mo4+ (tetrahedral) oxide phase (Leyva et al., 2008; Qu et al., 2003; Yu et al., 2008). The lower reduction temperatures of the MWCNT catalysts indicate less interaction with metal catalyst compared to γ-alumina thus making it easier for transition to sulphide phase for MWCNT catalysts in hydrotreating application. The TPR temperature being nearly similar for cat-70 and cat-75 showed that two supports were similar enough in surface features. The specific consumption peak temperatures for each catalyst were 456°C for cat-60, 453°C for cat-75, 450°C for cat-70, and 445°C for cat-65.

2.11 Boiling Point Distribution of CLGO Feed and Products:

The results of the hydrotreating the CLGO feed samples for commercial catalyst and prepared samples show that average pore diameter lowered from 75 to 60 nm, the percentage of carbon numbers ≤C15 increased steadily from 25.4% to 28.4%. The inverse trend was found for carbon numbers ranging from C16 to C19 and ≥C20, as the percentage of molecules within this range decreased steadily from 32.9% to 30.3% and 41.7% to 40.9% as the average pore diameter of the catalysts dropped from 75 to 60 nm resp. Hence, by making the pores of the MWCNT supports more confined leads to more extensive cracking of CLGO (coker liquid gas oil) molecules during the hydrotreating process. Refer Fig.18.

2.11 Catalyst Performance Based on N and S Conversion:

HDS & HDN steady-state activities of each catalyst from the CLGO feed can be found numerically in Fig. 19 and Fig. 20 for 4 MWCNT-based catalyst with a commercial catalyst with γ-alumina support under equivalent loading condition to evaluate the applicability of the NiMo/CNT sulphide catalyst.

Here the steady-state activity is obtained by averaging the conversions found for all the samples at steady-state (got after 24 hours) at 370°C 350°C and 330°C. The catalyst displayed sulfur conversions of 77.0%, 61.0%, and 47.4% and similarly for nitrogen, steady-state conversions of 37.5%, 26.7%, and 19.3% were found at respective temperatures of 370°C, 350°C, and 330°C for cat-65 rep. Under the same reaction conditions and at equal mass loading, the commercial γ–Al2O3 catalyst performed steady-state sulfur conversions of 51.1%, 37.8%, and 30.7%, along with nitrogen conversions of 26.6%, 16.2%, and 8.7%, for the three descending reaction temperatures. The cat-60, cat-70 and cat-75 had decreasing order of activities and long term detrimental effects was due to precooling taking place for the smaller pores for cat-60 and large inner diameters of cat-70 and cat-75 resulted in less dispersion of catalyst metals or direct mass transport of gas oil through catalyst without any reaction. Cat-65 had sufficient mass transfer of the reactant liquids and gases though maintaining high surface areas necessary for sulphide metal dispersion.

Fig.19 HDN activities in steady-state1 Fig.20 HDS activities in steady-state1

2.12 Variation of CVD Parameters:

Carbon yields from C2H2 achieved for nine chemical vapor deposition conditions and 0.250 grams of AAO template per run. Depict that highest yield of CNT was obtained at 650oC and yield also decreased with the temperature of the reaction as shown in Fig. 21 and Fig.22 .

Fig.21 MWCNT yield at different temperatures Fig.22 Carbon mass deposited during CVD

3. Conclusions:

From the first study it can be concluded that, the pore diameter variation of the AAO film influenced the preparation of the CNTs with desirable structural characteristics (at 40V potential and 0.4M oxalic acid concentration). The CNTs have average pore diameter of 10-11.5 nm, 229 m2/g BET surface area and a 0.658 cc/g single-point pore volume.

The MWCNTs with approximate inner diameters of 65-67 nm were found to provide the optimum HDS and HDN activities for a Ni (2.5 wt.%) Mo (13.0 wt.%) catalyst. The catalyst displayed sulfur conversions of 77.0%, 61.0%, and 47.4% and nitrogen conversions of 37.5%, 26.7%, and 19.3% at reaction temperatures of 370°C, 350°C, and 330°C, respectively which is relatively high compared to commercial alumina catalysts.

Moreover for the chemical vapour deposition process operation conditions at 650°C reaction temperature produced highest carbon yield of 85% MWCNT product.

Now from the second study, the conclusions can be drawn as follows, the TPR results show that active species of Co-Mo/CNT catalysts can be reduced at relatively lower temperature compared to that of Co–Mo/γ-Al2O3. Moreover the CNTs favour s the reduction of the active species than the γ-Al2O3. The Co/Mo ratio greatly affected the reducibility of catalysts where catalyst with Co/Mo ratio 0.7 having the highest reducibility and the reducibility decreasing for the 0.2, 0.35 or 0.5 resp.

The HDS experiments prove that the Co-Mo/CNT catalysts have higher activity and selectivity for the hydrogenation compared to Co–Mo/γ-Al2O3 where the Co/Mo atomic ratio of 0.7 has highest activity and 0.35 ratio having the highest selectivity for the Co-Mo/CNT catalyst. The weak interaction between the active phases of CNT result in greater dispersion, surface atomic concentration of Co, Mo and S and chemical environment for Mo atom compared to γ-Al2O3 catalyst. Thus it can also be concluded that there is relation between the reducibility and HDS activity of the catalysts.

0102030405060708090

100

1 2 3 4 5 6Yield at 4mL/min.g flow rate at 450, 550,650 deg CYield at 5mL/min.g flow rate at 450, 550,650 deg CYield at 6mL/min.g flow rate at 450, 550,650 deg C

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

1 2 3Carbon Mass at 4mL/min.g flow rate at 450, 550,650 deg C

Carbon Mass at 5mL/min.g flow rate at 450, 550,650 deg C

Carbon Mass at 6mL/min.g flow rate at 450, 550,650 deg C

4. Refernces:

1. Stefan Kasey Sigurdson “Hydrotreating of Light Gas Oil Using Carbon Nanotube Supported NiMoS Catalysts: Influence of Pore Diameters” December,2009

2. Ancheyta J., Speight J.G., Hydroprocessing of Heavy Oils and Residua, Boca Raton, FL: CRC Press, 2007

3. Chen Y., Ciuparu D., Lim S., Haller G., Pfefferle L.D., “The effect of the cobalt loading on the growth of single wall carbon nanotubes by CO disproportionation on Co-MCM-41 catalysts”, Carbon, 67-78, 44, 2006

4. Dai H., “Carbon nanotubes: opportunities and challenges”, Surface Science, 218-241, 500, 2002 5. Diaz-Real R.A., Mann R.S., Sambi I.S., “Hydrotreatment of Athabasca bitumen derived gas oil

over Ni-Mo, Ni-W, and Co-Mo catalysts”, Industrial and Engineering Chemistry Research, 1354-1358, 32, 1993

6. Dong K., Ma X., Zhang H., Lin G., “Novel MWCNT-support for Co-Mo sulfide catalyst in HDS of thiophene and HDN of pyrrole”, Journal of Natural Gas Chemistry, 28-37, 15, 2006

7. Dresselhaus M.S., Dresselhaus G., Saito R., Jorio A., “Raman spectroscopy of carbon nanotubes”, Physics Reports, 47-99, 409, 2005

8. Eklund P.C., Holden J.M., Jishi R.A., “Vibrational modes of carbon nanotubes; Spectroscopy and theory”, Carbon, 959-972, 33, 1995

9. Eswaramoorthi I., Sundaramurthy V., Das N., Dalai A.K., Adjaye J., “Application of multi-walled carbon nanotubes as efficient support to NiMo hydrotreating catalyst”, Applied Catalysis A: General, 187-195, 339, 2008

10. Furimsky E., “Role of MoS2 and WS2 in hydrodesulfurization”, Catalysis Reviews - Science and Engineering, 371-400, 22, 1980

11. Gras R., Duvail J.L., Minea T., Dubosc M., Tessier P.Y., Cagnon L., Coronel P., Torres J., “Template synthesis of carbon nanotubes from porous alumina matrix on silicon”, Microelectronic Engineering, 2432-2436, 83, 2006

12. Gruia A., “Chapter 8: Hydrotreating”, Handbook of Petroleum Processing, 321-354, Dordrecht: Springer, 2006

13. Lee O.J., Hwang S.K., Jeong S.H., Lee P.S., Lee K.H., “Synthesis of carbon nanotubes with identical dimensions using an anodic aluminum oxide template on a silicon wafer”, Synthetic Metals, 263-266, 148, 2005

14. Sigurdson S., Sundaramurthy V., Dalai A.K., Adjaye J., “Effect of anodic alumina pore diameter variation on template-initiated synthesis of carbon nanotube catalyst supports”, Journal of Molecular Catalysis A: Chemical, 23-32, 306, 2009

15. 207 (2001) 421 from Fuel Processing Technology, M.C. Abello, M.F. Gomez, O. Ferretti, Appl. Catal., A Gen.

16. 187 (1999) 213 from Fuel Processing Technology, A Gen. 17. 105(1987) 299 from Fuel Processing Technology, F.T.J. Hayden, A. Dumesic, R.D. Sherwood,

R.T.K. Baker, J. Catal.