Applied Catalysis A: General 411–412 (2012) 35–43
Contents lists available at SciVerse ScienceDirect
Applied Catalysis A: General journal homepage: www.elsevier.com/locate/apcata
Study of the role of the catalyst and operating conditions on the sediments formation during deep hydroconversion of vacuum residue Charles Marchal a , Denis Uzio a,∗ , Isabelle Merdrignac a , Loïc Barré b , Christophe Geantet c a b c
IFPEN, Rond-point de l’échangeur de Solaize, BP 3, 69360 Solaize, France IFPEN, 1 & 4 avenue de Bois-Préau, 92852 Rueil-Malmaison Cedex, France Institut de Recherches sur la Catalyse et l’Environnement de Lyon, IRCELYON, UMR 5256 CNRS Université Lyon 1, 2 avenue Albert Einstein, F-69626 Villeurbanne, France
a r t i c l e
i n f o
Article history: Received 29 July 2011 Received in revised form 15 September 2011 Accepted 13 October 2011 Available online 21 October 2011 Keywords: Hydroconversion Vacuum residue Sulfide catalyst Fluorine Effluent stability Sediment formation Asphaltenes Resins
a b s t r a c t The aim of this work is to investigate the influence of the catalyst and the operating temperature on the sediments formation occurring during vacuum residue hydroconversion. At high conversion, instability phenomena occurring due to carbonaceous sediments formation is a tricky issue to address and has detrimental effects on the operability of industrial units. The impact of the catalyst properties on this phenomenon has been scarcely studied. Therefore, a doped F doped NiMo/Al2 O3 catalyst has been prepared and catalytic tests performed at two temperature levels (430 and 390 ◦ C) in order to study the effect of catalytic hydrocracking or thermal conversion on the sediment formation. For both temperature regimes, it has been observed that the amount of sediments is strongly reduced with the F-NiMo catalyst compared to the undoped one and attributed to the improvement of the conversion of the sediment precursors (asphaltenes). This improvement is mostly due to the promotion of the acidic and hydrogenation functions by F addition. At 390 ◦ C, coke deposition on F-NiMo catalyst is reduced which limits the plugging of the porous volume and preserve the intragranular diffusion of large molecules in the porosity. It has been shown that deep asphaltenes conversion can be achieved using an improved hydrogenation catalyst by reducing their polarity through the removal of heteroatoms and the saturation of some aromatic rings by hydrogen addition. The residual molecules exhibit lower interactions and self association ability resulting in a lower sediment formation. The temperature has also a great impact on the sediment formation in the effluents since the decrease of the temperature favors hydrogenation and catalytic hydrocracking route which is more appropriated to produce stable effluents compared to radical chemistry predominant at high temperature. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Conversion of heavy oils still attracts a lot of attention from researchers in the refining industry since it faces major challenges in the next future. Indeed, this field is strongly stimulated by several incentive factors such as more stringent specifications, increasing overall naphtha and middle distillates demand, the decline of the production of light crude oil gradually replaced by heavier non conventional resources. These challenges highlight the need to improve efficiency of the existing processes in order to meet these goals. Different options are available to upgrade heavy oils containing high level of asphaltenes such as solvent deasphalting by liquid–liquid extraction, contacting the feed with a non polar solvent, or converting them to coke through a thermal cracking process like coking [1,2]. However, such routes are not fully satisfactory since a lot of raw material is not converted into valuable
∗ Corresponding author. Tel.: +33 04 37 70 22 43; fax: +33 04 37 70 20 66. E-mail address:
[email protected] (D. Uzio). 0926-860X/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2011.10.018
products. Among the existing solutions, deep hydroprocessing is nowadays well established and is widely spread in the refining industry worldwide. In the specific case of vacuum residue conversion containing high metal (Ni, V) and asphaltenes loadings, ebullated bed technology is particularly suitable to convert up to 60–70% of the 540 ◦ C+ fraction [3–5]. The reaction is performed at high temperature and pressure with continuous addition and withdrawal of catalyst which enable to control the deactivation by coke formation as well as thermal exchanges in the reactor. Nevertheless, improvements of such processes need to overcome a tricky limitation due to the instability of effluents leading to sediment formation at high residue conversion. Indeed, these carbonaceous sediments promptly deposit on the equipment causing severe operability issues as well as on the catalyst deactivation [5,6]. Many attempts have been made in order to get a better understanding on the instability phenomenon observed at high hydroconversion levels [7]. Asphaltenes are known to be the major precursors of sediments [8]. Structure and chemical composition of asphaltenes such as hydrogen content and aromaticiy factor are parameters governing the self aggregation [9]. Recently, Mullins [10] revisited the
36
C. Marchal et al. / Applied Catalysis A: General 411–412 (2012) 35–43
“Yen” model, in order to account for most of the properties of heavy molecules known today. Among these properties, the chemical composition, namely heteroatoms or metals, of the asphaltenes and resins is also supposed to govern the interactions because polarity and heteroatoms content are closely related [11,12]. According to Mitra-Kirtley et al. [13], nitrogen is mostly present in aromatic structures as basic nitrogen (case of pyridine type species), or as neutral nitrogen (case of pyrrolic cycle). Because of its electronegativity, nitrogen improves the charges delocalization, aromatic rings polarization is then increased which in turn strengthen the attractive interactions. The same effect may be expected with other elements such S or O and metals such as Ni and V contained in porphyrinic structures. The composition of the liquid product is also of primary importance. Using a thermodynamic modelling approach [14], Rogel studied the influence of the molecular structure of the resin/asphaltene system on the aggregation process. Results show that asphaltenes solubility depends on resins molecular structures which strongly impact the asphaltene aggregation state. The stabilizing effect is supposed to be due to the formation a mixed asphaltene–maltenes (mainly resins and aromatics) entity [15] rather than a micellar structure with adsorption of the resin around the asphaltene nano-aggregate like in the initial model proposed in the pioneer work of Nellensteyn [16]. Resins concentration plays a key role in asphaltenes aggregation [17], if the total amount of resins interacting with asphaltenes is not high enough, auto-association of asphaltenes take place forming aggregates. Asphaltene structure (such as the peri or cata-condensed types) is also an important feature to take into account when studying sediment formation [18] since peri-condensed asphaltenes are often considered as strong deposit precursors [19]. Deeper conversion leads to smaller but more aromatic molecules, Ancheyta et al., using vapour-phase osmometry and NMR analysis, showed that aromaticity of residual asphaltenes increases and molecular weight decreases raising temperature (especially above 420 ◦ C) for Maya and Isthmus residues [20]. Other studies also reported such tendency for different feedstocks and technologies (fixed or ebullated bed reactors) with the formation of highly polycondensed dealkylated aromatic structures [21–23]. Bartholdy and Andersen [24] studied the effect of the temperature on the stability followed by a titration method of hydrotreated asphaltenes from an Arabian heavy atmospheric residue. A strong increase of the precipitation onset of asphaltenes was observed above 380 ◦ C (13 MPa) correlated with a deep change of their chemical composition (decrease of H/C). [25]. However, this limit is feed dependent as show by Ancheyta et al. [20] who reported that thermal cracking of Maya and Isthmus asphaltenes is dominant above 420 ◦ C. Condensation reactions leading to coke formation are favored at high temperatures because of the rapid increase of radical concentration with temperature [26,27]. Kinetic of hydroconversion of these different species is also a parameter to consider since stabilizing molecule and flocculating ones may have different reactivity and unbalanced system can lead to fast auto-association of asphaltenes and sediments formation. It is then important to consider the selectivity between these two families of molecules, selectivity which may be oriented by the reaction pathway itself (thermal cracking or catalytic hydrocracking) [28,29]. Few studies have been devoted to the catalyst effect on effluents stability in vacuum residue hydroconversion process. Catalysts porosity has been identified to strongly influence on sediments formation [30,31]. Concerning the effect of surface acido-basic properties and active phase on sediments formation, it is still a matter to explore. To summarize, asphaltenes stability has been extensively studied and models describing the phenomenon have reached a high
Table 1 Main characteristics of the Safaniya Residue feedstock (SVR). Specific gravity at 15 ◦ C, g cm−3 Viscosity at 100 ◦ C, cSt 520 ◦ C+ cut, wt% Saturates Aromatics Resins Asphaltenes S, wt% N, ppm Ni, ppm V, ppm Carbon Conradson, wt%
1.0277 1516 84 11.2 40.3 33.9 12.4 4.94 3724 42 143 20.1
level of complexity. Progresses in the comprehension of instability phenomena have been made recently thanks to the development of new characterization techniques. Nevertheless, if it is now possible to list the different features responsible of the stability of asphaltenes, it is still a challenge to identify which parameters are the governing ones for a special set of operating conditions in order to make reliable predictions on the sediment formation onset. Moreover, as far as catalytic hydroconversion is concerned, few studies have been focussed on the impact of the catalyst properties on the sediment formation in function of the temperature controlling the reaction pathway (hydrocraking versus radical). Therefore, our study aims to explore and figure out the role of the catalyst acido-basic properties on the sediment formation in catalytic (roughly T < 400 ◦ C) and thermal regimes (T > 400 ◦ C). 2. Experimental 2.1. Catalysts preparation The reference catalyst is a standard NiMo/␥-Al2 O3 hydrotreating catalyst containing 10 wt% of MoO3 and 4.5 wt% of NiO with a meso–macro bimodal pore size distribution, a 0.9 mm diameter cylindrical extrudate shape and a length of 3 mm. Fluorine doped catalyst has been prepared by incipient wetness impregnation of an aqueous solution of NH4 F, followed by a maturation step during 24 h at room temperature in a water controlled humidity cell, dried for 12 h at 120 ◦ C, and finally calcined for 2 h at 500 ◦ C under a 2 Lh−1 gcata −1 air flow. The fluorine loading on the final catalyst is 5.1 wt%. Before the catalytic tests, catalysts have been ex situ sulfided at 450 ◦ C during 2 h under a flow of 2 Lh−1 gcata −1 of 15% (v/v) H2 S/H2 . Using a Castaing JEOLJXA8100 electron-probe microanalyzer, it has been checked that all the elements, including fluorine, were uniformly dispersed along the length of the extrudates. Catalysts acidity and active sites of sulfided catalysts has been analyzed by CO adsorption followed by IR spectroscopy at low temperature (77 K) using a Nexus Nicolet Fisher Scientifique spectrometer. 2.2. Catalytic tests Hydroconversion catalytic tests have been performed in a 300 mL stainless-steel batch reactor with 15 mL of presulfided NiMo/␥Al2 O3 catalyst and 90 mL of Safaniya Vacuum Residue (SVR). The main characteristics of SVR are given in Table 1. During the catalytic tests, the catalyst/oil weight ratio is 0.1, the pressure was kept constant at 14.5 MPa and the stirring rate set to 900 rpm. The tests were performed during 8 h at 390 ◦ C (catalytic regime), and 2 h at 430 ◦ C for conditions representative of thermal cracking. The spent catalysts, recovered after SVR hydroconversion tests, were washed 3 times with toluene using a Soxtherm device (Gerhardt, Germany) at the boiling point of the solvent (110 ◦ C at atmospheric pressure) under reflux for 7 h in order to remove the
C. Marchal et al. / Applied Catalysis A: General 411–412 (2012) 35–43
soft part of coke. The solid is then dried in an oven at 150 ◦ C for 3 h. Coke is defined in this work as the carbon content of the spent catalyst after washing by hot toluene. The specific surface area and mesopore size distribution of fresh and used catalysts have been determined from N2 isotherms at 77 K using a Micromeritics ASAP2420 devices after degassing at 350 ◦ C for 4 h (300 ◦ C for the coked catalyst). Hg porosimetry has also been performed to obtain the whole pore size distribution using a Micromeritics Autopore IV 9500 series porosimeter after a thermal treatment at 250 ◦ C for 4 h.
As =
aro Cqsub aro Caro − Cqcond
HDX =
HDAs =
mf · [Asph]f − mp · [Asph]p mf · [Asph]f
× 100
(1)
mf is the feed mass (g); mp the product mass (g); [Asph]f the asphaltene concentration in the feed; and [Asph]p is the asphaltene concentration in the product. ASTM D1160 normalized distillation is performed in order to obtain 370 ◦ C− and 370 ◦ C+ cuts, the residue conversion is obtained making a simulated distillation IFP9522 on the 370 ◦ C+ fraction X520+ (%) =
mf × (X520+ )f − mp × (X520+ )p mf × (X520+ )c
× 100
(2)
q
aliphatic carbons CH3 , CH2 , CH and the quaternary aliphatic Csat . The main structural parameters calculated from 13 C NMR results are the aromaticity factor (fa ), CH2 /CH3 ratio for saturated carbons, and the percentage of substitution of aromatic rings (As ), which are described below: fa =
Caro Caro + Csat
mc · [X]c − mp · [X]p mc · [X]c
(3)
(4)
× 100
(5)
mc is the asphaltene mass in the feedstock (g), mp the asphaltene mass in the product (g), [X]c the % weight of X in the asphaltene of the feed, and [X]p is the % weight of X in the asphaltene of the products. The relative repeatabilities are 2 and 3% for of HDS and HDN respectively. Asphaltenes and resins molecular mass have been determined by Size Exclusion Chromatography (SEC) on a Waters Alliance 2695 system using a refractive index detector. SEC Calibration has been performed using 10 monodispersed polymer standards with a mass in the range 162–120 000 g/mol (Polymer Laboratories, United Kingdom). 50 L at 5 g/L in tetrahydrofuran was injected, the column temperature was fixed at 40 ◦ C and the flow rate at 0.7 mL/min. Three columns packed with polystyrene–divinylbenzene supports (PS-DVB, Polymer Laboratories) were chosen. The corresponding porosities are 100, 1000, and 10,000 A˚ and the column characteristics are as follows: particle size, dp = 5 m; column length, L = 300 mm; As a consequence, resins and asphaltenes molecular weight Mw mentioned in this study are expressed as equivalent polystyrene and are evaluated as follow: Mw =
NM2 i i i i
(X520+ )f is the mass fraction boiling above 520 ◦ C in the feed; (X520+ )p the mass fraction boiling above 520 ◦ C in the product (liquid + gas); mf the feed mass (g); and mp is the product mass (liquid + gas) (g). SARA fractionations of the 370 ◦ C+ cut have been performed according to the following protocol: asphaltenes have been firstly recovered by precipitation through a derived method of the NF T60-115. The resulting maltenes are then splitted into saturates, aromatics and resins by preparative liquid chromatography: maltenes are injected into a silica–alumina chromatographic column. Saturates aromatics and resins are obtained by increasing solvent polarity. Three solvents of increasing polarity are used to separate the saturates, the aromatics and resins and are respectively: heptane, heptane/toluene and toluene/dichloromethane/methanol. The carbonaceous molecular structures of resins and asphaltenes have been investigated by 13 C NMR using a 600 MHz Bruker Avance spectrometer. Based on this method proposed by Bouquet and Bailleul [33], it is possible to determine the proportion of the different types of aromatic carbons: CH and aro ) or quaternary Cqaro , including the quaternary condensed (Cqcond aro ) and the fraction of the different quaternary substituted (Cqsub
× 100
with: Caro is the total aromatic carbons, Csat the total saturated cararo the alkyl-substituted quaternary aromatic carbons and bons, Cqsub aro Cqcond the condensed quaternary aromatic carbons. CHNS elemental analysis have been determined using an EA 1100 CE Instruments analyzer. Hydrodesulfurisation and denitrogenation of asphaltenes have been determined (HDX, X for S or N) with Eq. (5):
2.3. Effluents analysis Sediment formation is followed measuring the total mass of sediments in the final products (total liquid effluent or 370 ◦ C+ fraction) using the normalized ASTM D4870-88 (IP375/93) hot filtration procedure [32]. After the catalytic test, effluents are collected and splitted into four samples in order to avoid the filter plugging. The √ accuracy is given by r = 0.123 x with x representing the sediments percentage in the filtered product. Asphaltenes amount in the feed or in the products has been determined using the AFNOR T60-115 method. Asphaltenes conversion has been calculated according to Eq. (1):
37
Ni Mi
(6)
where Ni represents the number of molecules, and Mi their molecular mass. The relative repeatability of the overall procedure (including deasphalting step) is 20%. It is well known that asphaltene dispersed in maltene self associate to form aggregates at few nanometer length scale. Due to their higher molecular weight than maltenes, they give rise to electron rich regions whose correlation can be probed using small angle X-ray scattering (SAXS). Intraparticle correlations are related to shape and size of nano-aggregates whereas interparticle correlations depend on concentration and interactions. Converted asphaltenes have been dissolved in their corresponding maltenes for concentrations ranging from 0.01 to 0.05 g cm−3 . Density measurements of diluted solutions of asphaltenes and maltene in toluene were performed using an Anton Paar DMA500 densitometer. The specific volume as a function of solute mass fraction show a linear behavior which allows extrapolating data at null dilution and estimating asphaltenes and maltenes mass densities da and dm . Scattering length densities of asphaltenes a , and maltenes m have been deduced from mass density and elemental analysis according to Eq. (7): =
Ndle xi Zi 100
(7)
i
where xi and Zi are respectively the wt% and the electron number of atom i, d is the mass density, N the Avogadro number and le the scattering length of one electron (0.281 × 10−12 cm). The corresponding contrast term 2 = (a − m )2 is 6.4 × 1020 cm−4 . SAXS experiments have been performed using a cell heated at 330 ◦ C in order to obtain a dispersion of asphaltenes in maltenes
38
C. Marchal et al. / Applied Catalysis A: General 411–412 (2012) 35–43
mixture at sub-micrometer scale and to perform measurements under temperature conditions close to hydroconversion ones. The SAXS equipment used for this study is a homemade device: the X-ray radiation is reflected by a multilayer parabolic mirror to ˚ parallel beam. Two pairs of provide a monochromatic ( = 1.5418 A) crossed slits are used to define the beam size (≈1 mm2 on the sample) and another pair located just before sample is used to remove parasitic scattering. The scattered intensity in a wave vector range of 8 × 10−3 up to 0.25 A˚ −1 is measured using a 2D proportional detector located at 1.55 m from the sample. The wave vector q, defined by q = 4/ sin(), where 2 is the angle between the incident and scattered beams, has the dimension of a reciprocal length and can therefore be regarded as an inverse meter stick. For each sample, the scattering intensity has been found isotropic allowing an azimutal averaging for each value of q. Each spectrum has been normalized according to sample thickness, and transmission. Finally scattering from windows and maltene has been subtracted and data converted into scattering cross section per unit volume I(q) (cm−1 ) [34]. Information about asphaltenes aggregates can be obtained using the Zimm approximation [35] relevant for moderately dilute solutions of particles. This approach has been successfully applied for asphaltene in toluene investigations [36,37]. The following Eq. (8), valid for small q and c values, links up experimental measurements to relevant parameters Mw ; Rg and A2 of aggregates 2 c 1 = Mw I(q)
1+
q2 Rg2
3
+ 2A2 Mw c + · · ·
qRg < 1; A2 Mw c < 0.25
(8)
where Mw and Rg represent respectively the mass and radius of gyration. A positive (resp. negative) value of A2 , the so called second virial coefficient, correspond to repulsive (resp. attractive) interactions between aggregates whereas the amplitude of A2 is related to the interaction strength. For each spectrum, an extrapolation to q = 0 has been performed to estimate I(0). Eq. (8) can then be simplified according to (9): 2 c 1 1 = = [1 + 2A2 Mw c + · · ·] Mw MAPP (c) I(0)
(9)
The slope of 2 c/I(0) versus c plot allow to estimate A2 . 3. Results and discussion 3.1. Physico-chemical characterization of the fresh and used catalysts Fluorine has already been reported as an efficient promoter of the hydrogenation and acidic functions of NiMo catalysts via induced electronic effects [38,39]. Depending on the nature of the carrier and active phase loading, the maximum effect is observed for contents ranging from 3 to 6 wt%. IR(CO) spectroscopy shows two characteristic bands of CO on alumina observed at 2155 and 2190 cm−1 corresponding respectively to CO hydrogen-bonded to hydroxyl groups (Brønsted acid sites) and to CO coordinated to the Lewis acid sites of alumina (Al3+ ) [40]. Table 2 shows that fluorine addition increases the total number of Brønsted acidic sites by a factor 2.4 with a slight red shift effect of the wavenumber (2154 cm−1 for NiMo to 2162 cm−1 for F-NiMo). This effect of fluorine on the catalyst acidity, already described in previous works [41–44], is usually ascribed to an attractive electronic effect of the fluorine modified alumina carrier enhancing the protons acidity [45]. The absence of Lewis acidity (2189 cm−1 ) after fluorine addition is usually attributed to the coordination of F on Al Lewis acid sites by formation of terminal Al–F groups [41,46].
Table 2 IR(CO) characterization results of the NiMo and F-NiMo catalysts. CO Band area (normalized)
Acid sites
Lewis acid sites (CO) (cm−1 ) Brønsted acid sites (CO) (cm−1 ) Sulfur sites NiMoS promoted sites (CO) (cm−1 ) Unpromoted MoS2 sites Unpromoted NiS sites (CO) (cm−1 )
NiMo
F-NiMo
100 2189 100 2154 100 21 2127 45 35 2110
0 – 237 2162 105 51 2129 38 11 2114
As illustrated in Fig. 1, the amount of promoted NiMoS sites strongly increased after fluorine addition and simultaneously, bands corresponding to unpromoted MoS2 and NiS phases decreased. The formation of NiMoS phase is therefore favored in presence of fluorine but the electronic properties of the sites are not significantly modified as far as we consider the constant values of the (CO) (Table 2). Finally, no modification of the length or the stacking of the sulfur slabs has been evidenced by TEM analysis before and after F addition. The pore size distributions obtained from Hg porosimetry and N2 physisorption measurements for the fresh and the used catalysts are presented in Figs. 2 and 3. After the catalytic tests performed at 430 ◦ C, the total porous volume is greatly reduced from 0.59 mL/g for the fresh catalysts to 0.22 mL/g and 0.23 mL/g for NiMo and FNiMo respectively. Pore size distributions are very close for both used catalysts and the coke content is almost the same (20 wt%, Table 3). If we now consider the catalytic tests performed at 390 ◦ C, one can observe a significant disparity between the residual mesoporous volume between the NiMo (0.13 mL/g) and F-NiMo catalysts (0.18 mL/g) related to the lower amount of coke deposited for the latter solid. Pore size distributions (Fig. 3b) show the higher contribution of mesopores in the range 7–15 nm for the F-NiMo catalyst whereas macropores are affected in a lesser extent by coke deposition (Fig. 2a and b). The aforementioned differences mostly observed at 390 ◦ C of the textural properties and coke contents between NiMo and F-NiMo used catalysts may have a strong influence on mass transfer properties, in particular in the case of large molecules such asphaltenes. Using the following expression taken from the hydrodynamic model for hindered diffusion described in Refs. [47,48], it is possible to estimate the asphaltenes effective diffusion coefficient in the porous network of each catalyst: 2
Deff = cata · Vmeso · (1 − ) · (1 − 2.104 + 2.0893 − 0.9485 ) · D∞
(10)
where cata is the catalyst density (kg/m3 ), Vmeso the mesoporous volume (mL/g), the ratio Rh /r0 with Rh the hydrodynamic radius of asphaltenes nano-aggregates in the Safaniya feedstock (2.95 nm according to a previous study [49]), and r0 the catalyst average mesopore radius (nm), and D∞ the translational phase diffusion coefficient (m2 /s) (also determined in Ref. [49] at room temperature and high dilution regime in order to avoid large aggregation and to measure nano-aggregate self diffusion properties). For the catalysts tested at 390 ◦ C, the ratio Deff,F-NiMo /Deff,NiMo of the effective diffusion coefficients of asphaltenes is 1.4 whereas it is close to 1 for the catalysts tested at 430 ◦ C. The effective diffusion of the asphaltenes is significantly improved mainly due to the lower amount of coke deposited inside the porous network of the F-NiMo. This point may have a strong impact on the overall
C. Marchal et al. / Applied Catalysis A: General 411–412 (2012) 35–43
39
Table 3 Textural properties and wt% of coke deposited after SVR conversion for the fresh and used catalysts. Catalyst
Mesoporousa volume (mL/g)
Macroporousb volume (mL/g)
Total porous volume (mL/g)
SBET (m2 /g)
Catalyst density (kg/m3 )
NiMo, fresh F-NiMo, fresh NiMo, 430 ◦ C F-NiMo, 430 ◦ C NiMo, 390 ◦ C F-NiMo, 390 ◦ C
0.41 0.41 0.10 0.11 0.13 0.18
0.18 0.18 0.12 0.12 0.13 0.14
0.59 0.59 0.22 0.23 0.26 0.32
305 262 20.2 19.8 15.0 12.8
1009 1092 1399 1375 1288 1227
a b
C wt% (H/C molar ratio) – – 20.2 (0.77) 19.8 (0.79) 15.0 (1.12) 12.8 (1.60)
For pore diameter (nm) in the range [2,50]. For pore diameter above 50 nm.
conversion of these large molecules since conversion of heavy fractions is strongly hindered by internal diffusion limitations. 3.2. Influence of the fluorine addition to the NiMo catalyst on the effluents properties Effluents filtrations show that the amount of sediments is roughly divided by two when using the F-NiMo catalyst for both
reaction temperatures which underlines the benefit brought by fluorine addition (Table 4 and Fig. 4). Since sediments are concentrated in the heavy fractions, results are also normalized to the 370 ◦ C+ wt%. This improvement can be explained by different factors: firstly, a lower concentration of residual asphaltene concentration (higher HDAs see Table 4), known to be the major sediment precursors [10]. This concentration decrease may limit the auto-association and sediment formation (cf. Fig. 4b) [50].
Fig. 1. IR(CO) spectra for (a) NiMo and (b) F-NiMo sulfided catalysts.
Fig. 2. Pore size distributions obtained from Hg porosimetry of used catalysts after hydroconversion tests at 430 ◦ C (a) and 390 ◦ C (b).
Fig. 3. Pore size distributions obtained from N2 physisorption for the used catalysts after hydroconversion tests at 430 ◦ C (a) and 390 ◦ C (b).
40
C. Marchal et al. / Applied Catalysis A: General 411–412 (2012) 35–43
2,5
F-NiMo
2
2,5
b
NiMo
430°C 1,5 1 0,5
NiMo
Sediments wt% in 370+
Sediments wt.% in 370+
a
390°C
F-NiMo
2
430°C 1,5 1
390°C 0,5 0
0 40
60
80
0
100
SVR conversion (wt.%)
1
2
3
Residual asphaltenes concentration (wt.%)
Fig. 4. Catalyst and temperature effects on sediment formation versus SVR conversion (a) and residual asphaltenes concentration (b).
100
HDN, HDS Asphaltenes (%)
As a consequence, the resins/asphaltenes weight ratio will be higher for the F-NiMo catalyst which is usually (but not always) a benefit for the stability [17,51]. This lower sediment formation for F-NiMo can also be correlated to the lower content of heteroelements (S, N) in asphaltenes as shown in Table 6. Indeed, asphaltenes polarity depends on heteroelements concentration, as suggested by Wattana et al. [12], and highly polar compounds are supposed to exhibit a higher tendency to self aggregation via coulombic electrostatic interactions [52]. The chemical composition of asphaltenes shows that F-NiMo catalyst improves HDS and HDN conversions (Fig. 5 and Table 6) in particular in the catalytic regime (390 ◦ C) and for the different cuts (maltene, asphaltenes and 520◦ C+ cut). Metal (Ni, V) contents in the different polar fractions (resins and asphaltenes) were not analyzed in this study but they are also supposed to change the polarity of the molecules. Purification reactions improvement may be explained by the higher hydrogenation activity of F-NiMo catalyst which contains more NiMoS active sites as shown by IR(CO) previous results. This higher hydrogenation activity of F-NiMo catalyst is also experimentally illustrated by the higher H2 consumption or lower Conradson Carbon (Table 4). HDS improvement has already been observed in the case of F-NiMo catalyst by Kim et al. [53] for dibenzothiophene and 4,6-dimethyldibenzothiophene HDS. It is also well documented in literature that in the case of heavy molecules, heteroatoms are mostly located in rings. As a consequence, the purification reactions (HDS, HDN, and HDM) occur via a hydrogenation pathway [54] rather than direct desulfurization, which underlines the key role of hydrogenation function of the catalyst in such process.
HDS
98
96
95
93
90
90
HDN 93
85 78
80 75
71
70 64
65 60 55 50 F-NiMo 430°C
F-NiMo 390°C
NiMo 430°C
NiMo 390°C
Fig. 5. HDS and HDN conversion of the asphaltenes for the F-NiMo and NiMo catalysts at 430 and 390 ◦ C.
In order to study the interactions between residual asphaltenes, SAXS measurements have been performed on the effluents obtained in the catalytic regime at 390 ◦ C (Table 5). Results suggest a slightly attractive like asphaltene–asphaltene interactions with a negative second virial coefficient (A2 = −1) in the case of NiMo catalyst and a repulsive interaction for F-NiMo catalyst (A2 = +24). The attractive interactions in the case of NiMo catalyst are consistent with the higher concentration of asphaltene and their higher
Table 4 Effects of catalyst composition and temperature on products properties.
Sediments (wt% of the total liquid) Sediment (wt% in the 370 ◦ C+ ) H2 consumption (mol) Yield(±2 wt%) 100 ◦ C− cut 100–220 ◦ C cut 220–370 ◦ C cut 370–520 ◦ C cut 520 ◦ C+ cut Conversion 520 ◦ C+ (%) Conradson Carbon (wt%) SARA fractionation 370 ◦ C+ cut (wt%) Saturates Aromatics Resins Asphaltenes Asph. conversion (wt.)% n.d., not determined.
Feed
390 ◦ C NiMo
390 ◦ C F-NiMo
430 ◦ C NiMo
430 ◦ C F-NiMo
– –
0.12(±0.02) 0.16 0.71
0.06(±0.02) 0.09 0.78
0.6(±0.04) 1.98 0.57
0.28(±0.03) 1.02 0.60
0 0 0 16 84 – 20.1
9 8 18 31 34 55 6.7
9 9 19 31 33 57 4.9
20 23 26 18 12 80 5.5
21 25 27 18 10 84 5.0
11.2 40.3 33.9 12.4 0
17.0 28.9 16.7 2.1 83(±4)
19.4 28.8 14.5 1.0 92(±2)
n.d. n.d. n.d. 2.6 79(±4)
n.d. n.d. n.d. 1.9 85(±3)
C. Marchal et al. / Applied Catalysis A: General 411–412 (2012) 35–43
41
Fig. 6. SEC profiles of asphaltenes obtained at 430 ◦ C (a) and 390 ◦ C (b) with NiMo and F-NiMo catalysts.
heteroatoms content increasing their intrinsic polarity and van der Walls interactions leading to more sediment formation. For F-NiMo catalyst, the repulsive interactions which are the consequence of a lower asphaltene concentration and an intrinsic lower polarity of residual asphaltenes, are more effective to counterbalance auto association. The carbon structure of the asphaltenes has been analyzed by 13 C NMR (Table 5). For both temperatures, the aromaticity factor (AF) and the CH2 /CH3 ratio for saturated carbons are very close for the two catalysts, suggesting similar residual asphaltenes structures. Analysis of the resins at 390 ◦ C gives the same trend, only a higher CH2 /CH3 ratio can be noticed for the F-NiMo which may mean that the effluent is enriched in naphtenic rings or contains longer aliphatic chains are present as it was not possible to distinguish between those two species. Furthermore, analysis of resins extracted from the effluent at 430 ◦ C was not possible due to instability issues. Higher HDAs conversions are reached when using the F-NiMo catalyst for both temperatures, nevertheless residual molecules have the same 13 C NMR signature which means that it is possible to depart from the trend usually observed in other studies obtained varying HDAs (or HD540 ◦ C+ ) raising the temperature [7–10]. Asphaltenes SEC profiles (Fig. 6) show that fluorine does not have a significant impact on asphaltenes apparent mass for both temperatures. These results are consistent with 13 C NMR ones which showed that the residual asphaltenes have similar carbon structures.
1,4
430°C 1,2
390°C
dXi/dlogM
1
Feed
0,8 0,6 0,4 0,2 0 100
1000
10000
100000
Mi (g/mol, eq.PS) Fig. 7. SEC profiles of asphaltenes extracted from the feed and effluents obtained after conversion at 390 and 430 ◦ C with NiMo catalyst.
significantly higher than the value obtained after the test carried out at 390 ◦ C (15 wt% for NiMo and 12.8 wt% for F-NiMo). Kinetic and thermodynamic considerations can explain this result: reactions leading to coke formation are highly temperature activated (condensation, dehydrogenation) and thermodynamic equilibrium strongly disfavors hydrogenation reaction at high temperature especially in the case of polycondensed aromatic structures. A less graphitic carbonaceous deposit on the catalyst is obtained in these conditions (390 ◦ C) as illustrated by the higher H/C molar ratios (Table 3). Fig. 7 compares the asphaltenes SEC profiles at the two temperatures for the F-NiMo catalyst. Despite a higher HDAs conversion (92 versus 85%), a higher apparent size is obtained for residual asphaltenes in the catalytic regime. Even if SEC response may be affected by the aggregation state and polarity as well, molecular size in this case is likely to be the predominant factor since polarity
3.3. Effect of temperature on effluents properties Analysis results of the used catalysts are presented in Table 3. Results show that the coke content after the catalytic tests performed at 430 ◦ C, close to 20 wt% for the two catalysts, is
Table 5 SEC, 13 C NMR and SAXS characterization of resins and asphaltenes from feed and effluents.
Asphaltenes Molecular weight (g mol−1 eq. PS.) Aromaticity factor AF % of substitution of aromatic rings As CH2 /CH3 saturated Second virial coefficient A2 Resins Molecular weight (g mol−1 eq. PS.) aromaticity factor AF % of substitution of aromatic rings As CH2 /CH3 sat
Feed
390 ◦ C NiMo
390 ◦ C F-NiMo
430 ◦ C NiMo
430 ◦ C F-NiMo
5819 49 0.70 3.3 n.d.
1361 72 0.58 2.4 −1
1540 72 0.51 2.6 24
623 85 0.46 1.6 n.d.
563 86 0.43 1.4 n.d.
2162 40 0.57 3.9
791 53 0.50 3.3
819 54 0.53 4.3
n.d. n.d. n.d. n.d.
n.d. n.d. n.d. n.d.
42
C. Marchal et al. / Applied Catalysis A: General 411–412 (2012) 35–43
Table 6 Chemical analysis of the effluents for the different catalysts and reaction temperatures. Maltene and asphaltenes (Asph.) obtained from 370 ◦ C+ fraction. S(±0.05 wt%)
NiMo – 390 ◦ C F-NiMo – 390 ◦ C NiMo – 430 ◦ C F-NiMo – 430 ◦ C
N(±0.03 wt%)
H/C
Maltene
Asph.
520 ◦ C+
Maltene
Asph.
520◦ C+
Maltene
Asph.
520 ◦ C+
1.5 0.9 2.0 1.7
3.1 2.2 2.4 2.1
1.7 1.2 2.6 2.2
0.3 0.2 0.6 0.4
1.2 0.9 1.2 1.0
0.4 0.3 0.8 0.4
1.6 1.6 1.2 1.2
0.9 0.9 0.7 0.7
1.2 1.6 0.8 0.8
of the molecules is not supposed to be strongly affected between the two temperatures as shown by the close HDS and HDN conversion levels. That means that high asphaltenes conversion can be achieved without necessarily a strong reduction of their apparent size. At this point, one must keep in mind that asphaltene molecules are defined by their solubility properties (soluble in toluene or insoluble in n-paraffins) which are function of their intrinsic polarity and interactions they may create with their surrounding medium. Therefore, their conversion is the result of all chemical modifications modifying their polarity such as the removal of groups able to create coulombic interactions like heteroatoms or saturation of polycondensed aromatic rings instead of the only reduction of their size and boiling point as conversion is usually associated to in hydrocracking of heavy hydrocarbons. Finally, one also can notice that SEC size distribution is narrower for converted asphaltenes than for asphaltenes of the feed, indicating that the polydispersity of converted asphaltenes decreases with their conversion, as already pointed out by other research teams [55,56]. 13 C NMR analysis shows that, the aromaticity factor (AF) of the asphaltenes is higher for the effluents than for the feed, especially those obtained at high reaction temperature (Table 5). The percentage of substitution of aromatic rings (As ) and the CH2 /CH3 ratio of aliphatic carbons are higher for residual asphaltenes at 390 ◦ C in agreement with results reported by Buch [57]. This result is consistent with previous SEC results and is due to dealkylation reactions promoted at high temperature involving radical ˇ scissions of aliphatic carbons. Aromatic ring opening is only possible via a hydrogenated intermediate which is not thermodynamically favored at such high temperature. On the other hand, at 390 ◦ C radical cleavage is slower and thermodynamic equilibrium is displaced towards hydrogenation so that asphaltene hydrocracking via saturation of aromatic ring and consecutive opening via carbenium ion reaction pathway should no longer be neglected and is supposed to participate to the overall conversion. This point is supported by the fact that the used F-NiMo (after HDC at 390 ◦ C) exhibits a rather high residual activity for the cyclohexane isomerisation (not presented here) which is a well known Brönsted acid site catalysed reaction. On the other hand, in the case of the NiMo, the acidic function is no longer active after the HDC tests whatever the temperature. Higher hydrogen consumptions in the catalytic regime and H/C ratio are also in agreement with this hypothesis (Tables 4 and 6). For both catalysts, the decrease of temperature dramatically reduces the sediment formation as already observed in literature [25,58]. In this case, heteroatoms concentration cannot be considered anymore to account for the stronger asphaltene interactions since S and N wt% are rather close at both temperatures (even slightly more S wt% in asphaltenes after 390 ◦ C for each catalyst). Here, sediment formation differences is more likely to be due to the differences of the carbon structures of residual molecules: highly condensed and dealkylated structures obtained at 430 ◦ C promoting aggregation phenomena via – interactions between aromatic sheets [52,59] and more hydrogenated and branched rings obtained in the catalytic regime.
4. Conclusion The aim of this work was to investigate the influence of the catalytic functions of NiMo on alumina sulfide catalysts as well as the role of temperature on the sediment formation occurring during vacuum residue hydroconversion. We studied the effect of fluorine addition for two levels of temperature: 430 ◦ C where radical chemistry is the driving process for C–C bond cleavage and 390 ◦ C where hydrocracking via hydro-dehydrogenation step and carbenium ion species prevail. F-NiMo catalyst significantly decreases the sediments formation for both temperature regimes. Addition of F increases the number of hydrogenation NiMoS sites as showed by IR spectroscopy of adsorbed CO. Heteroelement removal is thus enhanced and as a result decreases asphaltene polarity and concentration. Their auto association is then likely less favored leading to a lower sediment formation. It is worth noticing that with the F-NiMo catalyst higher asphaltene conversions are reached keeping the same carbon structure as shown by 13 C NMR, contrary to the increase of aromaticity and dealkylation usually reported when increasing residue conversion by the operating temperature. If we now consider the effect of the type of reaction pathway (radical versus catalytic one), highly condensed and dealkylated molecules are produced via the thermal pathway instead of more substituted (and bigger) hydrogenated compounds in the catalytic conditions. Moreover, because it occurs on the catalytic surface where competitive adsorption takes place, hydrocracking is more selective than thermal reactions for the conversion of asphaltenes versus resins, which keep a suitable asphaltene/resin ratio to avoid sediment formation. Acknowledgments Authors would like to warmly thank collaborators from the Physics and Analysis Division: P. Paul, D. Barrallon, N. Cellier, A. Saunier, D. Blache, C. Plassais (Distillations), A. Olivier, C. Fayolle (SEC), E Sorbier, AA. Quoineaud, L. Lemaitre and M. Vidalie (RMN and IR(CO)). Jalel M’Hamdi from Applied Chemistry and Physico-Chemistry Division for SAXS experiments, and G. Fernandes of the Catalysis & Separation Division for helpfull contribution on hydroconversion tests. References [1] J.F. Le Page, S.G. Chatila, M. Davidson, Resid and Heavy Oil Processing, Eds Technip, Paris, 1992. [2] J. Ancheyta, J.G. Speight, Hydroprocessing of Heavy Oils and Residua, CRC Press Taylor & Francis Group, 2007. [3] F. Morel, S. Kressmann, V. Harle, S. Kasztelan, in: G.F. Froment, B. Delmon, P. Grange (Eds.), Hydrotreatment and Hydrocracking of Oil Fractions, Stud. Surf. Sci. Catal., Elsevier, Amsterdam, 1997, pp. 1–16. [4] E. Furimsky, App. Catal. A: Gen. 171 (1998) 177–206. [5] J. Martinez, J.L. Sanchez, J. Ancheyta, R.S. Ruiz, Catal. – Rev. Sci. Eng. 52 (2010) 60–105. [6] M. Abu-Khader, J. Speight, Oil Gas Sci. Technol. 62 (2007) 715–722. [7] I. Merdrignac, A.A. Quoineaud, T. Gauthier, Energy Fuels 20 (2006) 2028–2036. [8] J. Ancheyta, F. Trejo, M.S. Rana, Asphaltenes: Chemical Transformation During Hydroprocessing of Heavy Oil, CRC Press Taylor & Francis Group, 2009. [9] O. Leon, E. Rogel, J. Espidel, G. Torres, Energy Fuels 14 (2000) 6–10.
C. Marchal et al. / Applied Catalysis A: General 411–412 (2012) 35–43 [10] O.C. Mullins, Energy Fuels 24 (2010) 2179–2207. [11] T.J. Kaminsky, H.S. Fogler, N. Wolf, A. Mairal, Energy Fuels 14 (1999) 25–30. [12] P. Wattana, S.H. Folger, A. Yen, M. Del Carmen Garcia, L. Carbognani, Energy Fuels 19 (2004) 101–110. [13] S. Mitra-Kirtley, O.C. Mullins, J. Van Elp, S.J. George, J. Chen, S.P. Cramer, J. Am. Chem. Soc. 115 (1993) 252–258. [14] E. Rogel, Energy Fuels 22 (2008) 3922–3929. [15] M. Sedghi, L. Goual, Energy Fuels 24 (2010) 2275–2280. [16] F.I. Nellensteyn, The Science of Petroleum, vol. 4, Oxford University Press, London, 1938. [17] G. Ali Mansoori, J. Petrol. Sci. Eng. 17 (1997) 101–111. [18] H. Seki, F. Kumata, Energy Fuels 14 (2000) 980–985. [19] I. Gawel, D. Bociarska, P. Biskupski, App. Catal. A: Gen. 295 (2005) 89–94. [20] J. Ancheyta, G. Centeno, F. Trejo, G. Marroquin, Energy Fuels 17 (2003) 1233–1238. [21] K. Le Lannic, I. Guibard, I. Merdrignac, Petrol. Sci. Technol. 25 (2007) 169–186. [22] R. Wandas, Petrol. Sci. Technol. 25 (2007) 153–168. [23] J.M. Purcell, I. Merdrignac, R.P. Rodgers, A.G. Marshall, T. Gauthier, I. Guibard, Energy Fuels 24 (2009) 2257–2265. [24] J. Bartholdy, S.I. Andersen, Energy Fuels 14 (2000) 52–55. [25] J. Bartholdy, R. Lauridsen, M. Mejlholm, S.I. Andersen, Energy Fuels 15 (2001) 1059–1062. [26] F. Derbyshire, A. Davis, M. Epstein, P. Stansberry, Fuel 65 (1986) 1233–1239. [27] L. Petrakis, D.W. Grandy, Fuel 59 (1980) 227–232. [28] F.M. Dautzenberg, J.C. De Deken, Symposium on Development of Hydroconversion Catalysts, Div. Petroleum. Chem. ACS Miami Beach, 1985, p. 8. [29] J. Ancheyta, G. Centeno, F. Trejo, J.G. Speight, Catal. Today 109 (2005) 162–166. [30] A. Stanislaus, M. Absi-Halabi, Z. Khan, Stud. Surf. Sci. Catal. 100 (1996) 189–197. [31] K.M. Sundaram, U. Mukherjee, M. Baldassari, Energy Fuels 22 (2008) 3226–3236. [32] Standard Methods for Analysis and Testing of Petroleum and Related Products, Methods IP281-398, vol. 2, J. Wiley, Institute of Petroleum, London, 1993. [33] M. Bouquet, A. Bailleul, Fuel 65 (9) (1986) 1240–1246.
43
[34] J.P. Cotton, in: P. Lindner, T. Zemb (Eds.), Neutron, X-Ray and Light Scattering, Elsevier, North Holland, 1991, pp. 19–31. [35] B.H. Zimm, J. Chem. Phys. 16 (1948) 1093–1116. [36] L. Barré, S. Simon, T. Palermo, Langmuir 24 (2008) 3709–3717. [37] L. Barré, J. Jestin, A. Morisset, T. Palermo, S. Simon, Oil Gas Sci. Technol. – Revue d’IFPEN 64 (2009) 617–828. [38] Fischer, L., Ph.D. Thesis, Modification of Hydrotreating Sulfide Catalysts by Fluorine Addition, University of Paris VI, 1999. [39] N.A. Startsev, O.V. Klimov, A.V. Kalinkin, V.M. Mastikhin, Kinet. Catal. 35 (1994) 552–558. [40] F. Maugé, J.C. Lavalley, J. Catal. 137 (1992) 69. [41] V.R. Choudhary, Ind. Eng. Chem. Prod. Res. Des. 16 (1977) 12–22. [42] Z. Sarbak, Appl. Catal. A: Gen. 159 (1997) 147–157. [43] P.O. Scokart, P.G. Rouxhet, J. Colloid Interface Sci. 86 (1982) 96–104. [44] P.J. Chupas, C.P. Grey, J. Catal. 224 (2004) 69–79. [45] A.K. Ghosh, R.A. Kydd, Catal. – Rev. Sci. Eng. 27 (1985) 539–589. [46] E. Decanio, J.W. Bruno, V.P. Nero, J.C. Edwards, J. Catal. 140 (1993) 84–102. [47] W.M. Deen, AIChE J. 33 (1987) 1409–1425. [48] X. Yang, J.A. Guin, Chem. Eng. Commun. 166 (1998) 57–79. [49] Ch. Marchal, E. Abdessalem, M. Tayakout-Fayolle, D. Uzio, Energy Fuels 24 (2010) 4290–4300. [50] E. Rogel, Langmuir 20 (2004) 1003–1012. [51] P.M. Spiecker, K.L. Gawrys, C.B. Trail, P.K. Kilpatrick, Colloids Surf. A 220 (2003) 9–27. [52] J. Murgich, Petrol. Sci. Technol. 20 (2002) 983–997. [53] H. Kim, J.J. Lee, S.H. Moon, Appl. Catal. B: Environ. 44 (2003) 287–299. [54] A. Stanislaus, A. Marafi, M. Rana, Catal. Today 153 (2010) 1–68. [55] R.H. Heck, F.T. DiGuiseppi, Energy Fuels 8 (1994) 557–560. [56] T. Gauthier, P. Danial-Fortain, I. Merdrignac, I. Guibard, A.A. Quoineaud, Catal. Today 130 (2008) 429–438. [57] L. Buch, H. Groenzin, E. Buenrostro-Gonzalez, S.I. Andersena, C. Lira-Galeanac, O.C. Mullins, Fuel 82 (2003) 1075–1084. [58] M. Tojima, S. Suhara, M. Imamura, A. Furuta, Catal. Today 43 (1998) 347–351. [59] K. Matsushita, A. Marafi, A. Hauser, A. Stanislaus, Fuel 83 (2004) 1669–1674.