You are on page 1of 8

686 Energy & Fuels 1999, 13, 686-693

Color Degradation of Hydrocracked Diesel Fuel


Isabelle Bergeron,† Jean-Pierre Charland, and Marten Ternan*,‡
Energy Research Laboratories, CANMET, Natural Resources Canada,
Ottawa, Ontario, K1A 0G1, Canada, and Department of Chemical Engineering, University of
Ottawa, Ottawa, Ontario, K1N 6N6, Canada

Received October 5, 1998

A combination of two first-order reactions that had different rate constants was found to describe
the rate of color development in the 288-343 °C diesel fraction from hydrocracked Llyodminster
atmospheric bottoms. These two reactions were the conversion of color precursors which were
either (a) unsaturated compounds or (b) heteroatom compounds. The color precursors were
correlated with the composition of the diesel fraction. The heteroatom color precursors were
correlated with the combination of sulfur and nitrogen content of the diesel fraction. The
unsaturated color precursors were correlated with its density, since density can be related to the
aromatic content. The rate constants were related to catalyst composition. The fact that all of
the parameters in the correlation were related to physical phenomena supported the hypothesis
that color development is related to these two types of color precursors. The hydrocracking
conditions were 17.3 MPa, 415 °C, 0.59 LHSV, and 570 L (H2 at STP)/L of feed [2400 scf/bbl].
Eleven different hydrocracking catalysts were examined, most of which contained two components,
a conventional CoMo:γAl2O3 component and a zeolite component (sodalite containing nickel, SOD-
Ni). The same two rate expressions were found to describe the results obtained from all of the
catalysts. The conversion of the heteroatom compounds that were color precursors was associated
with the CoMo:γAl2O3 component of the catalyst component. The conversion of the unsaturated
compounds that were color precursors was associated with the SOD-Ni catalyst component.

Introduction peroxides) was necessary. Many of the studies per-


formed since that time are consistent with his views.
The production of distillate fractions by hydrocracking
Brooks noted that the organic peroxides, which are
heavy oil residue has been the subject of several studies
necessary for color formation, could be formed by the
performed in our laboratory. Under some conditions, the
autoxidation of olefins, including some that are formed
diesel fraction from hydrocracking has been observed
from cracking processes. Subsequent work on the au-
to change color, from a clear transparent hue when it
toxidation of olefins at low temperature and other
is first produced to a dark brown or black after several
autoxidation mechanisms has been reviewed by Wal-
days of storage. This degradation in color is not gener-
lace.2 However, Brooks proposed that the organic per-
ally accompanied by any apparent change in the other
oxides have a different role in color formation than in
measurable physical or chemical properties. While the
gum formation. Organic peroxides induce polymeriza-
dark color does not affect diesel fuel performance, it is
tion of unsaturates to form gum.
undesirable because consumers tend to associate dark
The compounds in fuels, which contribute to the
colors with poor fuel grades.
development of color, have been identified by Bhan et
Discoloration and gum formation were discussed in
al.3 Ion-exchange liquid chromatography was used to
an early paper by Brooks.1 On the basis of his experi-
separate six different diesel fuel samples into acidic,
ments with cracked gasoline, Brooks concluded that
basic, and neutral fractions. The acid fraction of each
color is formed by adding acid (HCl or SO2 which
diesel fuel was much more deeply colored than the other
becomes H2SO4, or sulfonic acids formed by oxidation
two fractions. This is consistent with the observations
of thiols or disulfides) to fuels containing organic
by Brooks. A gas chromatography mass spectrometry
peroxides. He reported that air alone did not cause color,
(GCMS) analysis of the entire acid fraction identified
since pure hydrocarbons could be oxidized sufficiently
three homologous series, indoles, cycloalkylindoles, and
to form gum but still had a light color. Similarly, only
benzoindoles (predominantly carbazoles). Subsequently,
slight decoloration occurred when sulfur dioxide was
the acid fraction was separated into subfractions by
added to samples that were not permitted to contact air.
performing high-performance liquid chromatoraphy
The presence of both acid and air (required to form
(HPLC) using a strong basic mobile phase so that the
* To whom correspondence should be addressed.
fractionation would occur according to acidity. Several
† Current Address: Pan Canadian Petroleum, Calgary, Alberta, T2P

2S5, Canada. E-mail: Isabelle•bergeron@pcp.ca. (2) Wallace, T. J. In Advances in Petroleum Chemical Refining;
‡ Current Address: 147 Banning Road, Kanata, Ontario, K2L 1C5, Wiley-Interscience: New York, 1964; pp 353-407.
Canada. E-mail: ternan@ncf.ca. (3) Bhan, O. K.; Brinkman, D. W.; Green, J. B.; Carley, B. Fuel 1987,
(1) Brooks, B. T. Ind. Eng. Chem. 1926, 18, 1198-1203. 66, 1200-1214.

10.1021/ef980214q CCC: $18.00 © 1999 American Chemical Society


Published on Web 04/17/1999
Color Degradation of Hydrocracked Diesel Fuel Energy & Fuels, Vol. 13, No. 3, 1999 687

of the fractions were analyzed by GCMS. Fraction 2 R• + O2 f ROO• (2)


contained hydroperoxides and the bulk of the pyrrolic
benzologues (indole, carbazole). Fractions 3 and 4 ROO• + RH f ROOH + R• (3)
contained hydroxyaromatics and carboxylic acids, re-
spectively. Oxygenated hydrocarbons were found in all Free-radical concentrations (R•) in diesel fuels have been
of the fractions. measured by Wright et al.14 The ROOH hydroperoxide
Color can develop from autoxidation reactions be- product from eq 3 provides an additional initiation
tween the fuel and air. Hydroperoxide formation was reaction
orders of magnitude greater in bottles of jet fuel that
were loosely covered compared to bottles that were ROOH f RO• + HO• (4)
tightly capped.4 This was a clear demonstration that
oxygen from the air does react with the fuel. Unsatur- The compounds, RH, that are susceptible to oxidation
ated hydrocarbons react particularly rapidly. Early were identified by Bahn et al.3 as partially hydrogenated
studies by Thompson et al.5 and Offenhauer et al.6 aromatic hydrocarbons, hydroxyaromatics, pyrroles, in-
associated much more fuel instability with catalytically doles, and polycylic aromatics, as well as olefins from
cracked distillates (which have large concentrations of catalytic cracking. Alkyl indoles have received attention
alkenes) than with straight run distillates (which are in many studies.7,8,15-18 Oxidation is the initial step in
generally 100% alkanes). A more recent example7 fuel degradation. Subsequently, the products of oxida-
showed that after 16 months there was no change in tion can combine with either polar compounds or neutral
the color of a straight run diesel fuel, while a blend of compounds in the fuel via chemical bonding (including
90% straight run and 10% catalytically cracked stock hydrogen bonding) to form sediment. Mushrush12 has
became substantially darker. Dorbon et al.8 also showed stated that nitrogen, sulfur, and oxygen heteroatoms
that much darker colors developed in light cycle oils contribute to fuel instability. An example is the reaction
than in straight run distillates. Although saturated of polycyclic aromatics (phenalene) that react with
hydrocarbons autoxidize more slowly than either olefins oxygen to form polycyclic aromatics with a carbonyl
or molecules with benzyl C-H structures,9 some nitro- group (phenalenone). This is followed by acid-catalyzed
gen-containing and oxygen containing molecules au- condensation of phenalenone and indole to form in-
toxidize readily to hydroperoxides. dolyphenalenes,15,19 which are soluble in the fuel and
sediment precursors. These precursors combine with
Three different mechanisms explaining fuel instabil-
acids (p-thiocresol) to form highly colored salts15,19 which
ity were mentioned in a recent review by Mushrush and
precipitated as sediment. Mushrush and Speight10
Speight:10 free-radical hydroperoxide initiated oxidation,
concluded that acid-catalyzed condensation reactions
electron-transfer initiated oxidation, and soluble mac-
are involved to a significant extent. These observations
romolecular oxidation. The free-radical hydroperoxide
can be generalized to suggest that (1) oxidized com-
initiated oxidation mechanism has been discussed in the
pounds can condense with heteroatom compounds to
greatest detail. Its initiation step has been described2
form fuel-soluble oligomers and (2) the fuel-soluble
as follows:
oligomers combine with organic acids to form colored
species.
RH + O2 f R• + HO2• (1) The electron-transfer initiated oxidation mechanism
for fuel instability was proposed by Beaver et al.20,21
It could occur in the presence of metal catalysts such Various molecular types (alkyl derivatives of pyrroles,
as copper.11 Different oxidation states for the metal indoles, phenols, and polycyclic aromatics) could be
during redox catalysis have been suggested.2,9 Alterna- oxidized by this mechanism. The role of alkyl indoles
tively, it has been indicated that the metal could be an has also been interpreted in terms of this mechanism.22
electron acceptor.12 Balster, Balster, and Jones13 have The soluble macromolecule oxidation mechanism was
pointed out that the fuel species RH, in eq 1, needs to proposed by Hardy and Wechter.23,24 It proposes that
have weakly bound H atoms and that aromatics can reactive fuel monomers combine via oxidation and
function as initiators. condensation to form species with molecular weights in
A simplified mechanism for the propagation steps has the 650-1000 range and that these soluble species are
been outlined:13 sediment precursors. One study that could be grouped
in this classification is the study of visbroken short
(4) Black, B. H.; Hardy, D. R.; Beal, E. J. Energy Fuels 1991, 5, 281-
282. (14) Wright, B. W.; Kalkwarf, D. R.; Smith, R. D.; Hardy, R. D.;
(5) Thompson, R. B.; Druge, L. W.; Chenicek, J. A. Ind. Eng. Chem. Hazlett, R. N. Fuel 1985, 64, 591-594.
1949, 41, 2715-2721. (15) Pedley, J. F.; Hiley, R. W.; Hancock, R. A. Fuel 1989, 68, 27-
(6) Offenhauer, R. D.; Brennan, J. A.; Miller, R. C. Ind. Eng. Chem. 31.
1957, 49, 1265-1266. (16) Dorbon, M.; Bernasconi, Fuel 1989, 68, 1067-1074.
(7) Pedley, J. F.; Hiley, R. W.; Hancock, R. A. Fuel 1987, 66, 1646- (17) Ali, M. F.; Nizami, A.; Hamdan, A. J. Fuel Sci. Technol. Int.
1651. 1995, 13, 655-678.
(8) Dorbon, M.; Bigeard, P. H.; Denis, J.; Bernasconi, C. Fuel Sci. (18) Hughes, J. M.; Beal, E. J.; Mushrush, G. W. Fuel Sci. Technol.
Technol. Int. 1992, 10, 1313-1341 Int. 1995, 13, 413-429.
(9) Walling, C. In Free Radicals in Solution; Wiley: New York, 1957; (19) Pedley, J. F.; Hiley, R. W.; Hancock, R. A. Fuel 1988, 67, 1124-
pp 411-416. 1130.
(10) Mushrush, G. W.; Speight, J. G. Rev. Proc. Chem. Eng. 1998, (20) Beaver, B. D.; Hazlett, R. N.; Cooney, J. V.; Watkins, J. M. Fuel
1, 5-28. Sci. Technol. Int. 1988, 6, 131-150.
(11) Finseth, D. H.; Schmidt, C. E.; Link, T. A. Prepr.sAm. Chem. (21) Beaver, B. D. Fuel Sci. Technol. Int. 1991, 9, 1287-1335
Soc., Div. Pet. Chem. 1984, 29, 980-982. (22) Beaver, B.; Gilmore, C.; Veloski, G.; Sharief, V. Energy Fuels
(12) Mushrush, G. W. Fuel Sci. Technol. Int. 1992, 10, 1523-1560. 1991, 5, 274-280.
(13) Balster, L. M.; Balster, W. J.; Jones, E. G. Energy Fuels 1996, (23) Hardy, D. R.; Wechter, M. A. Energy Fuels 1990, 4, 270-274.
10, 1176-1189. (24) Hardy, D. R.; Wechter, M. A. Energy Fuels 1994, 8, 782-787.
688 Energy & Fuels, Vol. 13, No. 3, 1999 Bergeron et al.

residues by Singh et al.25 They suggested that oxidized to eliminate the components with the greatest molecular
aromatics play a role in fuel stability. weights. The first two suggestions would increase the
Color degradation is not the only fuel instability phe- removal of unsaturated compounds (partially hydroge-
nomenon. Several reviews on fuel instability emphasiz- nated aromatics, polycyclic aromatics, and olefins) and
ing sediment formation have been published.2,21,26-28 heteroatom compounds (hydroxy aromatics, pyrroles,
Deposits on heat-exchange surfaces29 and the formation and indoles) which were identified above as being
of both soluble and insoluble gums30 are other types of responsible for color formation. Such strategies, which
fuel instability. Because storage sediment has been emphasize hydrogenation, can be successful. For ex-
studied more extensively than other instabilities, more ample, an unstable thermally cracked coker distillate
of those studies are available to compare with color derived from oil sands bitumen was hydrogenated to
degradation. produce a stable product.18
An equilibrium relationship between the colored Recently, there have been reports of the effect of
species remaining in the fuel (liquid phase) and the hydrocracking on fuel stability.32 Takatusuka et al.33
sediment that precipitates (solid phase) might be ex- stated that a large hydrogen pressure would hydroge-
pected. In one study,8 data for real fuel distillates with nate radicals and suppress color. In contrast, greater
no additives showed a linear relationship between color temperatures (increased hydroprocessing severity) re-
and the logarithm of the total insolubles. In contrast, tarded hydrogenation and facilitated color. They re-
another study3 found no relationship between the two ported that by increasing the hydrodesulfurization rate,
quantities. This apparent contradiction can be explained a yellow diesel fuel could become colorless. However, if
in terms of data presented in tables by Ali, Nizami, and the rate of hydrodesulfurization exceeded a certain
Hamdan.17 Their color data for pure compound additives value, the diesel oil product would become fluorescent
was plotted as a function of the sediment that was yellow-green. Polycyclic aromatic molecules were re-
produced by those compounds. Two observations were ported to be the source of the fluorescent color. In an
apparent. (1) Initially, the color increased markedly earlier study, Takatusuka34 suggested that color is
while little sediment was formed. Subsequently, the caused by the changing energy state of π electrons in
color remained at a constant value while progressively aromatic compounds. A study on the effect of hy-
larger amounts of sediment were formed. (2) For each drotreating severity on diesel fuel stability35 indicated
additive (pure compound) there was a different maxi- that the color became lighter with increasing severity,
mum color. The color did not become darker than this except that at the greatest severity the color became
value, even though more and more sediment was darker.
formed. This suggests a relationship between color in In this work, alterations in the composition of the
the liquid phase and the amount of sediment if the catalysts used to hydrocrack the heavy oils were made.
liquid is not saturated with color bodies. In contrast, if The objective was to determine whether changes in
the liquid is saturated with color bodies, all quantities catalyst composition could influence the color stability
of sediment greater than a particular value will cor- of the diesel fraction.
respond to the same color and there will be no relation-
ship between color and sediment. Thus, the explanation Experimental Section
arising from the data of Ali et al.17 can also provide a
Hydrocracking experiments were performed using a fixed-
consistent explanation for the earlier data.3,8 Further-
bed continuous flow reactor operating in the up flow mode.
more, it indicates that color develops prior to extensive Lloydminster atmospheric residue and hydrogen gas were
sediment formation. mixed, flowed through a preheater, and then flowed through
The distillates produced in our catalytic hydrocrack- a fixed bed of catalyst contained in a 316 stainless steel tubular
ing experiments with heavy oil will contain some of the reactor (length 0.305 m, inside diameter 0.025 m, volume 155
compound types described in the above studies. Both mL). The reactor furnace consisted of three separate heating
cracking and hydrogenation reactions occur when heavy zones, which were controlled to maintain the catalyst bed at
oil is hydrocracked. It is the cracking reactions that can a constant temperature of 415 °C. The other operating
produce compounds that are susceptible to oxidation. variables were 17.3 MPa, pressure; 0.59, liquid hourly space
velocity (LHSV); and 570 L (H2 at STP)/L of feed, hydrogen
These are the compounds that must either be treated
gas rate. The reactor product flowed into pressurized receiver
or removed to avoid color formation. vessels where the liquid was accumulated after being sepa-
Petroleum refiners31 have suggested the following rated from the gas. Vapor-phase products were vented con-
strategies to minimize color formation in hydrocracked tinuously and not analyzed. The temperature of the receiver
diesel. (1) Hydrocracking reactors should be operated vessels was maintained at 150 °C to eliminate the condensa-
at low temperatures, low space velocities, and large tion of water vapor in the liquid product. The feedstock used
hydrogen partial pressures. (2) A catalyst with a large for all experiments was obtained from Consumer’s Cooperative
hydrogenation activity should be used. (3) The final Refineries Ltd. (CCRL), Regina, Saskatchewan. Feedstock
boiling point of the diesel fraction should be decreased properties are listed in Table 1.
The diesel fraction (288-343 °C) was obtained from each of
the reactor liquid samples by distillation, using a manual
(25) Singh, I. D.; Kapoor, M. P.; Ramaswamy, V. Fuel 1991, 70, 195-
199.
(26) Batts, B. D.; Fathoni, A. Z. Energy Fuels 1991, 5, 2-21. (32) Rahimi, P.; Fairbridge, C. L.; Oballa, M.; Wong, C.; Somogyvari,
(27) Mushrush, G. W. Energy Fuels 1991, 5, 258-262. A. Prepr.sAm. Chem. Soc., Div. Fuel Chem. 1998, 43, 13-15.
(28) Mushrush, G. W. Energy Fuels 1991, 5, 749-753. (33) Takatsuka, T.; Inoue, S.; Wada, Y. Catal. Today 1997, 39, 69-
(29) Jones, E. G.; Balster, L. M.; Balster, W. J. Energy Fuels 1995, 75.
9, 906-912. (34) Takatsuka, T.; Wada, Y.; Suzuki, H.; Komatsu, S. Proc. Nat.
(30) Power, A. J. Fuel 1986, 65, 133-137. Petrol. Refin. Assoc., Washington, 1991; Paper AM-91-39.
(31) Anon, National Petroleum Refiners Association Question and (35) Waynick, J. A. Prepr.sAm. Chem. Soc., Div. Fuel Chem. 1998,
Answer Session. Oil Gas J. 1993, 88-91. 43, 8-12.
Color Degradation of Hydrocracked Diesel Fuel Energy & Fuels, Vol. 13, No. 3, 1999 689

Table 1. Properties of Lloydminster Atmospheric suspension consisting of 16.4 g of NaOH in 100 mL of H2O
Residuum (4.1 M), 8.2 g of Al2O3 (Catapal SB Alumina), and 7 g of SiO2
density 15 °C (kg/m3) 999.2 (Cab-O-Sil) was placed in a sealed autoclave and heated at
API gravity 10.11 160 °C for 120 h. After quenching, the resulting crystalline
ibp - 343 °C (wt %) 4.04 powder was filtered and rinsed with 3-4 L of deionized water.
343-525 °C (wt %) 42.76 The solids were dried for 10 h at 105 °C and calcined at 350
+525 °C residue (wt %) 52.9 °C for 4 h. The sodalite crystal structure was identified by
carbon (wt %) 84.4 X-ray diffraction measurements.
hydrogen (wt %) 11.4
The resulting sodalite was subjected to several processes:
sulfur (wt %) 3.74
nitrogen (wt %) 0.45 ion exchange of the sodium ions for nickel ions, impregnated
vanadium (ppm) 123 with nickel nitrate and its conversion to form nickel oxide,
nickel (ppm) 56.7 reduction of the nickel oxide to nickel metal, and ion exchange
ash (wt %) 0.04 of unreduced nickel ions with potassium ions. The objective
microcarbon residue (wt %) 13.4 was to prepare catalysts that contained reduced nickel metal
pentane insolubles (wt %) 14.6 inside catalyst cages. Incorporating large potassium ions would
toluene insolubles (wt %) 0.08 decrease the dimensions of the cage openings so that the access
of hydrogen sulfide to the nickel metal would be decreased.
Table 2. Chemical Properties of Fresh Catalysts
Cage openings that allowed hydrogen, but not hydrogen
catalyst wt % wt % Al2O3 MoO3 CoO sulfide, to reach the nickel metal might have been an interest-
no. Ni-SOD Ni in SOD (wt %) (wt %) (wt %) ing hydrogenation catalyst for hydrocracking sulfur containing
1 10.2 0 73.3 13.8 2.7 feedstocks.
2 10.2 13.1 73.3 13.8 2.7 The nickel exchange was performed by placing 40 g of
3 9.6 25.5 73.8 13.9 2.7 sodalite in 80 g of deionized water in a 1 L boiling flask. Each
4 10.2 34.1 73.3 13.8 2.7 exchange solution was prepared by dissolving 2.48 g of Ni-
5 10.2 38.5 73.3 13.8 2.7
(NO3)2 in 800 g of deionized water (2 wt % Ni). Approximately
6 10.2 44.8 72.0 15.0 3.0
7 0 0 83.1 14.1 2.8 150 mL of the exchange solution was added to the flask per
8 5.1 38.5 78.2 13.9 2.7 hour, and then the slurry was stirred for 20 h. It was
9 20.0 38.5 63.9 13.5 2.7 subsequently filtered and washed, until the wash water had
10 25.9 38.5 56.7 14.5 2.9 a concentration of less than 5 × 10-4 mol Ni2+/L, as measured
ref cat. 0 0 80.1 16.4 3.5 by a conductance meter. The powder was then dried at 120
°C for 400 min and heated to a calcination temperature of 550
Table 3. Composition of the Product Diesel Fraction °C at a rate of 1 °C/min and then maintained at 550 °C for 4
catalyst diesel fraction (288-343 °C) h.
wt % Ni wt % nitrogen sulfur density final color The exchanged powders were impregnated with Ni(NO3)2
in sodalite Ni-sodalite (ppm) (wt %) g/mL C ) Cuns + Chet solutions of varying concentration. The solution concentrations
were chosen to produce various nickel concentrations in the
0 10 673 0.1 0.886 4.25
sodalite. A volume of solution was added that was sufficient
13 10 672 0.094 0.887 4.75
26 10 597 0.16 0.8879 5.25 to fill the pore volume of the sodalite but insufficient to permit
34 10 529 0.14 0.8887 5.25 excess liquid to be present. After standing at 40 °C for 24 h,
39 10 544 0.076 0.8881 4.25 the solids were dried at 105 °C for 10 h and calcined at 350 °C
45 10 602 0.1 0.8799 3.75 for 4 h.
0 0 226 0.113 0.8844 4.75 The reduction was performed by loading the solid powder
39 5 379 0.12 0.8815 3.75 into fine mesh containers and placing them inside a 155 mL
39 20 667 0.14 0.881 3.75 autoclave having a gas inlet and gas outlet. The conditions
39 26 834 0.18 0.8814 4.25
used to attempt reduction were 435 °C in a stream of flowing
ref cat. 468 0.051 0.8848 3.25
no cat. 509 2.55 0.9006 8 hydrogen at 3.4 L H2(STP)/min at 17.3 MPa for 4 h. Subse-
quent hydrogen chemisorption measurements did not indicate
ASTM D-1160 procedure. The color of each diesel fraction was that the nickel had been reduced, and no additional reduction
measured using the ASTM D-1500 method. Measurements steps were performed. Feeley and Sachtler37 made similar
were made daily for the first week following hydrocracking observations of the nonreducibility of metals that are within
and then weekly for a period of 8-16 weeks. The color of the zeolites.
diesel fraction was compared visually with the colored disk of A KNO3 solution equivalent to 4 wt % K+ ions was used to
the ASTM scale, which varies from 0 (colorless) to 8 (dark red), exchange the unreduced nickel ions with potassium ions. The
by 0.5 increments. In accordance with ASTM 1500, when an reduced sodalite powder was slurried in twice its weight of
exact match cannot be found, the next higher ASTM color deionized water in a 1 L boiling flask equipped with a magnetic
value is to be used, preceded by the letter L. For the purpose stirrer and a heating jacket. A slurry of 0.101 g of KNO3/g of
of this study, the letter L was interpreted as being one-half reduced sodalite in 300-400 times its weight of deionized
an increment (0.25) lower than the color value chosen. For water was added to the agitated slurry during a period of 4 h,
instance, a color value of 3.75 would be assigned to an ASTM using five different additions. The resulting nickel sodalite
color of L4.0. powders (SOD-Ni) were dried at 105 °C for 10 h. Subsequently,
In this work, the composition of the catalyst used in the the temperature was increased at 5 °C/min, and the powders
hydrocracking reactor was varied to determine its influence were calcined at 350 °C for 4 h.
on color development in the diesel fraction of the hydrocracked The Co-Mo-γAl2O3 catalyst component was prepared by
product. Ten different catalysts were prepared having the mixing 180 g of pseudoboehmite alumina (Versal 250 from
compositions shown in Table 2. The catalysts were composed Laroche Chemicals) with three aqueous solutions. A solution
of two different principal components. One was a sodalite of ammonium paramolybdate (APM), (NH4)6Mo7O24‚4H2O, was
zeolite to which nickel had been added (SOD-Ni). The other prepared from 33.2 g of APM, 74 g of deionized water, and
was Co-Mo-γAl2O3, which is one of the common catalyst
compositions used for hydrotreating petroleum fractions. (36) Carr, R. M.; Larking, P. W.; Lyon, G. L. Aust. J. Chem. 1985,
The sodalite component (hydrosodalite, 6[NaAlSiO4]‚8H2O) 21, 2555-2559.
was synthesized using the method of Carr et al.36 A colloidal (37) Feeley, J. S.; Sachtler, W. M. H. J. Catal. 1991, 131, 573-581.
690 Energy & Fuels, Vol. 13, No. 3, 1999 Bergeron et al.

varied with the catalyst composition used in the reactor.


The diesel fraction with the darkest color was obtained
when hydrocracking occurred in the absence of a
catalyst. The lightest colored product was obtained
when using the standard reference catalyst. The various
laboratory prepared catalysts produced final colors
which had intermediate values. The reference catalyst
had a similar CoMo:γ-Al2O3 composition to the labora-
tory prepared catalyst that did not contain any nickel
sodalite 0(SOD-0Ni). The difference in the performances
of these catalysts is probably related to different prepa-
ration methods. For example, in a study38 unrelated to
this one, a CoMo:γ-Al2O3 catalyst prepared in the
laboratory by a two-step method (Co added after the Mo
had been calcined on the alumina) produced more
desirable results than the reference catalyst that was
used in this study. In this study, the Co and Mo
components were added during the same catalyst prepa-
Figure 1. ASTM color verus storage time (days). Diesel
ration step.
fractions prepared using catalysts containing Co-Mo-γ-Al2O3
as the major catalyst component plus 0%, 13%, and 26% of To explain the color effects shown in Figure 1, it was
the SOD-10Ni (10% nickel in sodalite) as the minor catalyst postulated that the CoMo:γ-Al2O3 catalyst component
component. These data are represented by open circles (0%), removed heteroatom (het) compounds (sulfur and ni-
solid circles (13%), and open inverted triangles (26%), respec- trogen) and that the SOD-Ni catalyst component re-
tively. Data obtained without a catalyst (solid triangles) and moved unsaturate (uns) compounds (olefins, aromatics,
with the reference catalyst (solid squares) are also shown. free radicals). Each of these two families of compounds
18.9 g of 30% NH4OH. A solution of cobalt nitrate (CN), Co- correspond to two different groups of color precursors
(NO3)2‚6H2O, was prepared from 20.7 g of CN and 24.7 g of (colorless species present in concentrations of nuns and
deionized water. They were placed in an oven a 125 °C for 10 nhet) which could be converted into colored species. After
min to speed dissolution. Finally, 74.7 g of a third solution conversion, each of the species in the two color precursor
composed of 25 volumes of deionized water to 1 volume of 70% groups would probably contribute a different color
HNO3 was added. The three solutions were mixed with the intensity, e.g., cuns and chet. The color of a particular
alumina to form an aqueous paste. diesel fraction would be derived from each of these two
The above water-based alumina paste was added to a components, Cuns ) cunsnuns (potential component of color
hydrocarbon-based SOD-Ni paste made from 1 part Ni-SOD
from the unsaturates) and Chet ) chetnhet (potential
and 1 part Zerice oil (Imperial Oil, 0.899 specific gravity, 245-
535 °C bp range, H/C ) 1.85). Surrounding the SOD-Ni with component of color from the heteroatoms). At any time
oil would inhibit water penetration of the sodalite pores and their respective conversions would be Xuns and Xhet.
subsequent pore blocking of the SOD-Ni by alumina. The Therefore, C(t), the color of the diesel fraction (the color
combined paste was extruded and then dried at 110 °C for 18 resulting from both converted color precursors), could
h. Calcination was performed at a series of sequential tem- be represented at any time t as
perature plateaus, 190 °C for 10 h, 200 °C for 10 h, 220 °C for
20 h, 285 °C for 7.5 h, 365 °C for 7.5 h, and finally 450 °C for C(t) ) CunsXuns(t) + ChetXhet(t) (5)
4 h.
Two different sets of laboratory prepared catalysts were where Cuns and Chet are precursor color components that
studied. Their properties are shown in Table 2. Each set was exist in the fuel and become colored when converted.
composed of two components, a sodalite component that
For a first-order reaction, conversion to colored species
contained nickel x(SOD-yNi) with the balance of the catalyst
being a CoMo:γ-Al2O3 component. The value x is the percent can be represented as
of SOD-Ni in the catalyst, and y is the percent Ni in the SOD-
yNi. The quantity of nickel in Table 2 is the weight percent X(t) ) 1 - exp(-kt) (6)
nickel expressed as nickel atoms. One set of catalysts contained
10% Ni-sodalite in which the amount of nickel, y, in the where k is the first-order rate constant and t is the time
sodalite was varied 10(SOD-yNi). In the other set, the amount that elapsed after hydrocracking the diesel fraction.
of sodalite component in the catalyst was varied but the Conversions using eq 6 for both species can be substi-
amount of nickel in the sodalite x was maintained constant at tuted into eq 5 to obtain:
39%, x(SOD-39Ni). The results obtained using the laboratory
prepared catalysts were compared with those obtained using C(t) ) Cuns[1 - exp(-kunst)] + Chet[1 - exp(-khett)]
a reference catalyst, 1442-B, obtained from American Cyana- (7)
mid.
Each of the lines in Figure 1 was obtained by fitting
Results and Discussion eq 7 to the color data points. Curve fitting of color data
was performed for each catalyst used in this study,
The colors of several diesel fractions are shown in
thereby obtaining the parameters Cuns, Chet, kuns, and
Figure 1 as a function of the time that had elapsed since
khet. The plots for each of the catalysts not shown in
they were removed from the hydrocracking reactor. In
Figure 1 matched their data just as well as the correla-
each case there was a rapid initial increase in the color
intensity. Thereafter it increased more slowly until it (38) Ternan, M.; Rahimi, P. M.; Clugston, D. M.; Dettman, H. D.
had attained its maximum color. The maximum color Energy Fuels 1994, 8, 518-530.
Color Degradation of Hydrocracked Diesel Fuel Energy & Fuels, Vol. 13, No. 3, 1999 691

Figure 2. Component of the color derived from unsaturated Figure 3. Component of the color derived from compounds
compounds that are color precursors, Cuns, and from com- containing heteroatoms that are color precursors, Chet, versus
pounds containing heteroatoms that are color precursors, Chet, f(S,N), the correlation with sulfur and nitrogen content of the
versus the percent of Co-Mo-γ-Al2O3 component in the diesel fraction (eq 8).
catalyst.
As shown in eq 8, Chet was found to correlate with the
tion lines that are shown in Figure 1. When only one sulfur and nitrogen contents of the diesel fractions
exponential term was used rather than both of the produced by the various catalysts
exponential terms that are in eq 7, the resulting lines
did not match with the data very closely. The agreement Chet ) 1.4 + (8.9 × 10-4)S + (1.8 × 10-4)N (8)
between the correlation in eq 7 and the data in Figure
1 supports the suggestion that color is produced by the where S is the sulfur content (ppm) and N is the
conversion of two different groups of color precursors, nitrogen content (ppm). The resulting correlation, shown
each of which may represent a different class of mol- in Figure 3, is also consistent with the postulate that
ecules in the diesel fraction. the color precursor parameters have a real physical
The diesel fuel fraction will contain many different significance.
compounds that could contribute to color. Each com- The above empirical relationship, eq 8, is consistent
pound will have a different rate constant, kuns or khet, with the free-radical hydroperoxide initiated oxidation
and different precursor color component, Cuns or Chet. mechanism used to explain fuel instability. That mech-
For example, in this work the rate constants reported anism requires three steps. First, air reacts with certain
will be some type of an average that represents the compound classes (partially hydrogenated aromatics
compounds that have been converted. Since different and olefins present in these diesel fractions are ex-
catalysts will convert different combinations of com- amples) to form hydroperoxide compounds, via free-
pounds, these rate constants will not be invariant. radical autoxidation. Second, the hydroperoxide com-
The precursor color components, Cuns or Chet, are pounds react via acid catalysis with polar or neutral
shown in Figure 2 as a function of the Co:Mo:γ-Al2O3 compounds to form oligomers or polymers. The type of
content in the catalyst. The other catalyst component compound that reacts with the hydroperoxide deter-
was SOD-39Ni. Chet decreased (from 3.6 at 74% Co:Mo: mines whether the condensation product is colored.
γ-Al2O3 to 2.5 at 100% Co:Mo:γ-Al2O3) and Cuns in- Compounds containing S and N heteroatoms are polar.
creased (from 0.8 at 74% Co:Mo:γ-Al2O3 to 2.0 at 100% They can also form color bodies. Therefore, the empirical
Co:Mo:γ-Al2O3). There is some physical significance to correlation in eq 8 is consistent with one of the estab-
these parameters. As the CoMo:γ-Al2O3 component of lished mechanisms used to describe fuel instability.
the catalyst increased, heteroatom removal would be The other parameter, the precursor color component
expected to increase and therefore the color precursors Cuns, was attributed to unsaturates (olefins, aromatics,
remaining in the diesel fraction that are attributed to and free radicals). Since, no direct measurement of these
heteroatoms, Chet, would be expected to decrease. The quantities was made, an indirect measurement (density)
decrease in Chet, shown in Figure 2, is consistent with was used to develop a correlation. For equivalent boiling
these expectations. Correspondingly, the catalyst so- point, aromatic molecules have greater densities than
dalite content decreased when the CoMo:γ-Al2O3 cata- aliphatic molecules. Since all the diesel fractions in this
lyst content increased. As the Co:Mo:γ-Al2O3 content of study had the same boiling-point range, changes in
the catalyst increased, there would be less sodalite density may be one indication of changes in unsaturates
(SOD-39Ni) component in the catalyst and therefore (aromatics). On this basis, Cuns was correlated with the
Cuns, the color precursors remaining in the diesel measured density of the diesel fraction, as shown in
fraction that are attributed to unsaturates, would be Figure 4. The greater the value of Cuns, the greater the
expected to increase. This was observed in Figure 2. quantity of unsaturates (as measured by greater density
The association of color with heteroatoms has been or aromatics). Thus, the correlation in Figure 4 is
suggested frequently in the literature on fuel stability. consistent with Cuns being representative of a physical
692 Energy & Fuels, Vol. 13, No. 3, 1999 Bergeron et al.

component increases. A discussion of these results must


account for the fact that the diesel fractions are a
mixture of compounds and that each compound in the
mixture has its own particular rate constant. The value
of kuns in Figure 5 is a combination of rate constants
for all of the compounds in the mixture that are
converted. Compounds that are converted easily will
have fast rates and large rate constants. The compounds
that are not converted will not contribute to kuns. Thus,
a small value of kuns indicates that compounds which
react slowly have been converted, while a large value
indicates that only the faster reacting compounds have
been converted.
When there was no nickel in the sodalite, the kuns
value was quite small (open circle in Figure 5). It can
be compared to a laboratory prepared catalyst that
contained only the Co-Mo-alumina component (solid
Figure 4. Component of the color derived from unsaturated circle). Adding 10% sodalite (with no nickel) to the
compounds that are color precursors, Cuns, versus the density catalyst decreased the kuns value substantially while
of the diesel fraction (g/mL). causing a comparatively smaller increase in the khet
value. This suggests that adding a SOD-Ni component
to the catalyst caused some of the slower reacting
unsaturated species to be converted.
The effect of adding sodalite without nickel to the
catalyst can be explained in terms of hydrogen-transfer
reactions. Acid catalysts are known to facilitate hydrogen-
transfer reactions, and zeolites such as the hydrogen-
sodalite used in this study are acid catalysts. For
example, previous work in our laboratory39 demon-
strated that when mordenite (a zeolite) was added to
catalysts, the hydrodemetallization reactivity increased.
Hydrogen addition is known to be the first step in
hydrodemetallization.40 A plausible explanation for that
observation is that the mordenite zeolite facilitated
hydrogen transfer. Similarly, the sodalite in these
catalysts might aid the transfer of hydrogen from a
hydrogen-rich compound in the diesel fraction to a color
precursor compound. In catalytic cracking reactions, it
is known that acidic sites in zeolites can transfer
Figure 5. Rate constants kuns (O) and khet (0) (s-1) versus hydrogen from olefins and naphthenes to aromatics and
percent nickel in the sodalite catalyst component 10(SOD-yNi). paraffins.41 Since 200 ppm of a compound can cause
The solid symbols represent a laboratory prepared catalyst color,17 even a small amount of hydrogen transfer could
that contained only the Co-Mo-alumina component.
have a beneficial effect. The observation that increasing
the amount of SOD-39Ni component (and decreasing the
quantity and not just a mathematical parameter that
CoMo:γ-Al2O3 component) in the catalyst caused a
arises from a correlation.
lighter color, and a decrease in the Cuns parameter
The increase in Cuns in Figure 2 as the SOD-39Ni (Figure 2) is consistent with this hypothesis.
component of the catalyst decreases (Co:Mo:γ-Al2O3 Initially, as nickel is added to the sodalite, the nickel
component increases) is also consistent with more will become incorporated into the sodalite structure,
unsaturates remaining unaltered. That would be ex- thereby blocking some of the hydrogen transfer sites.
pected if the SOD-39Ni component of the catalyst caused As the hydrogen transfer sites are blocked, fewer
the saturation. Again, this suggests that the precursor unsaturates will be altered and the value of kuns would
color parameters have physical meaning. be expected to increase, indicating that only the faster
The rate constants for the series of catalysts in which reacting species were converted. This is indicated in
the percent nickel y in the SOD was varied while the Figure 5 at the smaller % Ni values. Eventually no more
Co-Mo-alumina content and the SOD-yNi content of nickel can be accommodated within the sodalite struc-
the catalysts remained constant are shown in Figure 5 ture. The nickel content of the catalyst that exceeds this
(open symbols). The rate constant for heteroatom re- value will be comparatively easy to reduce to metallic
moval, khet (open squares), remained virtually constant. nickel and then to sulfide to nickel sulfide in the
A constant value would be expected because the catalyst hydrogen/hydrogen sulfide environment of the hydro-
component for heteroatom removal, Co-Mo-alumina,
was constant. (39) Minja, R. J. A.; Ternan, M. Energy Fuels 1991, 5, 117-122.
In contrast, in Figure 5 the rate constant for removal (40) Quann, R. J.; Ware, R. A.; Hung, C. W.; Wei, J. Adv. Chem.
Eng. 1988, 4, 95-259.
of unsaturates, kuns, was not invariant. It goes through (41) Bhatia., S. Zeolite Catalysis: Principles and Applications; CRC
a maximum as the percent nickel y in the SOD-yNi Press: Boca Raton, FL, 1988; p 83.
Color Degradation of Hydrocracked Diesel Fuel Energy & Fuels, Vol. 13, No. 3, 1999 693

is not just an arbitrary parameter from a mathematical


correlation. The solid symbols in Figure 6 are data
obtained using the commercial reference catalyst. Since
the reference catalyst did not include a sodalite com-
ponent, it can be compared to the laboratory prepared
catalyst that did not contain any of sodalite component.
The smaller values for kuns obtained with the com-
mercial catalyst indicate superior performance than the
laboratory prepared catalyst. As suggested previously,
this difference was probably related to differences in
catalyst preparation procedures rather than differences
in catalyst composition.
In summary, an overview of the color formation
phenomena can be suggested. The final color that forms
in hydrocracked diesel fractions can be attributed to the
amounts of two separate types of color precursors. One
type is the unsaturated compounds (Cuns), and the other
type is the compounds containing heteroatoms (Chet).
Figure 6. Rate constants kuns (O) and khet (0) (s-1) versus These color precursors were correlated with properties
percent of the sodalite component x(SOD-39Ni) in the catalyst. of the diesel fraction. Chet was correlated with the sulfur
The solid symbols represent the Co-Mo-alumina reference and nitrogen contents of the diesel fraction, while Cuns
catalyst.
was correlated with its density, since density can be an
processing reactor. NiS (p-type) can be a weak catalyst indication of aromatic content.
for the dissociation of hydrogen molecules into hydrogen Both rate constants were related to catalyst composi-
atoms. Additional hydrogen dissociation would alter tion. The amount of the CoMo:γ-Al2O3 catalyst compo-
more unsaturated molecules. With large amounts of nent that influences heteroatom removal was related
nickel in the sodalite, more hydrogen dissociation will to khet. The amount of the SOD-Ni catalyst component
occur and kuns will decrease since the slower reacting was related to kuns. The SOD-Ni component influences
species are being converted when the additional hydro- either hydrogen transfer (the number of SOD sites that
gen is available. This is observed in Figure 5 when there are not blocked by Ni) or hydrogenation (the number of
are large amounts of nickel in the SOD-yNi component Ni atoms that exceed the number of nickel atoms that
of the catalyst. can interact with the sodalite). The color at any time
For the other series of catalysts, the amount of nickel after the diesel fraction had been obtained from the
in the sodalite was maintained constant at 39% while processing unit was described in terms of four param-
the fraction of the SOD-39Ni component in the catalyst eters. Each of the four parameters, Cuns, Chet, kuns, khet,
increased. The rate constants for this series are shown were correlated with the following measurable quanti-
in Figure 6. The rate constant for unsaturates, kuns, ties (density, sulfur and nitrogen content, SOD catalyst
decreased as the amount of the SOD-39Ni component sites, and CoMo:γ-Al2O3 catalyst sites). These correla-
in the catalyst increased. This indicates that more of tions are consistent with the hypothesis that color in
the unsaturated molecules are being altered as the diesel fractions is caused the presence of two different
catalyst component causing hydrogen transfer and groups of color precursors.
hydrogen dissociation increased. This is also consistent
with the value of kuns having a physical significance.
Acknowledgment. The authors are grateful to Bill
The rate constant for heteroatom removal, khet, in-
Smith for sending the feedstock sample and for his
creased as the amount of SOD-39Ni component in the
encouragement. Funding from the Program for Energy
catalyst increased (CoMo:γ-Al2O3 component in the
Research and Development (PERD) and from the Natu-
catalyst decreased). The increase in khet indicates that
ral Sciences and Engineering Research Council was
faster reacting compounds are being converted. Again,
greatly appreciated.
this provides evidence that the rate constant has a
physical significance (even if it is not invariant) and it EF980214Q

You might also like