1232

Energy & Fuels 2000, 14, 1232-1239

Oxidative Desulfurization of Light Gas Oil and Vacuum

Gas Oil by Oxidation and Solvent Extraction
Shujiro Otsuki, Takeshi Nonaka, Noriko Takashima, Weihua Qian,
Atsushi Ishihara, Tamotsu Imai, and Toshiaki Kabe*
Department of Chemical Engineering, Tokyo University of Agriculture and Technology,
Tokyo 184-8588, Japan
Received May 15, 2000. Revised Manuscript Received September 5, 2000

The oxidation of model sulfur compounds (thiophene derivatives, benzothiophene derivatives,

and dibenzothiophene derivatives), straight run-light gas oil (SR-LGO, S: 1.35 wt %), and vacuum
gas oil (VGO, S: 2.17 wt %) were conducted with a mixture of hydrogen peroxide and formic
acid. The thiophene derivatives with 5.696 to 5.716 electron densities on the sulfur atoms could
not be oxidized at 50 C. Benzo[b]thiophene with 5.739 electron density and other benzothiophene
and dibenzothiophenes with higher electron densities could be oxidized. The sulfur compounds
in SR-LGO and VGO appeared to be oxidized to a detectable levels (c.a., 0.01 wt % S) by GCFPD analysis. The IR spectra of oxidized SR-LGO and VGO showed that sulfones were formed
by oxidation. The removal of sulfur compounds by extraction became more effective for the oxidized
samples than for the original samples. Lighter sulfur compounds were preferentially extracted.
The extraction efficiencies of solvents, i.e., N,N-dimethylformamide (DMF), acetonitrile (ACN),
methanol, varied greatly. The most effective solvent for the removal of sulfur compounds was
DMF. The recovery of oil was, however, lowest with DMF.

Introduction
A new method to reduce the sulfur content in fuel oil
is needed today to meet the future legislation on a sulfur
content, e.g., lower than 50 ppm. The conventional
method for reducing sulfur is catalytic hydrodesulfurization (HDS) under severe conditions. It has been
reported that methyl substitution at 4 and 6-positions
of dibenzothiophene (DBT) remarkably retards the rate
of HDS and, thus, that 4-methyldibenzothiophene (4MDBT) or 4,6-dimethyldibenzothiophene (4,6-DMDBT)
is very difficult to convert due to its steric-hindrance
even under deep desulfurization conditions.1,2 Namely,
the reactivities of DBTs decrease in the order of DBT
> 4-MDBT > 4,6-DMDBT.
Various studies on oxidative desulfurization (ODS)
have been reported.3-20 The reactivity of sulfur com(1) Kabe, T.; Ishihara, A.; Qian, W. Hydrodesulfurization and
Hydrodenitrogenation; KODANSHA WILEY-VCH: New York, 1999.
(2) Kilanowski, D. R.; Teeuwen, H.; de Beer, V. H. J.; Gates, B. C.;
Schuit, G. C. A.; Kwart, H. J. Catal. 1978, 55, 129.
(3) Ford, J. F.; Young, V. O. Prepr. Div. Petro. Chem., Am. Chem.
Soc. 1965, 10(2), C-111.
(4) Heimlich, B. N.; Wallace, T. J. Tetrahedron 1966, 22, 3571.
(5) Paybarah, A.; Bone, R. L.; Corcoran, W. H. Ind. Eng. Chem.
Process Des. Dev. 1982, 21, 426.
(6) Attar, A.; Corcoran, W. H. Ind. Eng. Chem. Prod, Res, Dev. 1978,
17(2), 102.
(7) Aida, T. Catalysts & Catalysis 1995, 37(3), 243.
(8) Zannikos, F.; Lois, E.; Stournas, S. Fuel Process. Technol. 1995,
42, 35.
(9) Smetana, R. D. US Patent 3 595 778, 1971.
(10) Takeyama, T. JP Patent 290 793, 1987.
(11) Shiraishi, Y.; Hara, H.; Hirai, T.; Komasawa, I. Ind. Eng. Chem.
Res. 1999, 38, 1589.
(12) Shiraishi, Y.; Hirai, T.; Komasawa, I. Ind. Eng. Chem. Res.
1998, 37, 203.
(13) Guth, E. D.; Diaz, A. F. US Patent 3 847 800, 1974.
(14) Guth, E. D.; Arledge, N. L. US Patent 3 919 402, 1975.

pounds for oxidation is increased with the increase of

electron density on sulfur atom. The reactivities of DBT
derivatives are influenced by the electron donation of
substituted methyl groups, and; therefore, the reactivity
decreases in the order of 4,6-DMDBT > 4-MDBT >
DBT, reversing the order of reactivities for HDS.7 After
oxidation, the sulfur compounds are transformed to
sulfones. The extraction of sulfones is considered to be
useful method for removal.13-17 Aida et al. have reported
the removal of sulfur compounds to a very low level from
fossil fuels by oxidation, distillation, extraction and
adsorption. They reduced the sulfur content of light gas
oil from 0.56% to 0.018%. After adsorption with alumina, the sulfur content decreased to 0.0038% after
oxidation and extraction with N,N-dimethylformamide.
However, the extent of the oxidation and removal of
oxidized sulfur compounds was not reported.7,18-20
Herbstman et al. have reported that the sulfur content
of a 288 C+ Atmospheric Reduced Arabian Crude was
decreased to 1.48% from 2.6%. That is, 43% desulfurization was achieved by selective oxidation followed by
a high temperature thermal or KOH treatment.21
As mentioned above, various ODS methods have been
reported. Nevertheless, the extent of the oxidation of
(15) Tam, P. S.; Kittrell J. R. US Patent 4 485 007, 1984.
(16) Tam, P. S.; Kittrell, J. R.; Eldridge, J. W. Ind. Eng. Chem. Res.
1990, 29(3), 321.
(17) Tam, P. S.; Kittrell, J. R.; Eldridge, J. W. Ind. Eng. Chem. Res.
1990, 29(3), 324.
(18) Aida, T.; Yamamoto, D.; Sakata, K. Trans. Mater. Res. Soc. Jpn.
1993, 18A, 391.
(19) Aida, T.; Yamamoto, D. Prepr. Am. Chem. Soc. Div. Fuel Chem.
1994, 39(2), 623.
(20) Aida, T.; Funakoshi, I. JP Patent 197 035, 1995.
(21) Herbstman, S.; Patel, J. Am. Chem. Soc., Div. Petro. Chem.,
Prep. 1982, 27, 826.

10.1021/ef000096i CCC: $19.00 2000 American Chemical Society

Published on Web 10/14/2000

Oxidative Desulfurization of Gas Oil

Energy & Fuels, Vol. 14, No. 6, 2000 1233

Table 1. Properties of Feed SR-LGO and VGO

feed

SR-LGO

sulfur (%)
nitrogen (%)
density at 15 C (g/mL)
IBP (C)
5 vol % (C)
10 vol % (C)
20 vol % (C)
30 vol % (C)
50 vol % (C)
70 vol % (C)
80 vol % (C)
90 vol % (C)
95 vol % (C)
EP (C)

1.35
0.011
0.8580
245.5
275.5
284.5
300.5
313.5
329.5
356
367.5
375.5

VGO
2.17
0.0518
0.9161
206.5
378.0
407.5
424.5
449.0
474.0
488.5
508.5
557.0

sulfur compounds and the existence of difficult-toextract sulfur compounds have not been reported. It is
desirable to find whether the incomplete removal of
sulfur compounds from oil is attributed to incomplete
oxidation of sulfur compounds involved or to ineffective
extraction.
In this study, the relationship between the electron
densities of sulfur atoms and reactivities was investigated for the oxidation of model sulfur compounds with
hydrogen peroxide and formic acid. Then, the oxidation
of real feeds, i.e., SR-LGO and VGO, was carried out to
determine the extent of the oxidation of sulfur compounds and the extent of sulfur removal by extraction.
Experimental Section
Materials. The feeds used in this study were SR-LGO and
VGO, and their properties are given in Table 1. The model
sulfur compounds, i.e., methyl phenyl sulfide, thiophenol,
diphenyl sulfide, DBT, benzo[b]thiophene (BT), 2-methylthiophene (2-MT), 2,5-dimethylthiphene (2,5-DMT), thiophene
(commercial GR grade), 31% H2O2, formic acid and a mixture
of cis- and trans-decahydronaphthalene (decalin) were obtained from Kishida Chemicals. The concentration of H2O2 was
determined by iodometric titration before use.3,22 DBT derivatives including 4-MDBT, 4,6-DMDBT, dibenzothiophene-1,
1-dioxide (DBT sulfone), 4-methyldibenzothiophene-1,1-dioxide
(4-MDBT sulfone), 4,6-dimethyl- dibenzothiophene-1,1-dioxide
(4,6-DMDBT sulfone) were synthesized by reported
methods.23-26 Polar solvents for extraction, i.e., N,N-dimethylformamide (DMF), acetonitrile, methanol, dimethyl sulfoxide
(DMSO), and sulfolane (commercial GR grade), were obtained
from Kishida Chemicals.
Oxidation of Model Sulfur Compounds. The oxidation
of model sulfur compounds was conducted in a flask with a
magnetic stirrer. A 100 mL portion of decalin solution containing 10 representative model sulfur compounds and 5 mL of
formic acid were added to the flask. The mixture was stirred
and heated to 50 C, and, then, 10 mL of 31 wt % H2O2 was
added. H2O2 /sulfur ratio was 160 mol/mol. The reaction was
continued for 360 min.
Oxidation and Extraction of SR-LGO. The oxidation of
SR-LGO was conducted in a glass flask. An excess amount of
H2O2 and formic acid was used to oxidize the sulfur compounds
thoroughly. The first, 800 mL of SR-LGO was charged, and,
then, 200 mL of 31 wt % H2O2 and 35 mL of formic acid were
(22) Ali, P.; Russel, L. B.; William, H. C. Ind. Eng. Chem. Prod, Res,
Dev. 1982, 21, 426.
(23) Gilman, H.; Jacoby, A. L. J. Org. Chem. 1938, 4, 108.
(24) Campaigne, E.; Hewitt, L.; Ashby, J. J. Heterocyclic Chem.
1969, 6, 553.
(25) Gredil, R.; Lucken, E. A. C. J. Am. Chem. Soc. 1965, 87, 213.
(26) Gilman, H.; Esmay, D. L. J. Am. Chem. Soc. 1952, 74, 2021.

added. The mixture was stirred for 12 h at 25 C. The mixture

was heated to 50 C, added with 160 mL of H2O2 and 30 mL
of formic acid and stirred for 12 h at 50 C. The last 120 mL
of H2O2 and 40 mL formic acid were added, and the mixture
was further stirred for 12 h before it was cooled to room
temperature.
The extraction of SR-LGO was conducted by using a 300
mL separatory funnel. 150 mL of SR-LGO and 150 mL of
solvent were charged, and, then, the mixture was shaken for
10 min at room temperature before the oil and solvent layers
were separated.
Oxidation and Extraction of VGO. The oxidations of
VGO were conducted in the same way as SR-LGO. The first,
200 mL of H2O2 and 35 mL of formic acid were added to 800
mL of VGO and stirred for 12 h at 25 C. After that, the
mixture was heated to 70 C, 160 mL of H2O2 and 30 mL of
formic acid were further added and stirred for 12 h at 70 C.
The last, 120 mL of H2O2 and 30 mL formic acid were added
and stirred for 12 h, and, then, the mixture was cooled to room
temperature. The oil layer was washed with 150 g of water 3
times and evacuated at 10-3 Torrs at 100 C for 0.5 h to dry
the sample.
The extractions of VGO were conducted by mixing in a twoneck glass flask with a stirrer. 100 mL of VGO sample and
100 mL of solvent were combined with stirring at 50 C, and,
then, oil and solvent were separated. The recovered oil was
further extracted with 150 mL of solvent up to 10 times.
-Al2O3 Adsorption of Oxidized SR-LGO after Solvent
Extraction. Adsorption of sulfur compounds remained in oil
layer after solvent extraction using -Al2O3 was carried out
in a glass flask with a stirrer. 60 g of oil layer and 10 g of
-Al2O3 were mixed at room temperature for 10 h, and, then,
oil and adsorbent were separated by filtration. The -Al2O3
after adsorption was extracted with CCl3 by Soxhlet extraction
method to recover the adsorbed sulfur compounds.
Analysis. The reaction products were analyzed by a gas
chromatography, i.e., GC/MS (Shimadzu GCMS-QP5050) and
GC-FID, GC-FPD (Shimadzu GC-14A). An identification of the
sulfur compounds in SR-LGO was carried out using a GC-AED
as reported previously.27 The total sulfur contents of SR-LGO
and VGO were analyzed by XRF (Rigaku Co.Ltd.). The total
sulfur contents of VGO were also analyzed by a GC-FPD
(Shimadzu GC-14A). Infrared spectroscopy analyses of SRLGO and VGO were carried out by using FT/IR-5300 (Japan
Spectroscopic Co. Ltd.).
Calculations of Electron Density on Sulfur Atom. The
Computer Aided Chemistry (CAChe) worksystem provided by
CAChe Scientific Inc. was used to calculate the electron density
of the sulfur atom in the sulfur compounds by using the Sony
Tektronixs CAChe system (CAChe, 1995) and the Molecular
Orbital Package (MOPAC, Version 6.00). The PM3 (Modified
Neglect of Diatomic Overlap, Parametric Method 3) semiempirical Hamiltonian developed by Stewart (1989) was employed
to solve the Schrodinger equation to calculate the optimum
geometry and electronic properties of the sulfur compounds
using the standard parameters.

Result
Oxidation of Model Sulfur Compounds. The
electron densities on the sulfur atom of sulfur compounds, i.e., thiol, sulfides, thiophenes, and their benzene and/or dimethyl-substituted derivatives, are shown
in Table 2. The electron densities are in the range of
5.696 to 5.915. Figure 1 shows the relationships between
the sulfone yields of model compounds and reaction
time. The electron densities of thiophenol and two
(27) Kabe, T.; Ishihara, A.; Tajima, H. Ind. Eng. Chem. Res. 1992,
31, 1577.

1234

Energy & Fuels, Vol. 14, No. 6, 2000

Otsuki et al.

Table 2. Electron Densities on Sulfur Atom and Rate Constants

Figure 2. Relationships between the rate constants, k of

model compounds, and the electron densities.

Figure 1. Relationships between the sulfone yields of model

compounds and reaction time.

sulfides, methyl phenyl sulfide, and diphenyl sulfide

were relatively higher values among these sulfur compounds. The electron densities of thiophene derivatives
were changed in the range from 5.696 (thiophene) to
5.760 (4,6-DMDBT). 2,5-DMT, 2-MT, and thiophene
have relatively lower electron densities, 5.716, 5.706,
and 5.696, respectively. The oxidation of these compounds did not take place. BT (electron density ) 5.739)
and other model compounds with higher electron densities than BT were oxidized to form corresponding

sulfones. The oxidation of every compound was treated

as a pseudo-first-order reaction, and the oxidation rate
constants were calculated and summarized in Table 2.
These apparent rate constants decreased in the order:
methyl phenyl sulfide > thiophenol > diphenyl sulfide
> 4,6-DMDBT > 4-MDBT > DBT > BT. The relationships between the rate constants of the oxidation of
these model sulfur compounds to corresponding sulfones, and the electron densities are shown in Figure
2. This result indicates that rate constant increases with
the increase of electron density.
Oxidation and Extraction of SR-LGO. The mixture of SR-LGO and oxidant became three layers after
oxidation: oil layer (top), residual layer (bottom), and
aqueous layer (middle). The recovery and the sulfur

Oxidative Desulfurization of Gas Oil

Energy & Fuels, Vol. 14, No. 6, 2000 1235

Figure 3. GC-FPD chromatograms of oil layers of SR-LGO before and after oxidation.
Table 3. Recoveries and Sulfur Contents in Three
Fractions after Oxidation of SR-LGO and VGO
distribution
of SR-LGO in

recovery of oil
(%)

sulfur content
(wt %)

oil layer
residual layer
aqueous layer

89.8
9.0
1.2

0.65
8.8
NDa

distribution
of VGO in

recovery of oil
%)

sulfur content
(wt %)

oil layer
residual layer
aqueous layer

87.9
11.2
0.90

1.48
6.5
NDa

ND: non determination.

contents of these layers are summarized in Table 3. The

recovery of the oil layer after oxidation was 89.8%. The
sulfur content of oil layer of the oxidized sample was
reduced to 0.65 wt % from 1.35 wt % in the original SRLGO. In addition, the color of oxidized SR-LGO was
darker than fresh samples.
Figure 3 shows the GC-FPD chromatograms of oil
layers before and after oxidation. The GC peaks of sulfur
compounds in oil layers of oxidized samples were shifted
to the heavier direction. The content of sulfur was
reduced after oxidation, as determined by GC-FPD and
XRF analyses.
The recoveries of oil layers from the extractions of
unoxidized and oxidized SR-LGO with DMF, acetonitrile, methanol, DMSO, and sulfolane are summarized
in Table 4. The recoveries of oxidized SR-LGO were in
the range 1.2-6.1% lower than that of unoxidized SRLGO, and the extent decreased in the order: acetonitrile
> methanol > DMSO > DMF > sulfolane. The sulfur
contents of oxidized SR-LGO after extraction with the
five solvents were in the range of 0.10 to 0.18 wt %,
whereas those of unoxidized SR-LGO were 0.57 to 1.13

Table 4. Recoveries of Oils and Solvent Layers after

Extraction of SR-LGO and Sulfur Contents

DMF
extraction
SR-LGO
oxidized SR-LGO
acetonitrile
extraction
SR-LGO
oxidized SR-LGO
methanol
extraction
SR-LGO
oxidized SR-LGO
DMSO
extraction
SR-LGO
oxidized SR-LGO
sulfolane
extraction
SR-LGO
oxidized SR-LGO

total yielda
(%)

recovery of oilb
(%)

sulfur contentc
(%)

99.7
99.5

88.2
85.9

0.57
0.11

99.7
99.8

96.2
90.1

1.13
0.10

97.8
99.5

95.8
90.5

1.13
0.18

99.7
99.4

95.4
91.1

1.06
0.11

99.7
99.2

96.0
94.8

1.08
0.12

a {(oil + solvent recovered)/(oil + solvent fed)} 100. b {(oil

recovered)/(oil fed)} 100. c XRF analysis.

wt %. In the extraction of oxidized SR-LGO, all of the

solvents except for methanol were the effective solvents
to extract the oxidized sulfur compounds up to c.a.
0.10%. In the extraction of unoxidized SR-LGO, DMF
was the most effective solvent among five polar solvents.
Nevertheless, the recovery of oil layer extraction with
DMF was the lowest.
The concentrations of DBT derivatives, i.e., DBT,
4-MDBT, 4,6-DMDBT in unoxidized SR-LGO and DBTO2, 4-MDBTO2 and 4,6-DMDBTO2 in oxidized SRLGO, are shown in Table 5. In the extraction of oxidized
SR-LGO, the concentrations of DBT derivatives decreased to less than one ppm level, whereas the total

1236

Energy & Fuels, Vol. 14, No. 6, 2000

Otsuki et al.

Table 5. Concentrations of Dibenzothiophene Derivatives in Oil Layers

unoxidized
SR-LGO

feed
DMF
acetonitrile
methanol
DMSO
sulfolane
a

oxidized
SR-LGO

dipole
moment ()

dielectric
constant ()

DBT
[ppm]a

4-MDBT
[ppm]a

4,6-DMDBT
[ppm]a

total S
[ppm]b

DBTO2
[ppm]a

4-MDBTO2
[ppm]a

4,6-DMDBTO2
[ppm]a

total S
[ppm]b

3.24
3.53
2.87
4.06
4.81

36.71
35.94
32.66
46.45
43.26

1620
26
165
315
28
34

3350
71
567
769
169
163

525
16
101
112
48
40

1.35
0.57
1.13
1.13
1.06
1.08

19
<1
1
<1
<1
<1

67
<1
<1
<1
1
<1

19
<1
<1
<1
<1
<1

0.65
0.11
0.10
0.18
0.11
0.12

Determined by GC-FPD analysis. b Determined by XRF analysis.

Figure 4. IR spectra of SR-LGO before and after oxidation, extraction with acetonitrile, and Soxhlet-extracted compounds adsorbed
on -Al2O3.

sulfur content decreased to c.a. 0.1%. The DBT derivatives became easier to remove by the increase in polarity
after oxidation to form sulfone.
Figure 4 shows the IR spectra of SR-LGO before and
after oxidation and extraction with acetonitrile. The
specific infrared absorption of sulfone (1165 cm-1, 1289
cm-1)28 was observed indicating that sulfone was formed
by oxidation, and this sulfone was effectively removed
with extraction. The colors of oxidized SR-LGO was
darker than fresh sample but became similar to that of
fresh sample after extraction.
-Al2O3 Adsorption of Oxidized SR-LGO after
Solvent Extraction. Figure 5 shows the four GC-FPD
chromatograms such as (A) unoxidized SR-LGO, (B)
oxidized SR-LGO, (C) the oil layer of the oxidized SRLGO after extraction with acetonitrile, and (D) Soxhletextracted compounds adsorbed on -Al2O3 in SR-LGO
after acetonitrile extraction. The sulfur compounds
remaing in SR-LGO after acetonitorile extraction (peak
D) shifted to the heavier direction. This result indicates
that most of the remaining sulfur compounds were
oxidized to sulfones.
(28) Nakanishi, K. IR Absorption Spectroscopy, Nankodo, Japan
1960, p 58.

In addition, the IR spectrum of Soxhlet-extracted

compounds adsorbed on -Al2O3 is shown in Figure 4.
The spectrum shows that the sulfur compounds remaining in SR-LGO after acetonitorile extraction is sulfones.
Oxidation and Extraction of VGO. The product of
VGO oxidation became three layers; oil layer (top),
residual layer (bottom), and aqueous layer (middle). The
recoveries and the sulfur contents are summarized in
Table 3. The recovery of oil layer was 87.9%. The sulfur
content of oil layer was 1.48 wt % compared with 2.17
wt % of the original VGO.
The extractions of sulfur compounds from the oxidized
oil layer and the original VGO were conducted with
acetonitrile, methanol, and DMF. Figure 6 shows the
changes in the recoveries of oil layers as a function of
number of extraction before and after oxidation. The
recoveries of oil layers decreased in the order: methanol
> acetonitrile > DMF. The recoveries of oil layers
decreased to about 10-15% lower than before oxidation.
Figure 7 shows changes in the sulfur concentrations
in oil layers. The sulfur contents are summarized in
Table 6. The sulfur removal became more effective after
oxidation. After 10 times of extraction with acetonitrile,
methanol and DMF, the sulfur content of oxidized
sample decreased from 1.48 wt % of the original VGO

Oxidative Desulfurization of Gas Oil

Energy & Fuels, Vol. 14, No. 6, 2000 1237

Figure 5. GC-FPD chromatograms of SR-LGO before and after oxidation, extraction with acetonitrile, and Soxhlet-extracted
compounds adsorbed on -Al2O3.

Figure 6. Changes in recoveries of oil layers as a function of

number of extraction before and after oxidation. (O-Unoxidized
VGO extracted with acetonitrile; b- Oxidized VGO extracted
with acetonitrile; 4-Unoxidized VGO extracted with methanol;
2-Oxidized VGO extracted with methanol; 9-Oxidized VGO
extracted with DMF).

to 0.12, 0.21, and 0.01 wt %, respectively. DMF is the

most effective solvent among the three solvents. Nonetheless, we did not find a significant difference in
extraction among the solvents for the unoxidized VGO.
GC-FPD chromatograms of unoxidized and oxidized
VGO after acetonitrile extraction are shown in Figure
8. The GC peaks of sulfur compounds in the oxidized
sample shifted to the heavier direction. The chromatograms showed that lighter sulfur compounds were
preferentially removed by extraction.
As shown in Figure 8, after 10 times extractions with
acetonitrile the sulfur compounds remaining in the
oxidized VGO sample are much heavier than those in
the unoxidized sample. Further, similar results were
obtained when DMF was used as extraction solvent.

Figure 7. Changes in sulfur concentration of oil layers as a

function of number of extraction before and after oxidation.
(O-Unoxidized VGO extracted with acetonitrile; b- Oxidized
VGO extracted with acetonitrile; 4-Unoxidized VGO extracted
with methanol; 2-Oxidized VGO extracted with methanol;
9-Oxidized VGO extracted with DMF).
Table 6. Sulfur Concentrations in Oil Layers of VGO
after Extractions
unoxidized VGO
2.17 wt %
feed
extraction times
acetonitrile
methanol
DMF

oxidized VGO
1.48 wt %

first
(wt %)

10 th
(wt %)

first
(wt %)

10 th
(wt %)

2.44
2.06

1.45
1.76

1.09
1.67
0.41

0.12
0.21
0.01

Especially, the sulfur compounds in VGO could be

oxidized with hydrogen peroxide and formic acid down
to the 0.01 wt % level (DMF extraction) as determined
by GC-FPD.
Figure 9 shows IR spectra of oil layers before and after
oxidation and extraction with acetonitrile. In these

1238

Energy & Fuels, Vol. 14, No. 6, 2000

Otsuki et al.

Figure 8. GC-FPD chromatograms of unoxidized and oxidized VGO after extractions with acetonitrile.

Figure 9. IR spectra of oil layers before and after oxidation and extraction with acetonitrile.

spectra, the sulfone was observed (1152 cm-1, 1298

cm-1) and this sulfone compounds decreased to undetectable levels by 10 times extraction with acetonitrile.
In addition, it should be noted that colors of oxidized
VGO were darker than fresh sample, but color of
oxidized VGO after extraction turned to pale white.
Discussion
Relationship Between the Reactivities and Electron Densities on Sulfur Atom. The oxidation of DBT
with hydrogen peroxide and formic acid has been

reported by Aida et al.7,18-20 The divalent sulfur of DBT

can be oxidized by the electrophilic addition reaction of
oxygen atoms to the hexavalent sulfur of DBT sulfone.
Hence, the reactivity of oxidation becomes higher for a
sulfur atoms with a higher electron density. Ford et al.
reported the similar results that the lower rate of
oxidation of thiophenic compounds than sulfides.3 The
observed trend of the rate with the structure is accounted for in terms of (a) reduced availability of the
lone pair electrons and (b) steric strain in the reaction
products, sulfoxides and sulfones. The sulfur atoms of

Oxidative Desulfurization of Gas Oil

2,5-DMT, 2-MT and thiophene have electron densities

of 5.716, 5.706 and 5.596, respectively. These low
electron densities are not sufficient to be oxidized by
hydrogen peroxide and formic acid. Nevertheless, BT,
of which the electron density is 5.739, and other model
compounds with higher electron densities can be oxidized to form corresponding sulfones. Therefore, the
minimum value to cause the oxidation under the conditions used for this study appears to exist between 5.739
and 5.716.
Generally, the oxidation of DBT is considered to be a
consecutive reaction, i.e., sulfur-containing compoundsf
sulfoxidefsulfone. In our study, no sulfoxide was
detected by GC analysis under all reaction conditions.
Therefore, the rate-determining step is the sulfidef
sulfoxide, and sulfoxide formation is considered to
dominate the reaction rate. In the present study, an
apparent rate constant was used to discuss the reactivity of several sulfur compounds.
Identification of Sulfur Compounds in Oxidation of SR-LGO. Alkyl-substituted BTs and alkylsubstituted DBTs are contained in SR-LGO.27 The
electron densities of these alkyl-substituted BTs and
alkyl-substituted DBTs in SR-LGO are higher than the
minimum value to cause the oxidation. Indeed, the GCFPD chromatogram of the oxidized oil layer shows that
sulfur compounds shifted to the heavier direction,
indicating the formation of corresponding sulfones, as
shown in Figure 3. The formation of sulfones are
detected by IR spectra (Figure 4). Moreover, Figure 4
and Figure 5 show that the sulfur compounds remaining
after acetonitorile extraction is oxidized to sulfones.
Identification of Sulfur Compounds in Oxidation of VGO. Figures 8 and 9 show that the sulfur
compounds were oxidized to form sulfone derivatives.
Ma et al.29 performed identification and quantification
peaks of organic sulfur compounds in VGO. This paper
reported that the four types of sulfur compounds, e.g.,
alkylbenzothiophenes with alkyl carbon atoms from 2
to 16, alkyldibenzothiophenes with alkyl carbon atoms
from 0 to 6, alkylbenzonaphthothiophenes with alkyl
carbon atoms from 0 to 5 and alkylphenanthro[4,5-b,c,d]thiophenes with alkyl carbon atoms from 2 to 7, were
basically present in the VGO. The electron densities of
these four types of sulfur compounds in VGO are more
than the minimum value to cause the oxidation under
the conditions used for present study. Therefore, the
oxidation of this sulfur compounds is expected to take
place.
Effect of the Polarity of Solvent on Extraction.
The dipole moment provides a means for measuring the
polarity, or electrical dissymmetry, of a molecule.30 The
dipole moment of solvents used for this study decreases
in the order: sulfolane > DMSO > acetonitrile g DMF
> methanol. In addition, dielectric constant decreases
in the order: DMSO > sulfolane > DMF > acetonitrile
> methanol as shown in Table 5. Therefore, the approximate value of polarity decreases in the order:
sulfolane, DMSO > DMF, acetonitrile > methanol.
The dipole moment, , is defined as follow.

) qr
where q is the electric charge, and r is the distance
between electric charges. Thus, the molecular polarity

Energy & Fuels, Vol. 14, No. 6, 2000 1239

of sulfones with lower molecular weight is greater than

that with higher molecular weight. Figure 8 shows the
GC-FPD chromatograms of unoxidized and oxidized
VGO after consecutive extraction with acetonitrile.
These results show that lower molecular weight sulfur
compounds were preferentially removed by extraction.
The remaining sulfur compounds after repeated extraction of oxidized sample appeared to be sulfones with
higher molecular weight, which is lower polarity. The
incomplete removal of sulfur compounds is, therefore,
not attributed to the insufficient oxidation but to the
insufficient extraction capacity of the solvent and low
polarities of higher molecular weight solfones. Hence,
the removal of sulfur compounds can be improved by
using a solvent with higher polarity.
The Side Reaction and Formation of Residual
Layer. Oxygen-containing functional groups such as
ketone (1715 cm-1), aldehyde (1725 cm-1), ester (1735
cm-1), and epoxy (3000-3050 cm-1) were not observed
in oil layer after oxidation of SR-LGO and VGO by IR
analysis. Nevertheless, the sulfur compounds and olefinic compounds in SR-LGO and VGO are expected to
oxidized selectively to form discolored gummy material
via chain reactions induced by hydrogen peroxide.31
Residue was 9.0 wt % of the feed SR-LGO containing
8.8 wt % sulfur. Therefore, the sulfur contents of oil
layer were reduced to 0.65 wt % from 1.35 wt % sulfur
in the original SR-LGO. Thus, 52% sulfur of the original
content was removed by oxidation via the formation of
sulfur-rich residue. In the same way, for VGO oxidation,
the residue with high sulfur concentration formed to
reduce the sulfur contents of original VGO to 1.48 wt
% from 2.17 wt %. Hence, 32% sulfur removal by
oxidation is due to the formation of residual layer.
Conclusions
(1) When the electron density on sulfur atom was
higher, the rate constant of oxidation was higher. The
oxidation of model sulfur compounds with H2O2 and
formic acid showed the minimum value of electron
density to cause the oxidation exists between 5.739 and
5.716.
(2) The oxidation of the sulfur compounds in SR-LGO
and VGO with H2O2 and formic acid took place to a 0.01
wt S level as determined by GC-FPD, XRF, and IR
analyses.
(3) The solvent with higher polarity extracted oxidized
sulfur compounds (sulfones) more effectively. DMF was
the most effective polar solvent among the solvents
used. The sulfur content of oxidized VGO was reduced
to 0.01 wt % S by consecutive 10-time extractions.
Acknowledgment. Authors are grateful to Mr.
Tatsushima, Mr. Iwasa, and Mr. Mizuno of NikkiUniversal Co. Ltd. for conducting XRF analysis for the
determination of sulfur contents.
EF000096I
(29) Ma, X.; Sakanishi, K.; Isoda, T.; Mochida, I. Fuel 1997, 76(4),
329.
(30) Riddick, J. A.; Bunger, W. B.; Sakano, T. K. Organic Solvents,
4th ed; John Wiley & Sons: New York, 1986.
(31) Collins, F. M.; Lucy, A. R.; Sharp, C. J. Mol. Catal. A: Chem.
1997, 117, 397.