P ROCESS T ECHNOLOGY
Catalysis
Part One of a Two-Part Series
The Challenging Chemistry of
Ultra-Low-Sulfur Diesel
By Salvatore Torrisi Jr., Criterion Catalysts & Technologies (Houston);
Tom Remans, Shell International Chemicals Research (Amsterdam);
Justin Swain, Criterion Catalysts & Technologies (Fareham, England)
In ultra-deep hydrodesulfurization (HDS) operation, the
reactivity of a diesel feedstock depends on the relative amounts
of less-reactive to more-reactive sulfur components present.
The levels of these components, in turn, depend on the crude
source, the degree of blending of cracked stocks, fluid catalytic cracker (FCC)/coker operation, and feed cutpoint. The boiling point of the feedstock, in particular the light-cycle oil
(LCO) and coker gasoil content, has a significant impact on the
required operating conditions for deep desulfurization.
As the boiling point of the feedstock increases, the concentration of the sterically hindered aromatic sulfur compounds increases as well. As the desired product sulfur in
treated gasoil is reduced, sterically hindered compounds will
have a more significant impact on the reactivity.1
T
General Sulfur Species in Diesel Feed
To meet the sulfur specification of ULSD, it is necessary to
first understand the different types of sulfur compounds
in diesel fuel and their relative reaction rates. Figure 1
depicts various sulfur compounds and their relative reaction rates as a function of boiling point.
In order to reduce sulfur levels below 500 ppm to
meet ULSD specifications, it is necessary to remove the
more-refractory, and therefore less-reactive, sulfur compounds based on multi-substituted dibenzothiophene
(DBT). For a typical medium-pressure diesel hydrotreater,
reducing product sulfur to 50 ppm can result in a more
than 50% loss in achievable catalyst cycle length using
conventional catalysts. In going one step further from 50to 10-ppm sulfur, one would be pushing the boundaries
of operation in units employing conventional catalysts, as
this requires removal of the even less-reactive, more-complex, multi-substituted DBT and higher homologues.
• World Refining • December 2002
1 0 0 0- 50 0
Re s idual S p leve l// ppm w
R
S
S
5 0 0 --3 0 0
3 0 0 --5 0
< 50
R
R
Relative Reaction Rate
he destiny of the diesel fuel sulfur specification is clear.
Legislators around the world will push for “zero”-sulfur
diesel fuel (ULSD) until it is a global reality. U.S. and
Canadian legislation requires a 15-ppm sulfur maximum by
2006, and initial implementation of the 10-ppm maximum in
the European Union is now set for 2005, with full compliance
required in 2009. Thanks to tax incentives, 50-ppm sulfur diesel
has already been introduced in a number of European markets,
and a few European refiners are even producing 10-ppm sulfur
diesel. Some Asia-Pacific countries, such as Japan and Korea, are
also now formulating plans for ULSD implementation.
Laboratory studies and commercial results from some
European refineries show that making the step from 50-ppm
to 10-ppm sulfur can be at least as difficult as the step
from 500-ppm to 50-ppm sulfur. This fact has led to many
questions about how to cost-effectively produce ULSD.
In a two-part article, we will discuss the potential for
advanced hydrotreating catalysts to provide a solution to
many of these challenges. In part one, we will discuss the
chemistry involved in producing ULSD and how it affects
the refiner’s course of action. The second part of the series,
which is scheduled to be published in the
January/February 2003 issue of World Refining, will examine how new hydrotreating catalysts are being designed to
more effectively exploit the ULSD chemistry and thus provide more cost-effective solutions to ULSD production.
S
S
S
N
S
S
S
110
230
330
220
430
340
Bo
oili n g Poi nt /°C
625 Boiling Point /°F 645
350
360
66 0
680
Figure 1. Reaction rates of various sulfur compounds as a function of
boiling point. More stringent specifications lead to more involved
chemistry.
Ultra-Deep HDS...Why it is different!
Ther
cs
mod
yn
Kineti
ULTRA-DEEP HDS CHEMISTRY
amic
s
ULTRA-DEEP HDS LIMITATIONS
Thermodynamic Effects
S
CH3
HDA
Usual
Indirect HDS
CH3
CH3
Increasing
Pressure
CH3
Decreasin
Decreasing
S
CH3
Direct
HDS
CH3
LHSV
CH3
CH3
WABT
Figure 2. Mechanistic pathways of desulfurization reactions.
www.worldrefining.com
P R O C E S S T E C H N O LO GY
% Refractive S: 94%
6
5
6
5
IBP-FBP = 392-817°F
Feed A
4
3
2
4
3
2
1
1
0
10
20
30
40
50
60
70
80
90
Minutes
Chn B Antek
TMC.10001.31
58
% Refractive S: 48%
40
30
20
10
40
30
IBP-FBP = 320-738°F
Feed B
20
10
0
0
10
0
20
30
40
50
60
70
80
90
Minutes
Chn
25
Feed
B
Antek
TMC
1001.36
25
200025613
% Refractive S: 37%
24
IBP-FBP = 232-741°F
Feed C
20
15
10
20
15
10
5
5
0
0
0
10
20
30
40
50
60
70
80
90
Minutes
Figure 3. Percentage of refractive sulfur compounds in feed as a function
of boiling point.
Removal of bulky basic nitrogen species is also a key element to efficient ULSD operation, as substances such as acridine
(Figure 1) effectively compete for the same active sites as the
sulfur molecules.2
Ultra-Deep HDS Reaction Mechanism
50-ppm sulfur diesel compared to 350- to 500-ppm sulfur diesel.
This also explains why a higher-pressure existing hydrotreater can
have more flexibility (greater temperature operating window)
with a simple catalyst drop-in solution to meet lower sulfur specifications. When higher pressure is not an option, then the lower
LHSV (liquid hourly space velocity) route must be followed.
Nitrogen removal follows a very similar reaction pathway,
requiring a pre-hydrogenation step before heteroatom removal. It
therefore competes with the hindered DMDBTs for active sites on
the catalyst surface. Since nitrogen species, particularly basic
nitrogen compounds, adsorb more strongly to that catalyst surface than sulfur species, near-complete removal of nitrogen is
required when targeting sulfur levels <15 ppm. We discuss the
importance of choosing a catalyst having this dual HDS/HDN
functionality in subsequent sections.
Sulfur Compound Identification in Gasoil
Many sulfur species are present in the feed and product gasoil,
and some of these are especially difficult to remove in the context of ULSD. The feedstocks shown in Table 1 represent a
variety of feeds processed in Europe and in some North
American refineries.
Feed A is an example of North Sea sweet straight-run gasoil
(SRGO) that might appear to be the easiest to treat to ULSD
because of its low sulfur content. However, note the unusual true
boiling point (TBP) distillation that shows both a heavy tail and
90% of the feed boiling above 572˚F (300˚C).
Feed B is a typical full-range 80/20 SRGO/LCO blend
derived from Arabian crude and is the most difficult feed in terms
of sulfur and polyaromatic content. The presence of the LCO
would suggest that this feed might be difficult to process to ULSD.
Feed C is a 70/30 SR/cracked blend derived from a Nigerian
crude with a distillation typical of a full-range distillate feed.
A detailed sulfur analysis of the three feedstocks is shown in
Figure 3 (page 25). From Figure 3, we could rank the feed reactiv-
Making ultra-deep HDS chemistry happen effectively depends on
both creating an environment for good kinetic response of the catalyst and steering the thermodynamics to ensure an economical
window of operation. Desulfurization reactions can follow two
mechanistic pathways, as shown in Figure 2 with a di-substituted
DBT as the model sulfur compound.
The traditional mechanism features direct extraction of the
sulfur, or hydrogenolysis (carbon-sulfur bond breakage), as the
rate-determining step. This direct mechanism is favored on conventional catalysts when HDS is required down to product sulfur
levels above 350 ppm. At this product sulfur level, all of the morereactive sulfur compounds, up to and including nonsubstituted DBT, have been removed.
TABLE 1. FEEDSTOCK ORIGIN AND PROPERTIES
When a sulfur specification of 10 ppm or 50 ppm
is to be met, then removal of the less-reactive and
Feed Origin
Feed A
Feed B
Feed C
bulky di-substituted DBTs, such as 4,6Feed Type
SRGO
80/20
70/30 SRGO/
dimethyldibenzothiophene (4,6-DMDBT), is necesSRGO/LCO
VB+ LCO
sary. The concentration of these species increases with
Crude Origin
North Sea
Arabian
Nigerian
the refractory nature (sulfur, nitrogen, boiling point,
Density, ˚API
32.1
28.4
39.4
percentage of cracked component) of the feed. These
Sulfur, wt%
0.196
1.29
0.517
species are more effectively removed via the alternaTotal Nitrogen, wppm
200
200
145
tive hydrogenation-controlled mechanism (Figure 2).
In order for the breakage of the carbon-sulfur bond to
Aromatics (IP 391), wt%
occur, saturation of one of the aromatic rings must
Mono
14.1
17.2
18.2
first take place. This partial saturation changes the
Di
4.6
15.3
5.8
spatial configuration of the molecule, making the prePri +
3.8
2.1
0.9
viously sterically hindered sulfur more accessible for
effective adsorption on the active site and subsequent
Distillation (TBP), ˚F
reaction. In this respect, the ultra-deep HDS mechaIBP
392
320
232
nism is very similar to the traditional hydrodenitro10
572
457
331
genation (HDN) mechanism, where pre-hydrogena50
673
561
484
tion precedes hydrogenolysis.
90
743
669
635
Clearly the influence of the polyaromatic saturaFBP
817
738
741
tion equilibrium increases when producing 10- to
P R O C E S S T E C H N O LO GY
MBT’s
MDBT’s
DMBT’s
BT
T’s
%:
DMDBT’s
C3-DBT’s
C3-BT’s
DBT
C4+-BT’s
Chn B Antek
25
Feed TMC 1001.36
25
200025613
24
20
2
10
4-MDBT
3-MDBT
2-MDBT
1-MDBT
C4+-DBT’s + 4-ring S-compounds
20-30
4 x 3-CDBT (1 st = 4-M,6-EDBT)
6 x 4-CDBT (1 st = 4,6 DEDBT)
Other poly-CDBT’s
20
4,6-DMDBT
Feed C
15
15
10
58-68
2,4-DMDBT
3,4-DMDBT
1,4-DMDBT
10
5
5
0
TMC10001.36: IBP-EBP=111-394˚C
0
10
20
30
40
50
60
70
80
0
90
Minutes
T:
Chn B Antek
TMC.10001.31
58
40
40
Feed B
30
T:
334˚C
630°F
340˚C
645°F
360˚C >360˚C
680°F >680°F
345˚C
650°F
30
20
20
10
10
30
40
50
60
70
80
90
Minutes
0
0
TMC10001.31: IBP-EBP=160-392˚C
0
10
20
30
40
50
60
70
80
90
Minutes
•
•
•
•
•
•
•
•
•
•
•
T’s:
BT:
MBT:
DMBT’s :
C3-BT’s:
thiophene-group
benzothiophene
methylbenzothiophene-group
dimethylbenzothiophene-group
benzothiophene-group with 3 carbons attached in whatever
configuration
C4+-BT’s:
benzothiophene-group with 4 or more carbons attached in whatever
configuration
DBT:
dibenzothiophene
MDBT’s :
methyldibenzothiophene-group
DMDBT’s :
dimethyldibenzothiophene-group
C3-DBT’s :
dibenzothiophene-group with 3 carbons attached in whatever
configuration
C4+-DBT’s + 4-ring S-compounds:
dibenzothiophene-group with 4 or more carbons attached in
whatever configuration, including S-compounds in 4-ring configurations
Figure 4. Compound identification in feed.
ity, from most to least reactive, as Feed C > Feed B > Feed A.
Feed A is characterized by having few or no sulfur species
boiling below DBT. This is in line with its rather skewed TBP
distillation showing 90% of the feed boiling above 572˚F
(300˚C) and the heavy tail extending to a feed boiling point
(FBP) of 817˚F (436˚C). The large concentration of bulky,
multi-substituted, less-reactive sulfur species, as also evident
by the large integrated area beneath the baseline, identifies this
feed as one to be avoided when operating in ULSD mode. In
fact, by characterizing refractive sulfur species as all those
beyond DBT, then 94% of the sulfur in this feed is refractive.
Feed B, with its more typical diesel boiling range, is characterized by a more broad-ranging sulfur chromatogram of higher
resolution. Note the smaller integrated area under the baseline,
which indicates a broader range of sulfur species present. The
amount of refractive sulfur in this feed was found to be 48%.
Finally, Feed C, with its wide boiling range, shows the
largest distribution of sulfur species, with (correspondingly)
the lowest concentration of the more difficult-to-remove sulfur compounds. This is evident by the high degree of peak
resolution and the small integrated area under the baseline.
The light front end of the feed skews the distribution to more
easy-to-remove sulfur species, as is further evidenced by the
lower amount of refractive sulfur (37%).
The detailed feed analysis of Feed C (compared with
Feed B) revealed the identification of sulfur compounds as
shown in Figure 4.
If Feed C is subsequently treated to <10 ppm Sulfur, the
resultant distribution is as shown in Figure 5. From the product sulfur-speciation, one can observe the distinct peak of 4,6DMDBT at the 645˚F (340˚C) boiling point, as well as the rest
•
•
•
•
•
•
•
•
•
4-MDBT:
3-MDBT:
2-MDBT:
1-MDBT:
4,6-MDBT:
2,4-MDBT:
3,4-MDBT:
1,4-MDBT:
4 x 3-CDBT:
•
•
4-M,6-EDBT:
6 x 4-CDBT:
•
•
4,6-DEDBT:
poly- CDBT’s :
4-methyldibenzothiophene
3-methyldibenzothiophene
2-methyldibenzothiophene
1-methyldibenzothiophene
4,6- dimethyldibenzothiophene
2,4- dimethyldibenzothiophene
3,4- dimethyldibenzothiophene
1,4- dimethyldibenzothiophene
4 dimethyldibenzothiophenes with 3 carbons attached, at
least two of which are in the 4,6-position, the first
compound being 4-M,6-EDBT
4-methyl,6-ethyldibenzothiophene
6 dimethyldibenzothiophenes with 4 carbons attached, at
least 2 of which are in the 4,6-position, the first
compound being 4,6-DEDBT
4,6- diethyldibenzothiophene
dimethyldibenzothiophenes with more than 4 carbons
attached
Figure 5. Compound identification in liquid product. X-CDBT =
dibenzothiophene with X carbon atoms attached to the ring in
unknown configuration.
of the unconverted substituted DBT beyond that boiling point.
Despite having the strongest peak in the chromatogram,
4,6-DMDBT accounts for only 10% of the sulfur remaining in
this hydrotreated SRGO. The majority (up to 85%) of the sulfur
is bound in the more-complex sulfur compounds extending all
the way to what are termed poly-CDBTs (carbon dibenzothiophenes) whose reactivity is similar to or lower than that of 4,6DMDBT. This distribution also illustrates why feed undercutting is such a powerful tool for meeting ULSD specifications.
Feedstock Effects on HDS
Having identified the different sulfur species present in the
feed and product, we can look at the impact of these sulfur
species on the “processability” of the different feeds to meet
the ULSD specifications.
An investigation was made to study the reactivity of different feedstocks as the product sulfur specification moves
from 300 ppm to 10 ppm. Figure 6 compares the reactivity
in the ULSD regime of four feeds: three SRGOs and one
80/20 SRGO/LCO blend, each derived from different crudes.
The data were generated by pilot plant processing of the feeds
over a Criterion cobalt-molybdenum (CoMo) catalyst.
In terms of overall processability, the highest-sulfur feed
(SRGO 1) has turned out to be the most difficult; it requires the
highest weighted-average bed temperature (WABT) at each
product sulfur level. However, when the feedstock reactivity
ranking is defined in terms of temperature increase required to
reach 30 ppm, then the order of most reactive to least reactive
feed is: SRGO 2 > SRGO/LCO ~ SRGO 1 > SRGO 3.
Feedstock reactivity ranking changes very differently as the
P R O C E S S T E C H N O LO GY
Feed Reactivity with CoMo
Base+54
Mid - East
Base+36
North Sea
Treq, ˚F
Base+18
Treq
for 300 ppm
ppm S S
Treq
for 30
ppm
S
Treq
for 10
ppm
S
North Slope
Mid - East
Base
Base - 18
Product Sulfur, ppm
1000
500
(1.0 hr - 1, Med P , 1200 SCF H2/B)
Base+72
35 °F Reduction Can:
500
- Increase Rate by 40%
- Lower S to 80 ppm
100
100
50
10
T90=680°F
T
90=680°F
(360°C)
Base - 36
Base - 54
T90=645°F
T
90=645°F
(340°C)
1
0
Base - 72
SRGO 1
(1.46 wt% S)
SRGO 2
(0.20 wt% S)
SRGO 3
(0.52 wt% S)
1
2
SRGO/LCO
(1.29 wt% S)
Figure 6. Temperature requirements (Treq) for low-sulfur diesel and ULSD.
shift from 30-ppm to 10-ppm product sulfur is considered. In this case
the ranking becomes: SRGO 3 ~ SRGO/LCO > SRGO 1 > SRGO 2.
SRGO 2, despite being derived from the sweetest crude,
proved to be the most difficult feed to hydrotreat from 300-ppm
to 10-ppm sulfur. While a 35˚F (20˚C) increase in temperature
was needed to go from 300-ppm to 30-ppm sulfur, a further
55°F (30˚C) was required to go from 30 ppm to 10 ppm.
Effect of Reducing Endpoint
Based on the study so far, it can be seen that the more-difficultto-remove sulfur species tend to be those substituted DBTs that
start from 4,6-DMDBT at roughly the boiling point of 645˚F
(340˚C). Especially if a sulfur level much lower than 50 ppm is
desired, then the most-refractory sulfur compounds and highermolecular-weight DBTs have to be reduced or removed.
One direct way to solve this problem is to reduce the cut
point of the feed. Figure 7 (page 27) illustrates the impact of
changes in feedstock properties, such as a decrease in T90 point
through undercutting of the feed. This example illustrates that a
reduction in T90 of 35°F (to 645°F) can exclude most of the difficult sulfur species and make it much easier to achieve ULSD. This
comes at a substantial cost, as yield losses can range 10%–20% or
higher depending on the particular feed. Each refiner’s circumstance must be considered individually to evaluate the merits of
endpoint reduction and subsequent economic impacts.
Importance of Nitrogen Removal to Produce ULSD
It is widely known that organo-nitrogen compounds inhibit HDS
reactions, especially those involving the two-step conversion of
refractory sulfur, like DMDBT. In particular, basic nitrogen species
strongly adsorb to the catalyst surface and prevent some HDS reactions from proceeding unless the nitrogen compound is removed.
Even non-basic nitrogen species having low adsorption constants,
such as carbazole, are also known to react in the hydrotreating
environment to form basic nitrogen. This will inhibit the effectiveness of HDS and add to the challenge of making ULSD.
Controlled experiments studying the effects of basic nitrogen
on deep HDS of a gasoil stream were conducted to quantify the
impact without interference from other variables. Two deeply desulfurized gasoils were mixed in different proportions, resulting in different basic nitrogen concentrations. The feed sulfur was then
adjusted to a constant 2,000 ppm by adding DBT, which was selected because its reactivity would be sufficiently different from the low
levels of refractory sulfur present in the deeply desulfurized feed.
Results presented in Figure 8 show that there is a marked dif-
3
4
Relative Reactor Volume
5
6
7
Figure 7. Relative reactor volume vs. feed T90.
ference in the rate of sulfur removal between feeds containing 2
ppm and 18 ppm of basic nitrogen.1 In part two of this article, we
will discuss how the CENTINEL catalyst functionality has been
modified to improve the HDN activity, which in turn enhances its
ability to achieve ULSD levels of <10 ppm. ●
REFERENCES
1. Ee, C.S., Parthasarathi, R.S, and van den Brule, T. “Sulfur
Species and Distribution in Gasoil,” paper presented at a
technical conference in Pusan, Korea, Oct. 18-19, 2001.
2. van Looij, F., van der Laan, P., Stork, W.H.J., DiCamillo, D.J.,
and Swain, J. “Key Parameters in Deep Hydrodesulfurization of
Diesel Fuel,” App. Catal. A, 170: 1-12, 1998.
ABOUT THE AUTHORS
Salvatore P. Torrisi Jr., clean-fuels technical specialist for
Criterion Catalysts & Technologies, has responsibility for cleanfuels project coordination in the Americas. He joined the CRI
International family in 1996, becoming lab manager with the
Research & Commercial Development Group. More recently he
has been involved with new-product market introduction, particularly with the application of CENTINEL catalysts.
Justin Swain joined Criterion in 1996 and is currently technical manager for distillate hydrotreating. Before joining Criterion,
he worked at the Milford Haven Refinery in Wales, where he was
responsible for operation and optimization of distillation and catalytic processes.
Tom Remans has been working for Shell since 1997 in R&D
and technical service for refinery catalysts marketed by Criterion
Catalysts & Technologies. He has worked extensively in the areas
of distillate hydrodesulfurization, first-stage hydrocracking, and
catalytic dewaxing.
1400
1300
1200
1100
1000
900
800
700
18
6
2
N (Basic) ppm
CoMo NiMo
Figure 8. Inhibition of HDS activity by trace organo-nitrogen compounds.
 Copyright, Hart Publications, 4545 Post Oak Place, Ste. 210, Houston, TX 77027 USA (713)993-9320, Fax (713)840-8585