Document 11105117

advertisement
PREDICTING ORGANOSULFUR CHEMISTRY IN FUEL SOURCES
ARCHE$
by
CALEB ANDREW CLASS
MASSACHUSETTS INSTITUTE
OF rECHNOLOLGY
M.S. Chemical Engineering Practice
Massachusetts Institute of Technology, 2011
B.S. Chemical Engineering
Purdue University, 2009
JUN 162015
LIBRARIES
Submitted to the Department of Chemical Engineering
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy in Chemical Engineering
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
February 2015
0 2015 Massachusetts Institute of Technology. All rights reserved.
Signature of Auth or
Signature redacted
Department of Chemical Engineering
January 29, 2015
Certified by__
ignature redacted
K
Accepted by
William H. Green
Hoyt C. Hottel Professor of Chemical Engineering
T hesis Supervisor
Signature redacted
j
Richard D. Braatz
Edwin R. Gilliland Professor of Chemical Engineering
Chairman, Committee for Graduate Students
Abstract
3
3
Abstract
Predicting Organosulfur Chemistry in Fuel Sources
by
Caleb A. Class
Submitted to the Department of Chemical Engineering on
January 29, 2015 in partial fulfillment of the requirements
for the degree of Doctor of Philosophy
Abstract
Desulfurization of fossil fuels with supercritical water (SCW) has been the topic of many studies
over the past few decades. This process does not require the use of any catalyst, eliminates the
need for a hydrogen feed, and minimizes coke formation. Previous research has shown that it has
the potential to be a viable commercial process, and recent experimental studies have proven that
water acts as one hydrogen source for sulfur removal in this process. However, the exact
desulfurization mechanism is largely unknown, as are many other reaction mechanisms
involving sulfur compounds. Recent work has greatly expanded our ability to build
comprehensive reaction mechanisms automatically for the decomposition of organic sulfur
compounds using the automated Reaction Mechanism Generator (RMG). This thesis presents the
implementation of this and other tools to investigate chemical processes relevant to our use of
fuel sources containing sulfur compounds, and it shows some steps that have been taken to
improve our predictions for these mechanisms and those that will be generated in the future.
Previous investigations had focused on the pyrolysis of small sulfur compounds containing less
than six heavy atoms, so RMG is first used to study the pyrolysis of t-butyl sulfide. A detailed
reaction mechanism is then presented for the SCW desulfurization of hexyl sulfide.
Comprehensive kinetic mechanisms for these larger molecules are likely to include thousands of
reactions, so RMG builds this model in a systematic and unbiased way using a database of ab
initio data. This database is expanded with potentially relevant thermochemical and kinetic
parameters using transition state theory and quantum chemical calculations at the CBS-QB3 and
CCSD(T)-F12 levels of theory. With these data, as well as previously calculated rates for
Abstract
4
hydrocarbon and sulfur kinetics, RMG is used to build a reaction mechanism for the conversion
of hexyl sulfide to hydrogen sulfide, pentane, and carbon monoxide in the presence of SCW.
This mechanism is validated with results from batch and flow reactor experiments, and
predictions are accurate within a factor of two for reactant and major product concentrations.
Analysis of the proposed mechanism shows that the molecular addition of water to the carbonsulfur double-bond in hexanethial is a key step in the SCW process, as this not only leads to the
desulfurization of the compound, but also prevents the thioaldehyde from undergoing addition
reactions with other hydrocarbons in a process that could eventually form coke. Thus, this work
not only has implications in the SCW desulfurization process, but in the overall crude oil
upgrading process as well.
The calculated kinetic and thermochemical parameters are used to generate predictive reaction
mechanisms for other processes relevant in fuel chemistry, such as the geological formation of
oil and gas from kerogen. This not only allows us to model experimental work investigating the
effect sulfur compounds have on the oil-to-gas process, but we also explore how these effects
differ at geological conditions and timescales. And as the possible applications of RMG grow,
the need for accurate parameters in mechanism generation become even more critical. A
thermochemical database is generated for a wide variety of sulfur compounds using the highaccuracy CCSD(T)-F12/cc-pVTZ-F12 method, and this provides a basis for the investigation of
organosulfur chemistry with tighter uncertainty.
Thesis Supervisor: William H. Green
Title: Hoyt C. Hottel Professor of Chemical Engineering
Acknowledgements
5
Acknowledgements
Thank you to my thesis advisor, Prof. William H. Green. Your encouragement through the ups
and downs of a difficult project has been critical to my success, and your input has pushed us to
exciting new discoveries. I'm extremely grateful for your mentorship, not only in helping me
achieve success in kinetic modeling, but in my future professional life as well.
Thank you to my thesis committee members, Profs. Ahmed Ghoniem, Yuriy Romdn, and
Michael Timko. Your suggestions at committee meetings, videoconference preparation meetings,
and other interactions have not only been valuable in guiding the direction of my research, but
have also improved my ability preparing and giving presentations to a variety of audiences. Prof.
Timko was also the director for the upgrading project for my first few years at MIT, and our
interactions during this period were critical to my success in the remaining years.
Thank you to Dr. Yuko Kida, whose curiosity, diligence, and attention to detail in the lab led to
the discoveries that made my thesis scientifically satisfying, and whose presence in the lab made
it a truly enjoyable place to work.
Thank you to Dr. Aaron Vandeputte for writing such a good thesis and providing a strong basis
for the predictive modeling of organosulfur chemistry. It was a pleasure working with you, even
when the models weren't such a pleasure.
Thank you the RMG developers, server administrators, and quantum chemists, including
Shamel, Connie, Jorge, Josh, Ray, Nick, Mike, Richard, Enoch, Nate, and Amrit. This thesis is
only possible thanks to your commitment to removing bugs, repairing hardware, and making
sure my rate calculations weren't ridiculous.
Thank you to the past and present members of the SCW Upgrading team, including Adam, AJ,
Lawrence, Raj, Andrew, and Ashwin. Your contributions have allowed us to overcome
numerous experimental and modeling challenges, and my thesis, as well as our overall
knowledge of the upgrading process, is better for it.
Acknowledgements
6
Thank you to Prof. Shuhei Ono and Dr. Eoghan Reeves. It has been both fascinating and
enjoyable parsing and interpreting your experimental results with you, and gaining new insight in
the oil-to-gas process.
Thank you to Prof. Ilse Ipsen for giving this chemical engineer the chance to be a mathematician
for a week.
Thank you to Gwen, Alison, and Barb, for helping the group's operations run smoothly
(probably in many more ways than I'm even aware of). Your problem-solving, willingness to
lend an ear, and even your pizza choices are greatly appreciated.
Thank you to all of the past and present members of the Green Group. You're not only an
intelligent, curious, and hard-working bunch, but a fun group of people to spend an evening with.
The value of going to a conference, meeting, or regular workday with helpful, kind, and pleasant
coworkers cannot be understated, and I'm incredibly grateful for this.
Thank you to all of the friends I've met while at MIT. Even when a day at the office went poorly,
I always looked forward to spending time with you guys, whether doing something simple like a
dinner, pub quiz, or Lame Night, or something incredibly stupid like a Pieathlon.
Thank you to my parents, siblings, and the rest of my family for loving me, supporting me, and
listening to me (even when I wasn't in the mood to talk).
This is the point where I would put a joke acknowledgement for a random celebrity, but I've
learned not to do that anymore.
Table of Contents
7
Table of Contents
Chapter 1: Introduction.................................................................................................................
11
1.1
Form ation and composition of fuels...........................................................................
12
1.2
Pyrolysis of sulfur compounds....................................................................................
13
1.3
Supercritical w ater (SCW ) treatment of crude oil......................................................
15
1.4
Autom atic m echanism generation..................................................................................
19
1.5
Param eter estim ation..................................................................................................
20
1.6
Ab initio calculations..................................................................................................
21
1.7
Thesis Overview .............................................................................................................
22
1.8
References ......................................................................................................................
24
Chapter 2: M odeling the pyrolysis of t-butyl sulfide...............................................................
27
2.1
Abstract ..........................................................................................................................
27
2.2
Introduction ....................................................................................................................
28
2.3
M ethods..........................................................................................................................
29
2.4
Results and Discussion................................................................................................
30
2.4.1
Calculation of Unim olecular D ecom position Rates ............................................
30
2.4.2
Pyrolysis of Neat t-Butyl Sulfide........................................................................
32
2.4.3
Pyrolysis of t-Butyl Sulfide with Cyclohexene ...................................................
35
2.4.4
M echanism Comparison .........................................................................................
37
2.5
Conclusions ....................................................................................................................
38
2.6
References ......................................................................................................................
)9
Chapter 3: A kinetic database for organic sulfur and oxygen compounds ...............................
41
3.1
Abstract ..........................................................................................................................
41
3.2
Introduction ....................................................................................................................
42
3.3
Methods..........................................................................................................................
44
3.3.1
Calculation of Rate Constants for Reactions with Submerged Transition States... 46
3.3.2
Basis Set benchm arking for CCSD(T)-F 12 Calculations ....................................
3.4
Results and D iscussion................................................................................................
47
48
3.4.1
Molecular Addition of Water (Hydration of Double Bonds)...............................
48
3.4.2
M olecular Addition of Hydrogen Sulfide ..........................................................
52
Table of Contents
8
3.4.3
Hydrogen Abstraction Reactions ........................................................................
56
3.4.4
Radical Addition to Double Bonds (Reverse Beta-Scission) .............................
60
3.4.5
Tautom erization of Thiocarboxylic Acids ...........................................................
63
3.4.6
Therm ochem ical Library ......................................................................................
65
3.5
Conclusions ....................................................................................................................
66
3.6
References ......................................................................................................................
67
3.7
Appendix: Calculations for reactions 21 and 37 ........................................................
70
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water......................
71
4.1
Abstract ..........................................................................................................................
71
4.2
Introduction ....................................................................................................................
72
4.3
M ethods..........................................................................................................................
73
4.3.1
Batch Reactor Experim ent ....................................................................................
73
4.3.2
Batch Reactor M odel ..........................................................................................
74
4.3.3
Continuous Flow Stirred Reactor (CSTR) Experiment ......................................
74
4.3.4
CSTR Reactor M odel...........................................................................................
75
4.3.5
Principles of Autom ated M echanism Generation...................................................
75
4.3.6
Quantum Calculations.............................................................................................
76
Results: Quantum Calculations ......................................................................................
77
4.4
4.4.1
Water-Catalyzed Elim ination of H 2 S .................................................................
77
4.4.2
Hydrogen M igration.............................................................................................
79
4.4.3
Radical Addition to M ultiple Bond......................................................................
80
4.4.4
Cyclic Sulfide Formation
4.4.5
Therm ochem istry Calculations .............................................................................
84
Results: RM G Model Perform ance .............................................................................
84
4.5
.......................................
4.5.1
Reaction Path Analysis ........................................................................................
84
4.5.2
M odel validation for hexyl sulfide conversion in a CSTR ..................................
89
4.5.3
Model validation for product distributions in a batch reactor..............................
90
4.5.4
Effect of W ater Concentration .............................................................................
93
4.6
Conclusions ....................................................................................................................
94
4.7
References ......................................................................................................................
95
Chapter 5: Modeling the decomposition of alkylaromatic compounds ....................................
98
Table of Contents
9
5.1
A bstract ..........................................................................................................................
98
5.2
Introduction ....................................................................................................................
99
5.3
M ethods..........................................................................................................................
99
5.3.1
Reaction Sim ulation.............................................................................................
5.3.2
Quantum Calculations
5.4
.........................................
Results and D iscussion.................................................................................................
99
100
101
5.4.1
Quantum Calculations...........................................................................................
101
5.4.2
RM G M odel Perform ance.....................................................................................
102
5.5
Conclusions..................................................................................................................105
5.6
References ....................................................................................................................
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition.........................
106
108
6.1
Abstract........................................................................................................................
108
6.2
Introduction
109
6.3
M ethods........................................................................................................................
...................................................
110
6.3.1
Reaction Sim ulation..............................................................................................
110
6.3.2
Quantum Calculations...........................................................................................
1-10
Results and D iscussion.................................................................................................
111
6.4
6.4.1
Diethyl Disulfide Decomposition M echanism ......................................................
111
6.4.2
Phenyldodecane D ecom position M echanism .......................................................
117
6.4.3
Phenyldodecane Decomposition in the Presence of Diethyl Disulfide ................ 120
6.5
Conclusions ..................................................................................................................
123
6.6
References....................................................................................................................
123
Chapter 7: A therm ochem ical database for organosulfur com pounds........................................
7.1
A bstract ........................................................................................................................
7.2
Introduction..................................................................................................................126
7.3
M ethods........................................................................................................................
7.3.1
7.4
Regression of Bond Additivity Corrections (BA C's)...........................................
125
127
128
Results and D iscussion.................................................................................................129
7.4.1
BAC Regression and V alidation........................................................................
7.4.2
Therm ochem ical Database..................................................................................131
7.5
125
Conclusions ..................................................................................................................
129
137
.................
Table of Contents
10
7.6
References ....................................................
138
7.7
Appendix: Calculated Thermochemical Parameters....................................................
140
C hapter 8: C onclusions ...............................................................................................................
150
8.1
S u m mary ......................................................................................................................
150
8.2
Recommendations for future work............................................................................
151
8.3
R eferen ces ....................................................................................................................
152
------------
-
..
Chapter 1: Introduction
Chapter 1: Introduction
11
11
Chapter 1: Introduction
As the need for affordable sources of energy increases throughout the developing and developed
world, high-value fuel sources like light crude oil have become rarer and more difficult to obtain.
Existing petroleum reservoirs contain heavier oils that are more expensive to process,' while
even more expensive processing is required to meet stricter environmental regulations on the
impurities that are allowed in commercial fuels. 2 This increased cost, in addition to the increased
cost of nontraditional drilling, makes petroleum exploration a much riskier affair, as the margin
of error has shrunk dramatically.
Understanding the chemistry of these fuel sources can greatly enhance our ability to find and
harness these hydrocarbons. Reservoir models currently employ very general models to make
billion-dollar drilling decisions. Refinery models use much more detailed kinetic mechanisms,
but these are also frequently fit to match their exact conditions, and may therefore be prone to
errors when applied to different conditions.
Sulfur compounds in fossil fuels and their precursors are of particular interest. Sulfur can be
present at up to 10 wt% in crude oil, and these compounds must be removed to comply with
environmental regulations and prevent the poisoning of catalytic converters. A detailed kinetic
Chapter I.: Introduction
12
understanding of how various sulfur compounds react with other hydrocarbons, as well as how
they can be removed most efficiently, would potentially improve our ability to find and process
fuel sources.
1.1
Formation and composition of fuels
Fossil fuels are produced by the decomposition of plant and animal remains at high pressures,
deep beneath the surface of the earth. While coal was mainly formed from the remains of ferns
and other plants in swampy areas, the organisms in the ocean mainly decompose to form
kerogen, a heavy, waxy hydrocarbon. 3 This kerogen reacts further to form natural gas and crude
oil, which may also continue to produce more natural gas. The kerogen-to-oil-to-gas process is
likely affected by the presence of metals, 4 water,' 6 and reactive sulfur compounds,7 and
geochemists use this information to make drilling decisions.
This research is focused on the compounds found in crude oil, which can be separated into four
main categories: saturates, aromatics, resins, and asphaltenes. 8 The first category includes simple
alkanes and organosulfur molecules like thiols, sulfides, and disulfides, while the aromatics
fraction includes thiophenes and benzothiophenes. The resins and asphaltenes are larger
compounds that generally contain multiple aromatic rings, a saturated portion, and some
heteroatoms. Asphaltenes are larger than resins, and they precipitate out of the mixture when nalkanes are added. While some constituent units in these larger species have been identified,9
their full molecular structures are not well known. One example of a possible asphaltene
molecule can be seen in Figure 1.
N
Figure 1. Hypothetical asphaltene structure.
The sulfur types present in crude oil vary widely from reservoir to reservoir. Table 1 shows the
relative abundance of five sulfur types in the various fractions obtained from two different crude
Chapter 1: Introduction
13
oils.' 0 Sulfides are generally the easiest to desulfurize, but recent work has shown that these
compounds can react to form aromatic compounds in simple pyrolysis." Thiophenes are much
more stable in noncatalytic treatment, but the removal of alkyl chains connected to these
aromatic rings may be a key to overall upgrading chemistry. A more detailed understanding of
the chemistry of each of these compounds will be necessary to improve our utilization of crude
oil.
Table 1. Relative abundance of different sulfur types in two representative crude oils.'0
Sulfide
1.2
Thiophene Sulfoxide Sulfone
Sulfate
CAL Asph
15
29
50
5
1
CAL Resin
CAL Oil
KUW2 Asph
KUW2 Resin
KUW2 Oil
11
24
40
40
45
27
27
55
52
47
59
46
2
5
5
1
1
1
1
1
1
1
1
1
1
Pyrolysis of sulfur compounds
An understanding of the decomposition mechanisms of individual organosulfur compounds,
including both aliphatic and aromatic sulfur, is critical to gaining insight on how these species
affect overall processes in fuel chemistry. Much of the research in the pyrolysis of aliphatic sulfur
compounds is focused on sulfides. In particular, a variety of alkyl t-butyl sulfides and alkyl allyl sulfides
have been studied by Martin et al., and the overall Arrhenius parameters for these reaction networks were
tabulated in Martin's review on organosulfur compound pyrolysis.12 The results for alkyl t-butyl sulfide
pyrolysis supported the transition state suggested by Benson and Haugen," which is shown in Figure 1.
In this type of scheme, the reaction would occur more quickly if the transition state is stabilized by
resonance effects; this is true in Martin's work, as the overall pyrolysis of t-butyl phenyl sulfide is found
to have a lower activation energy than the fully non-aromatic compounds. 4
14
Chapter 1: Introduction
(CH3) 2C----S--Ri
H-S-R
S-R
C- ----H
H2
-J
3
Figure 2. Proposed reaction pathway for the decomposition of an alkyl t-butyl sulfide.'
Researchers have also proposed free radical mechanisms for the pyrolysis of some sulfide
compounds. Many past studies were not able to quantitatively validate their proposed
mechanisms with experimental data, but there has been some recent progress in this area. Zheng
et al. proposed a kinetic mechanism for the pyrolysis of diethyl sulfide, and it compared
reasonably well with the experimental results.15 Work by Vandeputte et al. has provided
additional insight into decomposition of small organosulfur compounds, using computational
chemistry techniques to model the pyrolysis of dimethyl disulfide and diethyl sulfide.' 6' 17 The
study provided a validation of an expansive
thermochemistry of organosulfur compounds.' 8-21
second
database for the kinetics and
Thiols are important intermediates in many sulfur pyrolysis mechanisms.18-21 A few of the
compounds previously studied include t-butanethiol, I-butanethiol, and bicyclo[2,2, 1 ]heptane-2thiol. The pyrolysis of a thiol is believed to be initiated by fission of the C--SH bond, and H 2S is
quickly formed as a major product, along with an olefin.2 Some research has also been
conducted on disulfide pyrolysis. Available studies include the pyrolysis of dimethyl disulfide in
a static system, as well as di-t-butyl disulfide and a few aryl t-butyl disulfides, and these results
can be used to validate computational pyrolysis models.22 -24
As a major component of coal and heavy oil sands, a significant amount of research has been
done on the high temperature chemistry of aromatic sulfur compounds. In one of the coal-related
studies, Bruinsma et al. investigated the gas phase pyrolysis of thiophene, furan, pyrrole,
27
pyridine, and benzene, as well as each of their benzo- and dibenzo- derivatives.25- The main
finding of this work was that the sulfur-containing rings were significantly more stable than the
other compounds. As shown in Figure 2, temperatures of over 900'C were required for 10%
conversion of thiophene and benzothiophene in five seconds (dibenzothiophene required a
slightly lower temperature). The authors concluded that the initiation of thiophene pyrolysis
occurs by ring-H scission, which accounts for the similar stability of the benzene molecules; the
15
Chapter 1: Introduction
pyrolysis of other heterocycles is likely initiated by bond fission between a carbon and the
heteroatom. A computational study of thiophene decomposition predicted very high barriers in
the likely decomposition mechanism, suggesting that this decomposition would not appreciably
occur in refinery processes without catalysts. 28
0--a
'73-b
r-r
97
1073
1173
-
Q273
T/K
-
873
Figure 3. The temperature required for 10-50% conversion (in five seconds) of each heterocycle (a), benzo29
derivative (b), and dibenzo- derivative (c). Figure from Bruinsma et al.
More research has been done specifically on the topic of benzothiophene and dibenzothiophene
pyrolysis. Dartiguelongue et al.28 looked at dibenzothiophene pyrolysis between 375 and 500'C,
and this work showed that desulfurization occurred more at lower reactant conversions, while
radicals combined to form heavy products and high conversions. A mechanism was proposed to
explain this process, but it was not confirmed experimentally.
1.3 Supercritical water (SCW) treatment of crude oil
The critical point of water is at 374'C and 22.06 MPa.30 As this point is approached and
exceeded, the properties of water vary in ways that make it a candidate medium for crude oil
treatment. One of the most important properties for this application is the dielectric constant,
which continuously decreases until shortly after the critical temperature, where it settles at a very
low value (as seen in Figure 3). This is evidence that hydrogen bonding becomes less prevalent
as the temperature increases. The decrease is significant enough that supercritical water can be
16
Chapter 1: Introduction
used as a nonpolar solvent for hydrocarbons. The mixing of supercritical water and various
'
hydrocarbons has been further studied using transport models.3
a x 10; p /kg m
3
-logKw
10
1400 .
3
~14
1000
218
22
600 -
26
200
0
473
673
T)K
ionic product (3) of water
Figure 4. Temperature dependences of the dielectric constant (1), density (2), and
32,33
&
Lunin.
at a pressure of 24 MPa. Figure from Galkin
As the temperature increases within the subcritical region, the ionic product of water increases to
a maximum, three orders of magnitude greater than at standard conditions. However, this value
quickly decreases as you pass the critical point, and it monotonically decreases in the
supercritical region. This means that subcritical water (near the critical point) is a much greater
source of H+ and OH- ions, and ionic chemistry should be less important in the critical region.
Using this knowledge, multiple researchers have attempted to remove sulfur from organic
compounds by SCW treatment. Katritzky et al. studied the kinetics of thiophene and
benzothiophene, as well as a variety of sulfides and thiols containing aromatic rings, in SCW at
4600C." The thiols and sulfides were desulfurized to varying degrees, with diphenyl sulfide
being notably less reactive than the others, likely due to the lack of abstractable hydrogen to
form a reactive intermediate in decomposition. Thiophene was found to be completely unreacted
after one hour in SCW (and in 15% solutions of formic acid or sodium formate), while only
1.9% of benzothiophene was desulfurized after one hour in formic acid solution, which may be
within the margin of error of the experiment. In addition, Vogelaar et al. 4 studied pretreated
gasoil that had been spiked with benzothiophene, dibenzothiophene, diphenyldisulfide, and
octadecanethiol. The results found were similar to those of Katritzky et al., as only the thiol and
WS
Chapter 1: Introduction
17
disulfide were desulfurized at 673 K. Finally, Townsend et al.35 treated dibenzothiophene in
550'C SCW, and found that this also resulted in the same small conversion, about 2% after four
hours, as dry pyrolysis. These studies provide evidence that the addition of SCW alone is not
sufficient to desulfurize aromatic hydrocarbons like benzothiophene.
The desulfurization of aromatic hydrocarbons has also been explored using alkaline SCW, and
these experiments have provided significantly better results. Yoshida et al.36 treated thiophene in
NaOH concentrations ranging from 1.0 to 5.0 mol/dm3 , and a maximum conversion of 58% was
reported in 20 minutes at 400'C, with a NaOH concentration of 4.0 mol/dm 3 . Kishita et al.3 7 also
studied this type of desulfurization using NaOH and KOH. The model compound study in an
NaOH solution showed that about 70% of benzothiophene was converted in one hour at only
300'C, while 60% conversion of dibenzothiophene was reported at 430'C. Kishita also explored
the desulfurization of bitumen-hydrocarbon mixtures containing a much greater heavy
fraction-in alkaline SCW at 430'C. This resulted in about 60% desulfurization after one hour
using a 5 mol/dm 3 KOH concentration, and the reaction led to an overall decrease in the pH of
the aqueous phase. Because of this, the authors concluded that the base was consumed while
causing the bitumen to break down into benzothiophenes and dibenzothiophenes. From here, the
benzothiophenes were decomposed to result in most of the desulfurization, while the
dibenzothiophenes were desulfurized to a much lesser extent due to their greater stability. Kishita
proposed that the sulfur was removed as S 2 , and this conclusion was supported by Yoshida et
al., who found that this ion was the only sulfur product in the alkali SCW desulfurization of
thiophene. However, it is important to note that the metallic reactor walls may have been acting
as catalysts in these studies.
Recently, Patwardhan et al. used a continuous stirred tank reactor to study the kinetics of the
SCW treatment of hexyl sulfide and benzyl sulfide, and the resulting 3/2 order conversion
suggested that the decomposition rate was controlled by a free-radical mechanism. 38 ' 39 Kida et
al. proved that SCW is a reactant in the desulfurization mechanism of hexyl sulfide, and
intermediate studies were used to propose a combined free-radical and molecular mechanism,
presented in Figure 4, to explain the observed products.4 0 This not only has the effect of
removing some sulfur from the reaction mixture, but it also prevents the reactive thioaldehyde
18
Chapter 1: Introduction
compounds (containing a carbon-sulfur double-bond) from undergoing addition and cyclization
reactions that could form undesired heavy products.
CSH
COCO
1 1
S
C5
H11
C5H12
.
C 5H 11
CSH1I AO
CSH1
CA11-tS
C6H14
C 5H 1 _11
C6H 13
C 5H
11
S
OH
H2S
CSH 11SHH2
(H20/
Figure 5. Proposed mechanism for conversion of hexyl sulfide to pentane and hydrogen sulfide in SCW
treatment. Figure from Kida et al'
Work in the Green Group has agreed with the previous findings that thiophene rings are
recalcitrant to SCW treatment," but the alkyl chains connected to these rings are likely to break
at these conditions. Analysis of crude oil fractions (before and after SCW treatment) by twodimensional gas chromatography, as shown in Figure 5, has demonstrated a shift in the peaks of
sulfur compounds from the heavier to the lighter range, while a much heavier product
distribution was obtained from anhydrous pyrolysis of the same feedstock. This decrease in
density, or upgrading, would provide a significant increase in the value of the oil, suggesting that
40 41
SCW treatment has the potential to be a viable commercial process. '
Chapter 1: Introduction
Chapter 1: Introduction
19
19
Figure 6. GCxGC-SCD chromatograms for crude oil bottoms before (a) and after (b) SCW treatment at 450
*C. x-axis shows volatility (heavier compounds on right), y-axis shows polarity. Figure from Kida et al.42
1.4 Automatic mechanism generation
Over the past few decades, automatic mechanism generation has become a viable replacement to
the manual creation of free radical mechanisms. Systematization of the model-building process
allows for the use of the vast chemical knowledge available in literature, provides mechanisms
much more quickly than possible by hand, and prevents many of the common pitfalls in manual
work, such as biases and transcription errors. Several other groups have developed software that
also automates this process. 42 Our effort has focused on the development of the open-source
Reaction Mechanism Generator (RMG). 43 44 RMG-Java was used for most of the work in this
thesis, although a version is also under development in Python.
The algorithm and use of this software have been extensively described in the thesis of Joshua
Allen, 45 so only a general introduction will be provided here. The key feature of RMG is its
iterative, flux-based model expansion algorithm, which allows for the detailed pursuit of reaction
pathways that are predicted to have a greater flux, while placing less emphasis on those that are
predicted to be minor. Only a small amount of input is required from the user to generate an
initial mechanism, including the reactant geometries and concentrations, temperature, pressure,
and termination criteria. More advanced features have been implemented recently, including
20
Chapter 1: Introduction
pressure-dependent rate calculations, on-the-fly quantum calculations, solvation chemistry, and
chemistry for molecules containing nitrogen and sulfur, the latter if which is the topic of this
thesis.
1.5
Parameter estimation
Accurate kinetic and thermochemical data have been collected experimentally for thousands of
reactions over the past hundred years, while reasonably accurate calculations have been
conducted to calculate parameters for perhaps tens of thousands more. Unfortunately, the entire
database of the kinetics community does not come close to capturing the billions of reactions
possible in nature, or even the millions of reactions that may be considered in one RMG run!
Thus, algorithms are implemented to use the available data to estimate parameters that are as
accurate as possible. Enthalpies, entropies, and heat capacities can be captured by group
additivity schemes, of which the most widely-used was defined by Benson and Buss46 and
expanded by Vandeputte et al. for organosulfur compounds. 47 An example application of this
Group Additivity Value (GAV) scheme is presented in Figure 7. These GAV's were regressed
from experimental measurements and ab initio calculations for a variety of compounds, so they
can be updated when additional data are available. Additional corrections are available to
estimate thermochemical parameters for rings, radicals, and many other special cases.
0
lI
\C=
0
/CH
CH
2
S
0
0
AHfi
98 =
+ H1
98
+ Hf,298+ 2H 2 98 +4Hf 1 9 8
Figure 7. Application of group additivity scheme to estimate enthalpy of formation for
2,3-dihydrobenzothiophen-3-one.
21
Chapter I: Introduction
The estimation of reaction rate parameters can be more complicated, especially for bimolecular
reactions. When RMG identifies a species or set of species as matching a specific reaction type, a
tree structure is used to determine the most similar reaction for which reaction rate data are
available. This works well in some cases, but in others it leads to estimation by averaging a large
number of other rate constants, providing parameters with uncertainties of multiple orders of
magnitude. If a poorly-estimated reaction rate is determined to be important to the overall
predictions of a mechanism, either by sensitivity analysis or chemical intuition, improved
parameters must be obtained. This can be done via experiments that isolate the reaction of
interest, or via ab initio calculations. As the former method is generally more accurate but also
more expensive, it is most valuable when there is a small number of reactions that need to be
known with very high accuracy. The second method was utilized for this thesis, as rate
parameters needed to be calculated for a large number of new reactions.
1.6 Ab initio calculations
Ab initio is Latin for "from the beginning," and in the field of Physical Chemistry, it refers to the
determination of molecular properties by solving the Schr6dinger equation,
HW = EW,
where H is the Hamiltonian operator, E is the energy of state W, and T is the wave function,
which describes the state of the electrons in the system. The solution of this equation gives us
exact physical and thermochemical parameters for a molecule of interest, but unfortunately, this
exact solution is only possible for one-electron systems. A wide variety of models have been
proposed to approximate the solution, including Hartree Fock (HF) and Density Functional
Theory (DFT).2 Composite methods, such as CBS-QB3, have also been successful in the past,48
but CCSD(T)-F12 has recently been shown to provide more accurate energy calculations with a
similar computational cost. 49'50
Quantum calculations in this thesis were conducted using the Gaussian 0351,
52
and Molpr5 3
software packages. The exact methods used vary, so the exact details can be found in each
chapter. For species optimizations, N-dimensional scans, where N is the number of rotatable
bonds in the compound, at the B3LYP/6-3 IG(d) level of theory or higher, were first conducted
Chapter 1: Introduction
22
to roughly identify the conformer with the lowest energy. Local optimizations, frequencies, and
single-point energies were then calculated using higher levels of theory. The CanTherm software
package5 4 was used to calculate reaction rate constants using transition state theory, assuming
reactions were occurring in the gas phase.
1.7 Thesis Overview
This thesis focuses on the use of automatic mechanism generation to gain new insights into
processes relevant to our use of the planet's fossil fuels, which previously had only been
understood at a high level from bulk experimental data. In particular, we focus on generating
mechanisms for the decomposition of sulfur compounds, exploring how these compounds
interact with other species present in fuel sources, and improving the data available for
generating accurate predictions in the future.
Chapter 2 presents an application of the RMG algorithm on the pyrolysis of t-butyl sulfide.
Previous modeling work had focused on modeling the decomposition of sulfides containing an
alpha-hydrogen (hydrogen bonded to the carbon adjacent to the sulfur atom), the donation of
which provides a key propagation step in their decomposition mechanisms. This chapter
validates the database for sulfur compounds without an alpha-hydrogen, and additionally
demonstrates the ability of RMG to explain and quantitatively test the key reaction steps for
mechanisms that could only be hypothesized in the past.
In Chapter 3, the RMG databases for thermochemistry and reaction kinetics are expanded to
allow for the simulation of organic mixtures containing both oxygen and sulfur. Experimental
and calculated data for this type of chemistry are very sparse in literature, so accurate coupledcluster calculations were conducted to provide rate parameters for hydrogen abstraction and
radical addition reactions that involve both heteroatoms. A database was also built for the
addition reactions of water and hydrogen sulfide to double bonds, as these reactions were
proposed as key steps in the supercritical water desulfurization of sulfides.
Chapter 4 shows the application of the expanded RMG database to study the pyrolysis and
supercritical water treatment of hexyl sulfide. The resulting mechanisms provided elementary
Chapter 1: Introduction
23
reaction pathways to explain the formation of all of the major products from both treatments, as
well as most of the minor products, and product predictions agreed well with quantified
experimental product distributions. The main difference between the two mechanisms was found
to be the fate of the reactive thioaldehyde intermediate, which can continue to form aromatic
thiophenes in the absence of water, and potentially undesired heavy products as well. This
intermediate is desulfurized in SCW to form lighter products. Thus, this work not only shows
one of SCW's effects in desulfurization chemistry, but in the overall crude oil upgrading process
as well.
In Chapter 5, an RMG mechanism is presented for the dealkylation of hexyl benzene in pyrolysis
and SCW treatment. As thiophenes and benzenes are known to be particularly stable in the
absence of catalysts, the breaking of alkyl chains connecting multiple aromatic rings may be a
key to the overall crude oil upgrading process. The work in this chapter shows that SCW alone
does not directly react with hexyl benzene or its intermediates, as the model predicts the same
product distribution in the presence and absence of SCW, a result validated by experimental data.
Chapter 6 presents an application of RMG on understanding the effects of sulfur compounds on
the geological dealkylation of aromatic compounds. Decomposition mechanisms were built for
phenyldodecane and diethyl disulfide, which have previously been used as model heavy oil and
sulfur compounds in experimental studies. Mechanisms were generated to be valid both at
conditions previously studied experimentally and at conditions more relevant geologically,
where experiments cannot be completed on a reasonable timescale. As expected, the disulfide
and alkyl aromatic compound reacted on vastly different timescales, and this could have
important implications in organic geochemistry.
Chapter 7 provides a database of high-accuracy thermochemical parameters for organosulfur
compounds. Enthalpies, entropies, and heat capacities were first calculated for a training set of
species for which reasonably accurate experimental thermochemical parameters are available,
and the comparisons between calculated and experimental values were used to generate
additivity corrections for enthalpy calculations. The database was then expanded to include
parameters for additional sulfur compounds, which are used as a starting point for the generation
Chapter I.: Introduction
24
of additivity parameters for the estimation of sulfur compounds in pyrolysis and oxidation
chemistry.
Chapter 8 presents overall conclusions of this work, as well as some thoughts on future areas of
research in the predictive modeling of organosulfur chemistry. This not only includes challenges
that may be faced computationally, but also experimental work that will be critical to decreasing
the uncertainty, and increasing the value, of models built using these techniques.
References
1.
2.
3.
V. A. Gembicki, T. M. Cowan and G. R. Brierley, Hydrocarbon Processing, 2007, 86.
C. S. Song, in National Meeting of the American Chemical Society, Boston, MA, 2002.
H. H. Schobert, in Chemistry of Fossil Fuels and Biofuels, Cambridge University Press,
New York, 2013.
S. J. M. Butala, J. C. Medina, T. Q. Taylor, C. H. Bartholomew and M. L. Lee, Energy
Fuels, 2000, 14, 235-259.
M. D. Lewan, Geochimica et Cosmochimica Acta, 1997, 61, 369 1-3723.
M. D. Lewan and S. Roy, Organic Geochemistry, 2011, 42, 31-41.
M. D. Lewan, Nature, 1998, 391, 164-166.
C. A. Islas-Flores, E. Buenrostro-Gonzalez and C. Lira-Galeana, Energy & Fuels, 2005,
19, 2080-2088.
O. P. Strausz, T. W. Mojelsky and E. M. Lown, Fuel, 1992, 71, 1355-1363.
S. Mitra-Kirtley, 0. C. Mullins, C. Y. Ralston, D. Sellis and C. Pareis, Applied
Spectroscopy, 1998, 52, 1522-1525.
Y. Kida, C. A. Class, A. J. Concepcion, M. T. Timko and W. H. Green, Phys Chem Chem
Phys, 2014, 16, 9220-9228.
G. Martin, in The chemistry of sulphur-containingfinctionalgroups, John Wiley & Sons,
New York, 1993, pp. 395-437.
S. W. Benson and G. R. Haugen, J. Am. Chem. Soc., 1965, 87.
G. Martin, H. Martinez and J. Ascanio, InternationalJournalof Chemical Kinetics, 1989,
21, 193-206.
X. Zheng, J. W. Bozzelli, E. M. Fisher, F. C. Gouldin and L. Zhu, Proceedingsof the
Combustion Institute, 2011, 33, 467-475.
A. G. Vandeputte, C. A. Class, M.-F. Reyniers, W. H. Green and G. B. Marin,
Submitted, 2014.
A. G. Vandeputte, M.-F. Reyniers and G. B. Marin, J. Phvs. Chem. A, 2010, 114, 1053 110549.
A. G. Vandeputte, M.-F. Reyniers and G. B. Marin, ChemPhysChem, 2013, 14, 17031722.
A. G. Vandeputte, M.-F. Reyniers and G. B. Marin, ChemPhysChem, 2013, 14, 37513771.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
&
1.8
Chapter 1: Introduction
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
25
A. G. Vandeputte, M. K. Sabbe, M.-F. Reyniers and G. B. Marin, Phys Chem Chem
Phys, 2012, 14, 12773-12793.
A. G. Vandeputte, M. K. Sabbe, M.-F. Reyniers and G. B. Marin, Chemistry-A European
Journal, 2011, 17, 7656-7673.
W. Tsang, J. Chem. Phys., 1964, 40, 1498-1505.
T. 0. Bamkole, J. Chem. Soc. Perkin Trans., 1977, 2, 439-443.
D. E. Johnson and A. F. Dimian, J. Chem. Soc. Chem. Comm., 1987, 416-417.
J. A. R. Coope and W. A. Bryce, Can. J. Chem., 1954, 32, 768-779.
G. Martin and N. Barroeta, J. Chem. Soc. Perkin Trans. 2, 1976, 1421-1424.
G. Martin and J. Ascanio, J. Phys. Org. Chem., 1991, 4, 579-585.
0. S. L. Bruinsma, P. J. J. Tromp, H. J. J. D. Nolting and J. A. Moulijn, Fuel, 1988, 67,
334-340.
X. Song and C. A. Parish, J. Phys. Chem. A, 2011, 115, 2882-2891.
C. Dartiguelongue, F. Behar, H. Budzinski, G. Scacchi and P.-M. Marquaire, Org.
Geochem., 2006, 37, 98-1.16.
A. A. Galkin and V. V. Lunin, Uspekhi Khimii, 2005, 74, 24-40.
S. Dabiri, G. Wu, M. T. Timko and A. F. Ghoniem, JournalofSupercriticalFluids,
2012, 67, 29-40.
G. Wu, S. Dabiri, M. T. Timko and A. F. Ghoniem, Journal ofSupercriticalFluids,
2012, 72, 150-160.
A. R. Katritzky, R. A. Barcock, M. Balasubramanian, J. V. Greenhill, M. Siskin and W.
N. Olmstead, Energy & Fuels, 1994, 8.
B. M. Vogelaar, M. Makkee and J. A. Moulijn, Fuel ProcessingTechnology, 1999, 61,
265-277.
S. H. Townsend, M. A. Abraham, G. L. Huppert, M. T. Klein and S. C. Paspek, Ind. Eng.
Chem. Res., 1988, 27, 143-149.
S. Yoshida, K. Takewaki, K. Miwa, C. Wakai and M. Nakahara, Chem. Lett., 2004, 33,
330-331.
A. Kishita, S. Takahashi, F. Jin, Y. Yamasaki, T. Moriya and H. Enomoto, J. Jpn. Pet.
Inst., 2005, 48, 272-280.
A. Kishita, S. Takahashi, Y. Yamasaki, F. Jin, T. Moriya and H. Enomoto, J. Jpn. Pet.
Inst., 2006, 49, 177-185.
P. R. Patwardhan, M. T. Timko, C. A. Class, R. E. Bonomi, Y. Kida, H. H. Hernandez, J.
W. Tester and W. H. Green, Energy & Fuels, 2013, 27, 6108-6117.
Y. Kida, Massachusetts Institute of Technology, 2014.
Y. Kida, A. G. Carr and W. H. Green, Energy & Fuels, 2014, 28, 6589-6595.
W. H. Green, in Advances in Chemical Engineering, ed. G. B. Marin, Academic Press,
vol. 32.
A. S. Tomlin, T. Turinyi and M. J. Pilling, in Comprehensive Chemical Kinetics, ed. M.
J. Pilling, Elsevier, 1997, vol. 35, pp. 293-437.
W. H. Green, J. W. Allen, R. W. Ashcraft, G. J. Beran, C. A. Class, C. Gao, C. F.
Goldsmith, M. R. Harper, A. Jalan, G. R. Magoon, D. M. Matheu, S. S. Merchant, J. D.
Mo, S. Petway, S. Raman, S. Sharma, K. M. Van Geem, J. Song, J. Wen, R. H. West, A.
Wong, H.-W. Wong, P. E. Yelvington and J. Yu, Reaction Mechanism Generator
(R.MG), (2013).
Chapter I.: Introduction
46.
47.
48.
49.
50.
51.
52.
53.
54.
26
J. W. Allen, Massachusetts Institute of Technology, 2013.
S. W. Benson and J. H. Buss, Journalof Chemical Physics, 1958, 29, 546-572.
Y. Zhao and D. G. Truhlar, Accounts of Chemical Research, 2008, 41, 157-167.
J. A. Montgomery Jr., M. J. Frisch, J. W. Ochterski and G. A. Petersson, Journalof
ChemicalPhysics, 2000, 112, 6532-6542.
A. G. Vandeputte, M.-F. Reyniers and G. B. Marin, Theor Chem Account, 2009, 123,
391-412.
J. Aguilera-Iparraguirre, H. J. Curran, W. Klopper and J. M. Simmie, Journalof Physical
Chemistrv A, 2008, 112, 7047-7054.
W. Klopper, R. A. Bachorz, D. P. Tew, J. Aguilera-Iparraguirre, Y. Carissan and C.
Hattig, Journalof Physical Chemistry A, 2009, 113, 11679-11684.
M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman,
J. Montgomery, J. A., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar,
J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H.
Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T.
Nakajima, Y. Honda, 0. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J.
B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, 0. Yazyev,
A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokurna, G. A.
Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C.
Strain, 0. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V.
Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A.
Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham,
C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M.
W. Wong, C. Gonzalez and J. A. Pople, Gaussian 03, (2004) Gaussian, Inc., Wallingford
CT.
H.-J. Werner, P. J. Knowles, G. Knizia, F. R. Manby and M. SchUtz, Wiley
InterdisciplinaryReviews: ComputationalMolecularScience, 2011, 2, 242-253.
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
27
Chapter 2: Modeling the pyrolysis of
t-butyl sulfide
2.1
Abstract
The automated Reaction Mechanism Generator (RMG) is used to build reaction networks for the
thermal decomposition of t-butyl sulfide. Simulation results were compared with data from
pyrolysis experiments with and without the addition of a cyclohexene inhibitor. Purely freeradical chemistry did not properly explain the reactivity of t-butyl sulfide, as the previous
experimental work showed that the sulfide decomposed via first-order kinetics in the presence
and absence of the inhibitor. With additional reaction rate data from a library of unimolecular
decomposition reactions of sulfides and thiols, as well as other published kinetic parameters for
small-molecule sulfur chemistry, the agreement between model and data was significantly
improved for the inhibited case. The unimolecular decomposition of t-butyl sulfide to form
isopropene and t-butanethiol was found to be a key reaction in both cases, as it explained the
first-order sulfide decomposition. Cyclohexene had a significant effect on the available radical
pool, and this led to dramatic changes in the resulting product distribution.
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
2.2
28
Introduction
Sulfur compounds are important in many aspects of life, including food, fuels, and the
environment.1-3 Sulfur in fuel sources can lead to problems in processing and usage. Sulfur in
crude oil leads to some challenges, as a sufficient amount of these compounds must be removed
during refinement to satisfy governmental regulations and prevent the release of toxic sulfur
compounds into the environment and the poisoning of catalytic converters.4- 6 In pyrolysis and
steam cracking, sulfur compounds have a significant impact on the initiation, termination, and
product distribution7 , leading to undesired process variability.
Because of the importance of these compounds, significant experimental efforts have been
undertaken to understand their chemistry, from pyrolysis -14 to oxidation"'
'
to decomposition in
aqueous and supercritical environments. 3-16 Due to the complexity in building accurate
mechanisms to model these phenomena, most of the available literature on sulfur chemistry
includes only speculative mechanisms without quantitative product predictions, with the
exception of the study of diethyl sulfide pyrolysis by Zheng et al. 7 The development of
automated reaction mechanism generation software' 8' 19 has greatly aided in this effort. For
decades, this software was limited to the study of hydrocarbon species without sulfur.2 0
21
With
extensions to the software and the availability of improved estimation methods for the
thermochemistry and reaction rates of elementary organosulfur reactions, -24 it is now possible
to use computational tools to shed mechanistic insight on past experimental studies of sulfur
chemistry.
Martin et al. previously studied the pyrolysis of a variety of alkyl t-butyl sulfides,2 5 and most of
the observed product distributions supported the four-centered transition state suggested by
Benson and Haugen, which is presented in Figure 1. t-butyl sulfide8 , which was pyrolyzed
between 360 and 413 'C with and without the presence of a supposed radical inhibitor,
cyclohexene, was the main exception to the trend. The expected products of the four-center
reaction were observed in the presence of cylcohexene, but significantly different product
distributions were obtained in the case of neat pyrolysis. In this chapter, kinetic mechanisms built
by RMG are analyzed to clarify the key reaction steps for the two cases.
.
...
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
(CH
S-R
29
-S-R
I
]
+
H-S-R
H2
Figure 1. Proposed reaction pathway for the decomposition of an alkyl t-butyl sulfide.
2.3
Methods
The Java version of RMG was used to generate the mechanisms for this work.' 8 RMG uses an
iterative, flux-based algorithm to build reaction mechanisms. With a user-supplied set of "core"
species, RMG searches a database containing a set of specific reactions and a more general
library of reaction recipes, to determine all possible reactions for the given species. The products
of these reactions are then added to the model's "edge" if they are not already present in the
mechanism. The reaction mixture is then simulated with the user-specified conditions and initial
concentrations, until an edge species reaches the required flux to be added into the core. The
database is then searched for reactions with the new core, and the whole process repeats until the
specified termination condition is reached without any of the edge species reaching the required
flux for addition to the core. The final mechanism only contains the core species and reactions,
and this is output as a CHEMKIN input file. 27
Thermochemical parameters were calculated using the group additivity values developed by
Vandeputte et al.2 4 The CBS-QB3 database of rate parameters for organosulfur compounds,
including hydrogen abstraction, 2,
2
29
beta-scission,3 and homolytic substitution reactions, 3
was used to provide rate constant estimations for reacting sulfur compounds. Other parameters
for free-radical and molecular reactions, available in the RMG database, were used estimate rate
constants for reactions where sulfur was not present.
Reactions found to be of particular importance in the model predictions were refined further by
calculating single-point energies at the CCSD(T)-Fl2a/cc-pVDZ-F12 level of theory after
geometry optimizations, frequency calculations, and hindered rotor scans using B3LYP/6311 G(2d,d,p), with a scaling factor of 0.99 used for the frequency analysis. This coupled-cluster
method with F12 has been found to provide basis-set errors that are generally below 2 kJ/mol
using triple-zeta and greater basis sets for small molecule calculations.32-34 As will be shown in
Chapter 3, double-zeta calculations are within about 2 kJ/mol of the same using triple-zeta, and
30
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
should therefore be accurate within roughly 4 kJ/mol or 1 kcal/mol, although the true uncertainty
is likely somewhat greater. A double-zeta basis set was chosen for this work because of the
relatively large systems under consideration, which contain up to 10 heavy atoms.
After quantum calculations were completed in Gaussian 0335 and Molpro, 36 the open-source
CanTherm software package 3 7 was used to calculate rate constants between 300 and 2000 K,
including a tunneling correction using the Eckart method.38 The rate constants were then fit to
the modified Arrhenius form,
k(T) = A x T' x exp (
RTa,
where T is the temperature in Kelvin.
2.4
Results and Discussion
2.4.1 Calculation of Unimolecular Decomposition Rates
Martin & Barroeta proposed a set of unimolecular reactions for the decomposition of t-butyl
39
disulfide to explain the formation of isobutene and hydrogen disulfide from the initial reactant.
Transition states were found using quantum chemistry methods for these two reactions, and their
geometries can be seen in Figure 2. The analogous mechanism is also possible in the pyrolysis of
t-butyl sulfide, and optimized transition states for these reactions are presented in Figure 3.
[
~~H
S
ks
-CH2
H
-H
-H
2C
o
_
k
SH
SH
---
H2S2
H
1.
171.8
LOW
9
2.950
S
2.896
H
H
Figure 2. Proposed reaction pathway
39
(top) and optimized transition state geometries (bottom) for the
molecular decomposition of t-butyl disulfide. Distances (A).
31
31
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
H ---CH2
S
H2C- -H
10-
SH
jw-
H2S
L
1.9
.19
&I
=ME@
H
H
H1 6
H
Figure 3. Proposed reaction pathway (top) and optimized transition state geometries (bottom) for the
molecular decomposition of t-butyl sulfide. Distances (A).
Table 1. Calculated rate constants (using CCSD(T)-F12/cc-pVDZ-F12) for molecular elimination reactions.
A [s-], n [unitless], E. [kJ/molI, k s-1].
I
S "S<
l
H2 S2
S
logOA
n
Ea
k(380 'C)
11.65
1.04
214.5
2.6E-03
12.71
0.39
233.8
1.3E-05
12.51
0.89
239.4
7.7E-05
12.88
0.36
256.0
2.6E-07
+
SH
SH
+
S SH
+
S'SH
-
-
- H2 S
The calculated rate parameters are presented in Table 1. While the calculated unimolecular rate
constant for sulfide consumption falls within the experimental uncertainty of the overall rate
constant,8 the calculated rate for disulfide consumption is more than an order of magnitude
slower than what was observed experimentally. 39 Calculations using a larger basis set might
bring this prediction closer to the observation. The overall trend provides some insight into the
differences between the reaction mechanisms of t-butyl sulfide and t-butyl disulfide. While tbutyl disulfide undergoes the full molecular mechanism to form isobutene and hydrogen
disulfide-some of which can react further to form hydrogen sulfide-the elimination of H2 S
from t-butanethiol is slower than the other three reactions in Table I by two orders of magnitude,
suggesting that the consumption of t-butanethiol may occur more quickly by a free-radical
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
32
mechanism. This would explain the equal consumption rate of t-butyl sulfide with and without
the radical inhibitor, with the cyclohexene inhibiting the radical pathway for H 2 S formation from
t-butanethiol. The exact mechanism can be explored in more detail using RMG.
2.4.2 Pyrolysis of Neat t-Butyl Sulfide
RMG simulated the pyrolysis of neat t-butyl sulfide at 380 'C and 217 Torr, using no diluent, a
goal reactant conversion of 60%, and a core tolerance of 0.10. Complete convergence of the
mechanism was not possible with the available memory, so the simulation was terminated with
289 species and 2749 reactions. The resulting CHEMKIN input file was used to simulate the
conditions of the experiments by Martin & Barroeta.8 The main reaction pathways are presented
in Figure 4, with major products in boxes and intermediate products, which continue to form a
variety of other minor products, in dashed boxes. Pathway (a), homolytic scission of a C-S
bond, accounts for an appreciable amount (20%) of the overall sulfide decomposition, and it
provides most of the predicted isobutene production. This reaction occurs much quicker than
analogous bond-scissions of other hydrocarbons due to the weakness of the C-S single-bond;
this one is particularly fast due to the production of a tertiary t-butyl radical in addition to the
thiyl compound.
Pathways (b) and (c) provide the majority of the main product, isobutene. Pathway (b), which
starts with the molecular elimination reaction that directly forms t-butanethiol and isobutene, is
predicted to account for 66% of sulfide conversion. This is a sensible result based on the
experimental data, as this reaction would explain the overall first-order consumption of t-butyl
sulfide observed in the presence and absence of cyclohexene. Much of the thiol undergoes
abstraction of the hydrogen bonded to the sulfur to form a thiyl radical. This radical is also
produced in smaller amounts through pathway (c), which requires hydrogen abstraction from one
of the six methyl groups on t-butyl sulfide, prior to a beta-scission reaction that also forms
isobutene. The t-butanethiyl radical abstracts a hydrogen atom from one of the adjacent methyl
groups, and the resulting radical species undergoes beta-scission to form the mercapto radical,
which then abstracts hydrogen to form hydrogen sulfide.
Comparisons between the experimental data and RMG predictions are presented in Figure 5 and
Figure 6. These plots show excellent agreement between model and experimental data, as
reactant conversion is predicted within 20% of the experimental observation. Isobutene
4SH
33
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
production is predicted with slightly greater error, suggesting that the rate of production and
recombination of 2-methylallyl radicals from isobutene is slightly overpredicted.
ZS
R
(a)
(b)
11%
-W
+
--
(c)
11%
66%
20%
20%
RH
I'l -)
S
RH
33%
-
3%
RR
S S-
SH
H
4%
17%
R
RH
64%
SH
4%
17%R
RH
64%
RY
Rt-
R
Figure 4. Major reaction pathways for neat pyrolysis of t-butyl sulfide. Percentages represent proportion of
reacted sulfide proceeding through a pathway over 40 minutes.
34
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
)
-
80
40
-
0
U 20
-
-
,-,60
S
0
0
40
30
20
10
Time (min)
*1
100
-
100
-
Figure 5. Experimental 8 and simulated results for conversion of neat t-butyl sulfide at 380 *C.
0
)
0(
0
10
-
10
H 2S
0
1
0
-
S
0
1
10
30
20
40
-t
I
0
10
Time (min)
40
30
-
100
-
100
20
Time (min)
S
0
0
..........
I............
-
10
-
10
0
0
+SH
1
1
0
10
20
Time (min)
30
40
0
10
20
30
40
Time (min)
Figure 6. Experimental 8 and simulated results for major products of neat t-butyl sulfide pyrolysis, presented
in logarithmic scale as a percentage of initial reactant concentration.
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
35
2.4.3 Pyrolysis of t-Butyl Sulfide with Cyclohexene
A mechanism was generated for the pyrolysis of a 40:60 mixture of t-butyl sulfide and
cyclohexene (by mole), using the same termination criteria. The resulting CHEMKIN file from
the converged run includes 69 species and 392 reactions, and the predicted overall fluxes over 40
minutes are presented in Figure 7. The same three major pathways are predicted as in the
pyrolysis mechanism in the absence of cyclohexene. Again, pathway (b) dominates the sulfide
consumption mechanism, so we would expect this mechanism to follow the same first-order
kinetics (with roughly the same rate) as in the absence of the inhibitor, and this is what was
observed experimentally.
LI S
R
RH
(b)
(a)
R
C)
33%
2%
75%
22%
RH
2%
S.
+
8%
25%
32%
S
RH
-R*U
R
R
5%
RH
--
SH
3%
H*
+
SH
RH-N
11- 5%
c
RH R
0
RH->
3%
R
32%
RH
Figure 7. Major reaction pathways for pyrolysis of t-butyl sulfide in the presence of cyclohexene. Percentages
represent proportion of reacted sulfide proceeding through a pathway over 40 minutes.
36
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
The main difference between the predicted reaction mechanisms for the two cases is seen in the
relative rate of pathway (c) of Figure 7, in comparison with the pyrolysis case. While this radical
is produced at a similar rate as the pyrolysis case, cyclohexene readily donates hydrogen atoms
to reverse this reaction, causing the overall flux to be much lower. Cyclohexene also provides
hydrogen in the production of H2S and isobutane, resulting in a resonance-stabilized radical,
which can further donate hydrogen atoms to eventually form a small amount of benzene. These
compounds will also donate hydrogen atoms to the thiyl radical produced through pathway (a),
substantially decreasing the net flux from the thiol to the thiyl radical. Due to this decreased
consumption rate of the thiol, this compound should have a concentration roughly equal to
isobutene, as was observed experimentally.
A comparison of the experimental and predicted sulfide conversion is presented in Figure 8, and
the product predictions are compared with experiments in Figure 9. Conversion is predicted with
good accuracy, with the same small error as in the neat pyrolysis case. Major products are also
predicted reasonably accurately. The production of hydrogen sulfide is overpredicted with
approximately the same error as the underprediction of t-butanethiol, suggesting a slight error in
the t-butanethiol desulfurization pathway.
80
_
-
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
80
0
-
40
20
-
0
S
0
0
10
20
30
40
Time (min)
Figure 8. Experimental8 and simulated results for conversion of t-butyl sulfide in the presence of cyclohexene
at 380 *C.
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
37
37
100
S
0
-SH
10
1
0
40
30
20
10
Time (min)
100
100
S
0
10
0
0
-
:1<
0
10
-
S
0~
I
H 2S
I
0
10
20
30
Time (min)
40
0
10
20
30
40
Time (min)
Figure 9. Experimental and simulated results for products of t-butyl sulfide pyrolysis in the presence of
cyclohexene, presented in logarithmic scale as a percentage of initial sulfide concentration.
2.4.4 Mechanism Comparison
As seen in the previous section and in the experimental work, the addition of a cyclohexene
inhibitor has little effect on the overall sulfide decomposition rate, as the dominating reaction for
this process is the unimolecular decomposition reaction to form isobutene and t-butanethiol. This
is further emphasized using sensitivity analysis, which shows this reaction as the most sensitive
for sulfide decomposition by a wide margin. However, the presence of cyclohexene and the
radicals subsequently produced has been shown to have a significant effect on the resulting
product distributions. This can be seen in Figure 10, where major radical species concentrations
are plotted for the two cases. While the total concentration of radical compounds is similar in the
two cases, the presence of cyclohexene suppresses the concentration of radicals other than
38
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
cyclohexenyl. These stable radicals will abstract hydrogen more slowly than others, leading to a
slower radical-decomposition of intermediate products, primarily the thiol.
1.E-05
-
1.E-07
0
4a-J
L
1.E-09
i.E-13
0.00001
0.001
0.1
10
1000
Time (s)
Figure 10. Radicals produced in pyrolysis of t-butyl disulfide, in the presence (gray) and absence (black) of
cyclohexene. Isobutenyl (solid), t-butanethiyl (dashed), mercapto (dotted), and cyclohexenyl (double).
2.5
Conclusions
In this work, we have demonstrated the ability of automated mechanism generation software to
propose and validate mechanisms for organosulfur pyrolysis mechanisms, where only
postulations based on bulk experimental data were available in the past. In particular, RMG was
able to identify the most important reaction controlling the rate of t-butyl sulfide in the presence
and absence of a compound expected to inhibit the reaction rate. Coupled-cluster calculations
suggested that the unimolecular decomposition pathway could fully account for the
decomposition mechanism of the disulfide and the first step of the sulfide mechanism. RMG was
used to elucidate the reaction mechanism for t-butyl sulfide, demonstrating the free radical
mechanism for the decomposition of t-butane thiol and the lack of this decomposition in the
presence of the cyclohexene inhibitor.
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
39
References
1.
S. Patai and Z. Rappoport, The Chemistry of sulphur-containing functional groups,
Wiley, New York, 1993.
C. Song, Catalysis Today, 2003, 86, 211-263.
D. T. Johnston, Earth-Science Reviews, 2011, 106, 161-1,83.
V. Mohnen, Sci. Am., 1988, 259.
N. A. Fishel, R. K. Lee and F. C. Wilhelm, Environmental Science & Technology, 1974,
8, 260.
G. Corro, Reaction Kinetics and Catalysis Letters, 2002, 75, 89-106.
M. Bajus, Sulfur Rep., 1989, 9, 25.
G. Martin and N. Barroeta, International Journal of Chemical Kinetics, 1980, 12, 699716.
L. Xu, J. Yang, Y. Li and Z. Liu, Fuel Process. Technol., 2004, 85, 1013-1024.
J. K. Winkler, W. Karow and P. Rademacher, J. Anal. Appl. Pyrolysis, 2002, 62, 123141.
M. T. Timko, E. Schmois, P. R. Patwardhan, Y. Kida, C. A. Class, W. H. Green, R. K.
Nelson and C. M. Reddy, Energy & Fuels, 2014, 28, 2977-2983.
H. Mei, B. W. Mei and T. F. Yen, Fuel, 2003, 82, 405-414.
A. R. Katritzky, M. Balasubramanian and M. Siskin, Energy & Fuels, 1992, 6, 431-438.
A. R. Katritzky, R. A. Barcock, M. Balasubramanian and J. V. Greenhill, Energy
Fuels, 1993, 8, 498-506.
Y. Kida, C. A. Class, A. J. Concepcion, M. T. Timko and W. H. Green, Phys Chem
Chem Phys, 2014, 16, 9220-9228.
P. R. Patwardhan, M. T. Timko, C. A. Class, R. E. Bonomi, Y. Kida, H. H. Hernandez, J.
W. Tester and W. H. Green, Energy & Fuels, 2013, 27, 6108-6117.
X. Zheng, J. W. Bozzelli, E. M. Fisher, F. C. Gouldin and L. Zhu, Proceedings of the
Combustion Institute, 2011, 33, 467-475.
W. H. Green, J. W. Allen, R. W. Ashcraft, G. J. Beran, C. A. Class, C. Gao, C. F.
Goldsmith, M. R. Harper, A. Jalan, G. R. Magoon, D. M. Matheu, S. S. Merchant, J. D.
Mo, S. Petway, S. Raman, S. Sharma, K. M. Van Geem, J. Song, J. Wen, R. H. West, A.
Wong, H.-W. Wong, P. E. Yelvington and J. Yu, Reaction Mechanism Generator
(RMG), (2013).
A. S. Tomlin, T. Turitnyi and M. J. Pilling, in Comprehensive Chemical Kinetics, ed. M.
J. Pilling, Elsevier, 1997, vol. 35, pp. 293-437.
M. R. Harper, K. M. Van Geem, S. P. Pyl, G. B. Marin and W. H. Green, Combust.
Flame, 2011, 158, 16-41.
J. W. Allen, A. M. Scheer, C. W. Gao, S. S. Merchant, S. S. Vasu, 0. Welz, J. D. Savee,
D. L. Osborn, C. Lee, S. Vranckx, Z. Wang, F. Qi, R. X. Fernandes, W. H. Green, M. Z.
Hadi and C. A. Taatjes, Combust. Flame, 2014, 161, 711-724.
A. G. Vandeputte, M.-F. Reyniers and G. B. Marin, Theor Chem Account, 2009, 123,
391-412.
A. G. Vandeputte, M. K. Sabbe, M.-F. Reyniers and G. B. Marin, Phys Chem Chem
Phys, 2012, 14, 12773-12793.
A. G. Vandeputte, M. K. Sabbe, M.-F. Reyniers and G. B. Marin, Chemistry-A European
Journal, 2011, 17, 7656-7673.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
&
2.6
Chapter 2: Modeling the pyrolysis of t-butyl sulfide
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40
G. Martin, in The chemistry of sulphur-containing functional groups, John Wiley & Sons,
New York, 1993, pp. 395-437.
S. W. Benson and G. R. Haugen, J. Am. Chem. Soc., 1965, 87.
CHEMKIN-PRO 10131, (2013) Reaction Design, San Diego.
A. G. Vandeputte, M.-F. Reyniers and G. B. Marin, ChemPhysChem, 2013, 14, 37513771.
A. G. Vandeputte, M. K. Sabbe, M.-F. Reyniers, V. Van Speybroeck, M. Waroquier and
G. B. Marin, J. Phys. Chem. A, 2007, 111, 11771-11786.
A. G. Vandeputte, University of Ghent, 2012.
A. G. Vandeputte, M.-F. Reyniers and G. B. Marin, ChemPhysChem, 2013, 14, 17031722.
J. Aguilera-Iparraguirre, A. D. Boese, W. Klopper and B. Ruscic, Chemical Physics,
2008, 346, 56-68.
J. Aguilera-Iparraguirre, H. J. Curran, W. Klopper and J. M. Simmie, Journal of Physical
Chemistry A, 2008, 112, 7047-7054.
W. Klopper, R. A. Bachorz, D. P. Tew, J. Aguilera-Iparraguirre, Y. Carissan and C.
Hattig, Journal of Physical Chemistry A, 2009, 113, 11679-11684.
M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman,
J. Montgomery, J. A., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar,
J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H.
Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T.
Nakajima, Y. Honda, 0. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J.
B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, 0. Yazyev,
A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A.
Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C.
Strain, 0. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V.
Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A.
Liashenko, P. Piskorz, 1. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham,
C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M.
W. Wong, C. Gonzalez and J. A. Pople, Gaussian 03, (2004) Gaussian, Inc., Wallingford
CT.
H.-J. Werner, P. J. Knowles, G. Knizia, F. R. Manby and M. Schiitz, Wiley
Interdisciplinary Reviews: Computational Molecular Science, 2011, 2, 242-253.
S. Sharma, M. R. Harper and W. H. Green, CanTherm open-source software package,
(2010).
C. Eckart, Physical Review, 1930, 35, 1303-1309.
G. Martin and N. Barroeta, J. Chem. Soc. Perkin Trans. 2, 1976, 1421-1424.
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
41
Chapter 3: A kinetic database for organic
sulfur and oxygen compounds
3.1 Abstract
Potential energy surfaces and reaction kinetics were calculated for 40 gas-phase reactions
involving sulfur and oxygen. This includes 11 H 2 0 addition, 8 H2 S addition, 11 hydrogen
abstraction, 7 beta scission, and 3 elementary tautomerization reactions, which are potentially
relevant in the combustion and desulfurization of sulfur compounds found in various fuel
sources. Geometry optimizations and frequencies were calculated for reactants and transition
states using B3LYP/CBSB7, and potential energies were calculated using CBS-QB3 and
CCSD(T)-Fl2a/VTZ-F12. Rate coefficients were calculated using conventional transition state
theory, with corrections for internal rotations and tunneling. Additionally, thermochemical
parameters were calculated for each of the compounds involved in these reactions. With few
exceptions, rate parameters calculated using the two potential energy methods agreed reasonably,
with calculated activation energies differing by less than 5 kJ/mol. The computed rate
coefficients and thermochemical parameters are expected to be useful for kinetic modeling.
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
3.2
42
Introduction
Sulfur compounds can be found in almost every aspect of life, and their interactions with
oxygenated species play an important role in fuels, geochemistry, and environmental chemistry."
2 The formation of petroleum in geochemical reservoirs may be accelerated by the presence
of
weak carbon-sulfur bonds, and the reaction mechanisms of these species can be affected by the
presence of water.3 5
One of the most important sources of sulfur compounds is crude oil, and these compounds will
react to form toxic sulfur dioxide if not removed prior to combustion. The desulfurization of
crude oil has become a very important topic of study, as sulfur emission standards have tightened
and the availability of sulfur-lean feedstock has lessened.6 The current industry standard,
hydrodesulfurization, requires the use of hydrogen and expensive catalyst to achieve the proper
sulfur level, so multiple alternatives are being studied to potentially achieve similar results at a
lower cost. Oxidative desulfurization converts thiophenic compounds into more easily removable
polar compounds using hydrogen peroxide and a catalyst.7 Microbial desulfurization removes
sulfur from organic compounds at ambient temperature and pressure. 8 Treating oil with
supercritical water accomplishes desulfurization without the requirement of any catalyst.9
Work in supercritical water upgrading has demonstrated that water generates products with
reduced sulfur content and molecular weight.' 11 Water's involvement in this process has been
explored via model compound experiments, and some investigators have proposed pathways to
explain the reactivity of various sulfur compounds in aqueous and supercritical systems.12
Additional experiments and the advancement of computational chemistry techniques have
assisted in the elucidation of this mechanism, showing water to be both a reactant and a
hydrogen-transfer catalyst in the mechanism of alkyl sulfide desulfurization.9 Based on
intermediate studies and quantum chemistry calculations, a plausible pathway for water-aided
desulfurization was proposed, and this is shown schematically in Figure 1. In the proposed
mechanism, the water prevents the conversion of the reactive thioaldehyde (reactant 3) to an
oligomer, which is known to occur in the absence of water.1 3 Water participates by adding to the
carbon-sulfur double-bond in reaction c to form reactant 4, which readily reacts at high
temperature to form hydrogen sulfide, carbon monoxide, and a smaller alkane.
__WE4_z_
- __ -
. ",
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
-
43
0
Co9
CC.
a
12
CH
S8
*
C 5H
6
H14
11
1
'CSH 11
b
0
*
C
S
C 5H 1
QO
1 2
C5 H 11
O
d
H2 S
C 5 H 1< S
CH 13
C111
4
OH
C
C5HII> SH
H 2 0)
20
Figure 1. Proposed mechanism 9 for conversion of hexyl sulfide to pentane and CO 2
Many other pathways are possible, and a full kinetic mechanism of the system based on accurate
thermochemical and kinetic data is necessary to evaluate and validate them. Extensive libraries
of thermochemical data and reaction rate parameters for hydrogen abstraction, beta scission, and
substitution reactions involving organosulfur compounds have been generated by Vandeputte et
al. 14-16 Rate constants have also been calculated for small-molecule reactions involved in
combustion to form SOx compounds.1 7' 18 However, these data are not sufficient for accurately
modeling the reactions of thiols, sulfides, and thiophenes with oxygenated species. This work
focuses on the reactions of sulfur compounds and other species that are likely to be produced in
the presence of water at high temperatures. Many of the reactions considered here could also be
relevant to organosulfur combustion systems.
Rate parameters in modified Arrhenius form were calculated for 40 reactions that involve
organic sulfur and oxygen. These provide rate constants for use in simulations of hydrocarbon
mixtures including both sulfur and oxygen, as well as in training sets to develop more general
rate estimation rules. Thermochemical parameters, which are required for the calculation of
equilibrium constants and temperature changes in reacting systems, have also been computed for
each of the species involved in the reactions and compared to the limited data already available.
_-AAd
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
3.3
44
Methods
Thermochemical data were computed using the Gaussian 03 and Molpro quantum chemistry
packages.' 9 ',20 All species with an even number of electrons were calculated in their singlet state,
and radical compounds were calculated in their doublet states. Geometry optimizations and
frequency calculations were conducted using B3LYP/CBSB7 ,
2
and it was tested that all the
reactants and products were indeed minima on the potential energy surface and that all the
transition states showed one and only one imaginary frequency that corresponded to the expected
reaction coordinate. These geometries were then used for single point energy calculations at
higher levels of theory. Electronic energies were calculated using both the composite CBS-QB3
method22 '
23
and the explicitly-correlated CCSD(T)-Fl2a/VTZ-F12
method 24 -2 7 (this will be
referred from now on as CCSD(T)-F12).
The slow convergence of CCSD(T) with the basis set size has been known for a long time.28,
That restricted its application to very small systems. 3 0,
31
29
In the last few years explicitly-
correlated methodologies have been introduced to circumvent this problem. 2 ,'
3
They directly
address the fact that conventional coupled-cluster methods approximate wavefunctions based on
one-electron basis functions and can hardly describe the electron-electron correlation. This
drawback was overcome with the introduction of functions depending explicitly on the interelectronic distance, as used in the CCSD(T)-F12 family. That makes the basis set convergence
much faster and allows us to describe medium-sized systems with basis-set error of less than I
kcal/mol. These properties have allowed it to be successfully applied in all sorts of fields,
34 3 6
including thermochemistry and kinetics. CBS-QB3 has previously been used in a variety of kinetic studies, including some relevant to
sulfur chemistry, and the reaction barriers calculated have been shown to have an uncertainty of
a few kcal/mol. 2 2 ,
23, 37, 38
CBS-QB3 thermochemistry is usually more accurate due to the
availability of empirical Bond Additivity Corrections (BAC). 3 9 It appears that CBS-QB3 is
becoming obsolete, as new density functionals like M06 and BMK provide comparable accuracy
at a much lower cost, 4 0'
41
and CCSD(T)-F12 methods provide improved accuracy at a still-
reasonable cost. We include it here nevertheless since a big part of the data available in the
literature from the last two decades has been calculated in this way, so a good assessment of its
accuracy is still useful.
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
45
Partition functions were calculated using the CanTherm software package,42 using a scaling
factor of 0.99 for the frequency analysis. Enthalpies, entropies, and heat capacities were
calculated using CanTherm, including CBS-QB3 BAC's that are available in literature, 3 9 No
correction was available for the C=S bond due to the scarcity of experimental data for
thiocarbonyl compounds. Calculated parameters were used to generate NASA polynomials for
each of the reactants and products. These calculations were used to extend the group additivity
scheme for thermochemical properties, which was originally proposed by Benson and Buss, and
extended by Vandeputte et al. using CBS-QB3 for compounds containing sulfur.' 6 '
43,
44
Using
the thermochemical parameters calculated in this work, group additivity values (GAV's) of
enthalpy and entropy of formation, and heat capacities between 300 and 1500 K for 15 groups
containing both sulfur and oxygen were derived using the regression method discussed by
Vandeputte et al.' 6 Hydrogen Bond Increments (HBI's), as defined by Lay et al. ,
were derived
for two radical groups including sulfur and oxygen. The values for groups with previously
calculated GAV's (i.e. those that do not contain all of sulfur, carbon, and oxygen) were held
constant at the literature values.
Transition states were optimized for each elementary reaction, and transition state theory in
CanTherm was used to calculate rate coefficients under the ideal gas assumption, correcting for
the internal rotations of each single bond within the reactants, products, and transition states.
One-dimensional hindered rotations were used in the analysis, optimizing the geometries at the
B3LYP/6-31 1G(2d,p) level at 20-degree increments for each rotatable bond. Asymmetric Eckart
tunneling corrections were also calculated, and these corrections were applied to generate the
reaction rate constants between 300 and 2000 K.46' 47 Rate constants were fitted to the modified
Arrhenius form,
-Ea
k(T) = A * T' * eR*T,
where T is the temperature in Kelvin, R is the gas constant, A and n are fitted constants, and Ea is
the fitted activation energy. It is important to note that the fitted Ea is not the same as the reaction
energy barrier AEa, the calculated energy difference between the reactants and transition state
including zero-point energies (ZPE's). E, and AE, can differ by multiple kJ/mol. The modified
WAWMMM"" .I--_
.................
- --. 1-111111111
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
46
Arrhenius form has been demonstrated to fit rate constants for a variety of organic systems better
than the standard Arrhenius form without the
r
term.48 '49
These rate parameters were calculated assuming reactant activity coefficients a= 1. The activity
for water can vary significantly at supercritical conditions: for example, at the conditions of Kida
et al., 9 the activity coefficient of water is calculated to be approximately 0.5, reducing the rates
by this factor when water is a reactant. Thus, the rate parameters in this work should be adjusted
to account for the conditions being modeled to avoid introducing additional errors.
3.3.1 Calculation of Rate Constants for Reactions with Submerged Transition
States
The reaction barrier was calculated to be significantly negative (i.e. greater than the uncertainty
of the calculations) for two of the reactions studied in this work, implying the existence of
reactive complexes at lower energy levels than the reactants of the respective reactions. This type
of "submerged-barrier" reaction is illustrated in Figure 2. The same methods as discussed
previously for reactants and products were used to calculate energies and frequencies for the
reactive complex of each reaction.
k1
Reactantsk2O
Lk.
Complex
Products
Reaction Coordinate--Figure 2. Potential energy surface for a generic reaction with a submerged transition state.
The parameters for each submerged reaction were calculated for the high-pressure limit using
CanTherm. The rate k, for the formation of complex was assumed to be the collision rate, 1013
cm3/(mol* s), and k-I was calculated using thermochemical consistency. The rate of formation of
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
47
products from the pre-reactive complex, k 2, was calculated using transition-state theory. The
complex is short-lived, so it can be modeled using the quasi-steady-state approximation. The
overall rate of product formation for a reaction with two reactants is therefore
dC p
k1k 2
dt
k 1 + k2
1 R2
and the effective rate constant is
keff(T)
_k
=k
1
1
k
+2 k2
The effective rate constant keg(T) was calculated at temperatures between 300 and 2000 K, and
modified Arrhenius parameters were fit to these calculations to obtain the values reported in the
Tables for Reactions 21 and 37. As our primary interest is in supercritical water reactions (with
pressures greater than 200 bar), rate constants are reported in the high pressure limit. In some
gas-phase situations, the low-pressure limit might be more appropriate than the high-pressure
limit values reported here.
3.3.2 Basis Set benchmarking for CCSD(T)-F12 Calculations
A procedure to establish the accuracy of the basis set in the Coupled-Cluster calculations for our
particular set of reactions was defined as follows: in each of the different classes of reactions, the
one with the smallest number of electrons was used. These are reactions 1 for the molecular
additions of water, 12 for the molecular additions of hydrogen sulfide, 20 for the hydrogen
abstractions and 31 for the beta-scissions. The restriction on the size of the reactions allows us to
perform calculations on a bigger basis that would it be practical otherwise, and use its results as a
benchmark.
We obtained both reactants and transitions states for our set of reactions and performed a
consistent set of CCSD(T)-F12a calculations with the basis set series VDZ-F 12, VTZ-F12, and
VQZ-F 12.50 CCSD(T)-F12b energies were also calculated with the VQZ-F12 basis set, and these
agreed with the F 12a energies for the same basis set with an average error of 0.17 kJ/mol.
The convergence with respect to basis set is shown on Table 1. Triple-zeta F12 barrier heights
are converged to better than 1 kJ/mol. Double-zeta basis set on the other hand lead to errors
above 1 kJ/mol. This is in good agreement with previous studies.5l In a compromise between
48
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
accuracy and computational cost, we chose VTZ-F 12 basis set as the standard for this study. It is
important nevertheless to be aware of the error introduced by such a choice.
Table 1. Mean absolute difference in barrierheight (kJ/mol) calculated using double-, triple-, and quadruplezeta basis sets with CCSD(T)-F12a.
#
Reaction
|DZ-QZ|
ITZ-QZ|
1
1.21
0.01
12
20
31
Average
0.31
0.57
1.57
0.34
1.67
0.16
1.19
0.27
While the calculations reported here are converged with respect to basis set, this does not mean
they are exact. CCSD(T) is not full-CI, and there are several small neglected terms (BornOppenheimer breakdown, relativistic, anharmonicity) which can contribute errors on the order of
kJ/mol. Still, we expect that the numbers computed here are rather close to the true energies.
3.4
Results and Discussion
3.4.1 Molecular Addition of Water (Hydration of Double Bonds)
Reaction coefficients calculated for the ten reactions involving the molecular addition of water to
double bonds are presented in Table 2. These reactions progress via a four-membered ring
transition state, as depicted in Figure 3 for Reactions 1-9. Transition states for reactions 3, 8 and
9 were previously calculated.9,
52
All the other geometries were determined in this work and are
reported in the Supporting Information. Those geometries were used in this study. Reaction 8
corresponds to the addition of water to the thiocarbonyl group of carbonyl sulfide, while reaction
9 is addition to the carbonyl group. Reactions 10 and 11 are for the addition of water to the
carbon-carbon double-bond of thiophene, which also occurs through a four-membered ring
transition state.
S ---- H
o
R,S
+ 0I
2
~L
H
|
HO
|2
H
Figure 3. H 2 0 Addition to thiocarbonyl compound.
SH
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
49
The transition state of reaction I is presented in Figure 4. Calculated reaction parameters for the
molecular elimination of water from methanediol and ethanol are available in literature, and
using available thennochemistry data we can estimate the activation energy of these reactions in
the addition direction. 3 ,5 4 These are compared with the activation energies of reaction 1 in Table
3. The instability of thiocarbonyl compounds, which are known to polymerize at room
temperature, provides for a low-energy pathway for the conversion of this type of compound.' 3
Table 3 shows that the activation energy in both directions is lowest for the thiocarbonyl case, as
the 4-center reaction is much more facile for sulfur-containing systems than for C/H/O systems.
Lower A-factors and higher n-factors are calculated when a methyl or ethyl group is substituted,
as in reactions 2 and 3. This leads to rate constants that are within an order of magnitude at most
temperatures. Reactions 2 and 3 have very similar Arrhenius constant and n-factor, and a
difference of 4 kJ/mol for the activation energy in both directions, which suggests that increasing
the length of the thiocarbonyl compound has a minor effect, and will likely have a lesser effect as
this chain length increases. Reaction 4 has similar activation energies but a lower A-factor than
reactions 1-3 due to the presence of a methyl group on both sides of the thiocarbonyl group. We
compute slightly lower A-factors for the addition of water to benzenethial when compared with
reactions 1-3, but a significantly higher barrier height is predicted in the forward direction. The
transition state geometries for reactions 5 and 6, the addition of water to 2-propenethial and
benzenethial, are presented in Figure 5. The lengths of the C-S bonds in the two transition states
differ by less than 0.01 A, and this similarity is reflected in the rate constant calculations. CBSQB3 calculations on reaction 5 resulted in a reaction barrier of 145 kJ/mol, which is within 1
kJ/mol of the calculated barrier for hydration of benzenethial. As expected, very similar
Arrhenius parameters are calculated for the addition of water to a thioaldehyde bonded to an sp2
carbon.
I
J
WIiww
--------------------- .1.1 -------------------
50
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
Table 2. Modified Arrhenius coefficients for the molecular addition of water to sulfur-containing compounds.
A [cm 3 /(mol*s)i, n (unitless), E., AE, and AH*,,, (kJ/mol). Parameters for reaction 6 computed using
CCSD(T)-F12//B3LYP.
Forward Rate Parameters
Reaction
+
-
1.
SH
i0HO
11
log 10A
n
-0.62
AE 0
3.55
E1
101.8
122.7
AH*rxn
-54.1
-2.42
3.96
102.7
123.5
-48.1
-2.58
3.95
101.3
121.6
-46.7
-4.30
4.54
101.7
125.0
-46.9
-1.22
3.75
122.8
140.9
-28.4
-1.78
3.89
122.2
143.5
-29.6
-4.98
4.64
135.4
160.2
-8.3
-6.36
5.40
188.7
187.8
13.8
3.77
2.47
257.7
230.4
35.6
-26.0
11.0
191.9
269.3
33.9
-11.1
6.87
222.9
264.4
37.0
SH
8 + 10
2.
H
SH
1120
S
------
S
4.
12
-
+
1120
OH
+
- 11- -- i"............ ,
S
S
5.
H
SH
Ph
Ph
S+
1120
H
SH
- H2 0
7.
OH
H
H
0
f l()
8.
~
S
0
9.
OH
HS
cO
1-120
C//
OH
HO
S
+
10.
S
11.
OH
H2 0
S
+
11()0
H
51
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
1.784
1.4
1.572-
Figure 4. Optimized transition state for the hydration of thioformaldehyde. Distances (Angstroms).
Table 3. Reaction barriers (kJ/mol) for hydration of thioformaldehyde, formaledehyde, and ethene.
Reaction
H 2C
I
+ H2)0
S
0
+
H2C
CH2
SH
HO
OH
H-,O
HO
H0
O
+
H 2C
HO
AE0,
AE or
Ref.
123
173
this work
166
189
Kent [50]
209
254
Li [49]
1.
8
1.728
1 .747
A1.605
1.165
Figure 5. Transition states for the hydration of 2-propenethial (left) and benzenethial (right). Distances
(Angstroms).
Reactions 8 and 9 correspond to the addition of water to carbonyl sulfide, as investigated by
Deng et al.,5 and transition state geometries for these reactions (as calculated in literature) are
presented in Figure 6. The barrier height is calculated to be 43 kJ/mol greater for the addition to
the C=O bond than when water attacks the C=S bond. Comparing reaction 8 with the other
hydration reactions with thiocarbonyl groups, we see that addition to carbonyl sulfide requires an
activation energy more than 80 kJ/mol greater than reactions 1 through 4.
52
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
_L256
1,45
1
1. 15
1.128
I
1.152
1;641
09
1.
2
'V
Figure 6. Transition states for the hydration of carbonyl (left) and thiocarbonyl (right) group of carbonyl
sulfide. Distances (Angstroms).
Because of the aromaticity of thiophene, reactions 10 and 11 are endothermic, in contrast to the
exothermic addition of water to C-C double-bonds in alkenes. As such, these reactions proceed
via much higher-energy pathways, and the parameters calculated in this study show that water
will not appreciably react directly with thiophene at temperatures below 1500 K.
3.4.2 Molecular Addition of Hydrogen Sulfide
Reaction coefficients for eight reactions involving the addition of H 2 S to a carbonyl bond are
presented in Table 4. The optimized transition states for reactions 14 and 19 were available in
literature.:
The other geometries were determined in this work and are reported in the
Supporting Information. This type of reaction progresses in a similar fashion as the molecular
addition of water to a thiocarbonyl compound, via a four-member ring transition state as shown
in Figure 7.
0 ---- H
SH
HO
0|
R/ + H2S
R
H
Figure 7. H 2 S addition to a carbonyl compound.
53
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
Table 4. Modified Arrhenius coefficients for the molecular addition of hydrogen sulfide to carbonyl
compounds. A Icm 3 /(mol*s)I, n (unitless), E, AE,, and A H*,,,, (kJ/mol). Parameters for reaction 6 computed
using CCSD(T)-F12//B3LYP.
Forward Rate Parameters
Reaction
Reaction
12.
H2S
+
SH
HO
-
156.7
AE,
170.0
AH*rxn
-50.1
2.93
153.5
161.8
-34.1
1.49
2.96
152.0
159.9
-32.9
0.22
3.45
158.0
168.7
-26.1
2.58
2.72
151.8
159.1
-22.0
2.09
2.83
145.1
152.1
-20.5
-0.68
3.60
159.7
170.4
31.9
-9.20
6.38
199.1
204.0
40.7
logjOA
n
Ea
1.09
3.27
1.78
SH
HS
+
"
-
13.
H
SH
+
14.
Ii's
SH
15. 0
- ---
H2
OH
+
15.
.SH
1-12S
+
16.
H
SH
Ph
0 +
Ph
H2 S
-
*
17.
H
SH
112S
OH
-
+
18.
-
H
0
1
19.
C//
+
1's
HS
I
OH
This reaction occurs via a four-membered transition state, as in the addition of water to a doublebond, but the bond lengths and angles are greatly different. This is shown in Figure 8, which
shows the optimized transition state for reaction 12. An IRC scan confirmed that this transition
state corresponded to the expected reaction, and the potential energy surface was scanned using
stepping the C-S and 0-H bond distances while optimizing the
remaining variables. This is presented in Figure 9, and it shows that the reaction happens in a
somewhat sequential fashion, with the translation of the hydrogen atom to form an OH group
largely complete while the carbon and sulfur atoms are still separated by a distance of 2.7 A (in
b3lyp/6-311G(2d,p),
comparison with the final C-S bond length of 1.8 A). Thus, we expect that a separate
disproportionation pathway exists with a similar transition state, although the addition reaction's
transition state is over 100 kJ/mol more stable than the sum of the CH 2OH and SH radicals that
would be the intermediates of a disproportion-recombination pathway. This reaction type is also
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
54
a likely candidate for a roaming radical pathway, which has previously been investigated for the
decomposition of formaldehyde. 5 '
56
In addition, investigating the possibility of reaction
pathway bifurcation5 7 may be an area of future research for this type of reaction system.
The carbon-sulfur distance in Figure 8 is calculated to be 46% greater in the transition state than
the bond length in the product compound (compared to only a 12% difference for the carbonoxygen distance in reaction 1). This is reflected in the general trend of activation energies, where
the addition of water to a thioaldehyde is calculated to be a significantly more favorable reaction
than the addition of H2 S to an aldehyde.
t.84
1.
i
9Q9
2
2.728
2
Figure 8. Transition state for the molecular addition of H2 S to formaldehyde. Distances (Angstroms).
Similarly to the case with the addition of water to a thiocarbonyl compound, the reaction barrier
in both directions is slightly lower when an alkyl group is substituted on the carbonyl compound,
as shown by reactions 13 and 14. The transition states for these two reactions are presented in
Figure 10. Again, this effect decreases as the chain length increases, so the calculated rate
parameters for reaction 14 should be acceptable approximations for the addition of H 2 S to a
longer aldehyde. Substituting an alkyl group on both sides of the carbonyl group leads to the
prediction of a lower Arrhenius constant and greater n-factor is predicted for reaction 15.
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
H
SH + CHOH
S+CHO
220
-
3.6
55
200
3.4180
160
3
140
--
28
0
(I)
120
H--S
2-6H
100
6 2-4
80
22
60
-
1
1.2
1.4
40
-
2
2
18
16
0-H Distance (Angstrom)
2.2
2.4
Figure 9. Potential energy surface for Reaction 12. Energies (kJ/mol) relative to the mercaptoalcohol.
1.053
..
7 --
12
-
277-7-
2.762
Figure 10. Transition states for the molecular addition of H 2S to acetaldehyde (left) and 2-propanone (right).
Distances (Angstroms).
Substitution of a phenyl group stabilizes the transition state of this reaction. In contrast to
hydration reactions 5 and 6 which had very similar Arrhenius parameters, the energy barrier for
reaction 17 is calculated to be 7 kJ/mol lower than that calculated using CBS-QB3 for reaction
16. However, the rate constants estimated using these parameters agree within a factor of two at
temperatures above 600 K, and the disagreement will decrease at higher temperatures.
The optimized transition states of reactions 18 and 19 are presented in Figure 11. These are the
only ones in Table 4 calculated to be endothermic in the addition direction, as these require
addition to a stable carboxylic acid or carbon dioxide. The activation energies of these reactions
are calculated to be the greatest of the reactions calculated in the addition direction, but the
lowest in the H2S elimination direction. These transition states have the shortest carbon-sulfur
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
56
distance of any calculated for this type of reaction, and this length is 14% less for the addition of
H2 S to CO 2 than for the addition to acetic acid.
10681
1~~~
919.67
1.69
26982311'
Figure 11. Transition states for the endothermic addition of H 2 S to acetic acid (top) and carbon dioxide.
Distances (Angstroms).
3.4.3 Hydrogen Abstraction Reactions
Hydrogen abstraction reactions proceed when a hydrogen atom is abstracted by a radical species,
as shown in Figure 12. Modified Arrhenius parameters for the 10 hydrogen abstraction reactions
calculated in this work are presented in Table 5.
$t
H
/
+
R2
R,----H----R 2
H
R1 + R/
Figure 12. A hydrogen abstraction reaction.
Reactions 20-25 show the abstraction of hydrogen from a sulfur compound by an oxygen radical
center. The first four reactions are favored in the forward direction, due to the much greater
hydrogen-affinity of an oxygen atom relative to the sulfur atom. Linear transition states were
found for most of these reactions, which is typical for hydrogen abstractions. However, linear
and nonlinear transition states were found for reactions 20, and these are presented in Figure 13.
IRC scans were conducted for the converged geometries, and they showed that both versions of
each transition state corresponded to the correct reaction. Lower potential energies were
calculated using the bent transition state, so this geometry was used to calculate rate parameters
for this reaction.
57
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
Table 5. Modified Arrhenius coefficients for hydrogen abstraction reactions. Reaction 21 calculated for
overall pathway including pre-reactive complex at high-pressure limit. A [cm 3 /(mol*s)], n (unitless), Ea, AE,,
and AH*, ,, (kJ/mol).
Forward Rate Parameters
Reaction
Ea
AE 0
7.80
-2.8
4.4
-114.1
S+ H20
13.0
0.03
1.9
-12.3
-136.7
OH
4.32
2.44
5.0
14.9
-56.3
6.12
2.09
-2.0
-1.0
-78.9
-3.14
4.73
52.4
77.0
22.5
-4.59
5.08
54.1
79.4
-0.10
4.08
2.90
0.74
6.8
-9.8
-3.71
4.68
26.4
44.9
-19.4
SH + H
20. HbS + *0H1
21.
SH
22.
1
2S +
+
sH
23.
SH + I
24. H2S
+
25.
-SH
+
OH
S +
(
H
+
+
OH
OH
+
26.
+ SH
27.
+
K
SH
+ f2S
SH
0
OH
OH
28.
+
*CH3
+ *SH
29.
N
CH4
-0.29
3.74
15.6
35.2
-55.2
+
H2 S
5.04
2.47
3.1
18.1
-0.17
0.13
3.51
-3.6
4.9
-59.3
H
SH +S
o'
H+
OH
OH
30.
AH*0
n
1.71
1og, A
SH
+
'CH 3
0___
+
CH4
Reaction 20 has previously been studied in experimental
62
and theoretical 63 investigations. The
rate constants estimated in this work are compared with experimental data in Figure 13.
Although our TST calculation do not capture the negative temperature dependence at very low
temperatures (below 300 K), all of the methods employed in this work come within about 20%
of the experimental data at the temperatures relevant to combustion and pyrolysis.
58
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
0
1458
IL~
3
14340.
L434
3
1385
1.
1.621
1.
5
Figure 13. Linear (left) and angled (right) transition states optimized for reaction 20.
Distances (Angstroms) and angle (degrees).
-
-10.6
-10.8
-11
Q
-114
0.5
1
1.5
2
2.5
3
3.5
4
1000/T
Figure 14. Comparison of rate constant calculations (cm 3/molecules/s) with experimental data for Reaction
20. Lafage (o), Michael (x), Perry (0), Westenberg (+), Ellingson (hashed): M06-2X (black), MPWB1K (blue),
MPW1K (green), BBIK (red), This Work (solid): CBS-QB3 (red), CCSD(T)-Fl2a/VTZ-F12 (black),
CCSD(T)-Fl2a/VQZ-F12 (blue)
For reaction 21, a valid transition state was only found for the angled geometry. The energy of
the transition state for this reaction was calculated to be 12.3 kJ/mol lower than the initial
reactants, and a prereactive complex was optimized at an energy 19.4 kJ/mol lower than that of
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
59
the reactants, which is illustrated in Figure 14. The rate of reaction 21 approximately equals the
collision rate at temperatures above 400 K, and this is reflected in the optimized effective rate
parameters (the actual keis calculated are provided in the Appendix).
atL
TS
1379
-12.3
6
Reactants7
0
Complex
-19.4
Products
-135.1
0A~
it
Figure 15. Potential energy surface for reaction 21. Energies (kJ/mol), distances (Angstroms), angles
(degrees).
Nearly linear transition state geometries were found for reactions 24 and 25, as both saddle point
geometries had an O-H-S angle greater than 170'. Reaction 24 was the only one found to be
exothermic in the direction of hydrogen abstraction from the hydroxyl group, while reaction 25
is isothermal (within the margin of error for the calculations). This is in agreement with
-
published thermochemistry data, from which standard enthalpies of reaction are estimated to be
6
18.2 and -4.3 kJ/mol for reactions 24 and 25, respectively.1 '64-67
Reactions 26 and 27 represent the abstraction of an aldehydic hydrogen by a sulfur-containing
radical. Low activation energies are calculated for reaction 26 in both directions, while
abstraction of the hydrogen of the carbon adjacent to a thiol group is found to be significantly
less favorable. However, this activation energy is 27 kJ/mol lower than that calculated for the
abstraction of hydrogen from propane by acetyl radical to form isopropyl radical and
60
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
acetaldehyde 68 , as the alpha radical in a thiol or sulfide is stabilized by the presence of sulfur.
These two values are compared in Table 6.
Table 6. Forward reaction barriers (kJ/mol) for hydrogen abstraction reactions by the acetyl radical.
Reaction
SH
+
+
O
"
SH
+
AEO
Ref.
44.9
this work
67.9
Tsang [55]
Reactions 28-30 were calculated as possible intermediate steps in the desulfurization of alkyl
sulfides and thiols in supercritical water. Reactions 28 and 29 show significantly lower activation
energies than generally observed for the abstraction of a hydrogen from tetravalent carbon, as the
resulting radical is stabilized by the neighboring sulfur and oxygen atoms. Reaction 30 is highly
exothermic, and the radical formed in this reaction is stabilized by the carbonyl group. A
negative activation energy was fit to this reaction, but the ZE,) is positive and the positive
relationship between temperature and rate constant is expressed by the n-factor of 3.5.
3.4.4 Radical Addition to Double Bonds (Reverse Beta-Scission)
Modified Arrhenius parameters for the seven radical addition reactions calculated in this work
are presented in Table 7. These proceed via the pathway presented in Figure 15.
f
F-
R3
R1
R
3
+
2
-R 1
R2-
Figure 16. Radical addition to a double bond.
R3
R4
61
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
Table 7. Modified Arrhenius coefficients for radical addition to double bonds. Reaction 37 calculated for
3
overall pathway including pre-reactive complex at high-pressure limit. A (cm /(mol*s), n (unitless), E,, AE,
and AH*,.x
(kJ/mol).
Reaction
31.
HO
s
+
S
+
F
-
Forward Rate Parameters
i
Reaction
-
S.
HO
log 10A
n
Ea
AEO
AH*rxn
8.45
1.63
11.4
16.0
-132.1
4.36
2.35
23.0
28.5
-99.3
3.22
2.54
16.3
20.7
-96.0
9.30
1.21
-5.3
-0.41
-101.2
9.92
1.23
32.2
38.4
-38.8
6.91
1.68
54.2
59.1
-28.3
13.08
0.00
1.7
-9.3
-41.6
OH
32. HO
*CH3
OH
33- HO
S
+
'C2H
Sr
OH
34.
+
OH
SH
35.
+
36.
+
-
S
*CH3
Sr
SH
37.
+
'SH
_
OH
Optimized transition states for reactions 31-33 are presented in Figure 16. The reverse of
reaction 31, which forms thioformic acid and a hydrogen atom, is calculated to be significantly
less favorable than the beta scission reactions (reverse of 32 and 33) that form the same
thioformic acid and alkyl radicals. The transition state of reaction 34 is calculated to have a
slightly negative activation energy and barrier height.
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
62
LH Wj .'
1.
8
2.056
2.334
SJM
'WW
1.
1
Figure 17. Transition states for beta scission reactions 31 (top left), 32 (top right), and 33 (bottom). Distances
(Angstroms).
Reverse reactions 35 and 36 form the stable carbonyl sulfide. These are calculated to be
significantly less endothermic than reverse reactions 31-34; so as expected, much lower
activation energies are calculated in the beta scission direction, while greater activation energies
are predicted in the addition direction.
A significantly submerged reaction barrier was calculated for reaction 37, and a pre-reactive
complex was optimized near the transition state geometry. The potential energy surface of this
reaction is presented in Figure 17. The conversion of the pre-reactive complex to form the
product is calculated to occur significantly faster than the reverse reaction to reform the reactants
at temperatures greater than 400 K: thus, the overall kg is calculated to exhibit very little
temperature dependence and remain approximately equal to the collision rate (additional details
available in the Appendix).
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
63
*r . 5
2.327
TS
-9.3
Reactants
Complex
-14.1
2Product
1.
2.9051
Figure 18. Potential energy surface for reaction 37. Energies (kJ/mol), distances (Angstroms).
3.4.5 Tautomerization of Thiocarboxylic Acids
Three elementary tautomerization reactions were calculated in this work, and they are shown in
Table 8. These occur via the translation of a hydrogen atom from an alcohol group of a
thiocarboxylic acid to the sulfur atom, as shown in Figure 18.
R
SH
S,
S
OH
R
R
Figure 19. Tautomerization of a thiocarboxylic acid.
0
64
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
Table 8. Modified Arrhenius coefficients for elementary tautomerization reactions that include sulfur and
oxygen. A (s-1), n (unitless), E,, AE0 , and AH*,,,, (kJ/mol).
Forward Rate Parameters
Reaction
SH
S
38.
OH
H)
0
H"
S
OH
Ea
AEO
AH*rxn
1.72
3.26
83.6
112.2
-8.6
2.02
3.21
78.0
106.2
-9.9
1.94
3.23
78.9
107.1
-8.6
0
CH 3
S
SH
40.
C2 H5
n
SH
39.
CH 3
log 10A
OH
C 2 H5
0
The three reactions calculated in Table 8 proceed via very similar transition states, as shown in
Figure 19. Interatomic distances vary by less than 0.03 A between the saddle point geometries
for reactions 38 and 39, and the rate parameters calculated vary only slightly. The transition state
is stabilized to some extent by the substitution of an alkyl group, but this only leads to a
difference of 6 kJ/mol in the forward barrier height of reactions 39 and 40 in comparison with
reaction 38. Reactions 39 and 40 are calculated to have nearly identical Arrhenius parameters,
and Figure 19 shows that the relevant interatomic distances for these two reactions are nearly
identical. We expect that further increasing of the alkyl chain length should have a negligible
effect. Thus, the coefficients calculated for reaction 40 should be acceptable for elementary
tautomerization reactions of thiocarboxylic acids containing alkyl chains.
Based on the rate coefficients calculated for reaction 40, a thiocarboxylic acid with a C=S bond
would have a half-life of less than 0.1 s at temperatures above 500 K. It is recommended to
include this pathway in any model where this type of compound is likely to be produced.
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
65
1.650
1
2
1.
1.667
66
6
1.354
1
1.369
1.647
1.356
Figure 20. Transition states for reactions 38 (top left), 39 (top right), and 40 (bottom). Distances (Angstroms).
3.4.6 Thermochemical Library
Thermochemistry Group Additivity Values (GAV) 43 for the 15 groups calculated in this work
using CBS-QB3 are presented in Table 9, and Hydrogen Bond Increments (HBI) 45 for the two
radical groups are presented in Table 10. Previous comparisons with a small set of sulfur
compounds with experimental thermochemistry showed that these calculations are generally
accurate within 4 kJ/mol.16, 38 These groups are primarily relevant to the SCW pyrolysis of
sulfides and thiols; they represent a small subset of all possible groups containing carbon, sulfur,
and oxygen. Future expansion of this group library will be necessary for modeling more oxidized
69-73
systems, for which more extensive experimental data are available for benchmarking. - In
addition, regression of BAC and GAV using CCSD(T)-F12 for organic compounds should
provide more accurate estimates for thermochemical parameters, and these calculations are
discussed in Chapter 6.
66
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
Table 9. GAV for groups containing carbon, sulfur, and oxygen. Groups presented in Benson notation
AfH* (kJ/mol), S~in (J/mol/K) C,* (J/mol/K)
Benson Group Additivity Values
Group
C-(O)(S)(H)
'.
2
C-(C)(O)(S)(H)
CO,
AfH-
0
298 K
298K
300K
400K
500K
600K
800K
1000K
1500K
-11.58
4.58
7.48
9.54
11.00
11.91
12.85
13.54
14.93
-16.14
8.37
10.32
11.10
11.30
11.30
11.21
11.60
8.34
-11.10
sint
C-(C) 2 (O)(S)
-11.26
-39.73
8.16
10.15
10.69
10.52
9.74
9.01
C-(0) 2(S)(H)
-19.72
-13.26
6.36
8.72
10.13
10.88
11.56
11.91
12.53
C-(C)(O) 2 (S)
-21.41
-36.70
6.65
8.43
9.23
9.47
9.43
9.20
8.89
CO-(S)(H)
CO-(C)(S)
CO-(O)(S)
CS-(O)(H)
-9.84
-14.02
-11.53
2.85
29.36
8.55
5.51
4.37
4.94
4.48
6.16
5.04
6.70
5.51
6.35
6.09
7.17
5.83
6.99
6.82
8.06
6.29
7.59
8.05
8.79
6.48
7.76
8.99
9.83
6.38
8.18
10.37
-1.32
8.62
3.90
4.17
4.60
5.10
6.08
6.76
7.44
-22.72
2.67
3.08
3.59
3.90
4.03
3.99
3.75
3.23
6.98
5.56
8.05
5.78
8.35
6.31
9.10
6.73
9.48
7.00
9.95
7.60
10.38
7.61
10.65
8.34
11.59
8.52
11.62
9.31
12.26
8.99
12.26
9.77
12.99
9.29
13.25
10.14
CS-(C)(O)
CS-(0) 2
0-(CS)(H)
O-(CS)(C)
S-(CO)(H)
S-(CO)(C)
-31.38
-14.54
-21.06
-15.33
9.61
30.14
32.08
10.02
35.41
11.11
5.63
5.30
Table 10. HBI for radical groups containing carbon, sulfur, and oxygen. AfH* (kJ/mol), S'i, (J/mol/K) C*
(J/mol/K)
Hydrogen atom bond increment
Group
AfH 0
298 K
Ce-(C)(O)(S)
SO-(CO)
92.10
89.86
Sint
298 K
8.16
-1.26
C
300 K
-5.77
-9.75
400 K
-5.52
-11.80
500 K
600 K
800 K
1000 K
1500 K
-4.98
-13.39
-4.77
-14.85
-5.82
-17.41
-8.12
-19.29
-14.23
-21.42
3.5 Conclusions
Rate coefficients and thermochemical parameters were calculated for 40 reactions involving
sulfur and oxygen compounds. These have applicability in studies of sulfur chemistry in an
environment rich in water or other oxygenated species, such as the reactions of organosulfur
compounds in supercritical water reactors or in geological formations where water is present.
Although the calculation methods employed in this work are among the most accurate available,
rate coefficients calculated using these methods can still have greater than factor-of-2
uncertainty. In situations where more accurate rate parameters are required, experiments (if
possible) or calculations using higher-level quantum chemistry methods and improved treatments
of anharmonicity 74' 75 should be conducted. However, the parameters calculated in this work
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
67
provide a good starting point for the kinetic modeling of organosulfur chemistry in supercritical
water.
3.6 References
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
&
7.
C. Song, Catalysis Today, 2003, 86, 211-263.
D. T. Johnston, Earth-Science Reviews, 2011, 106, 161-183.
M. D. Lewan, Geochimica et CosmochimicaActa, 1997, 61, 3691-3723.
M. D. Lewan, Nature, 1998, 391, 164-166.
M. D. Lewan and S. Roy, Organic Geochemistry, 2011, 42, 31-41.
ETA. Retrieved 03/19/2014 from
http://www.eia.gov/dnav/pet/pet_pnp_crq_dcu -nusm.htm..
L. A. Gonzalez, P. Kracke, W. H. Green, J. W. Tester, L. M. Shafer and M. T. Timko,
Energy & Fuels, 2012, 26, 5164-5176.
J. D. V. Hamme, A. Singh and 0. P. Ward, Microbiology and Molecular Biology
Reviews, 2003, 67.
Y. Kida, C. A. Class, A. J. Concepcion, M. T. Timko and W. H. Green, Phys Chem Chem
Phys, 2014, 16, 9220-9228.
M. Morimoto, Y. Sugimoto, Y. Saotome, S. Sato and T. Takanohashi, The Journal of
SupercriticalFluids, 2010, 55, 223.
L.-Q. Zhao, Z.-M. Cheng, Y. Ding, P.-Q. Yuan, S.-X. Lu and W.-K. Yuan, Energy
Fuels, 2006, 20, 2067.
A. R. Katritzky, R. A. Barcock, M. Balasubramanian and J. V. Greenhill, Energy
Fuels, 1993, 8, 498-506.
N. J. Cooper, Compr. Org. Funct. Group Transform. 11, 2005, 3, 355-396.
A. G. Vandeputte, M. K. Sabbe, M.-F. Reyniers and G. B. Marin, Phys Chem Chem
Phys, 2012, 14, 12773-12793.
A. G. Vandeputte, University of Ghent, 2012.
A. G. Vandeputte, M. K. Sabbe, M.-F. Reyniers and G. B. Marin, Chemistry-A European
Journal, 2011, 17, 7656-7673.
K. J. Hughes, A. S. Tomlin, V. A. Dupont and M. Pourkashanian, Faraday Discuss.,
2001, 2001, 337-352.
P. Glarborg, D. Kubel, K. DamJohansen, H. M. Chiang and J. W. Bozzelli, International
Journalof Chemical Kinetics, 1996, 28, 773-790.
M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman,
J. Montgomery, J. A., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. lyengar,
J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H.
Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T.
Nakajima, Y. Honda, 0. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J.
B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, 0. Yazyev,
A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A.
Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C.
Strain, 0. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V.
Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A.
&
1.
2.
3.
4.
5.
6.
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
68
Liashenko, P. Piskorz, 1. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham,
C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M.
W. Wong, C. Gonzalez and J. A. Pople, Gaussian 03, (2004) Gaussian, Inc., Wallingford
CT.
H.-J. Werner, P. J. Knowles, G. Knizia, F. R. Manby and M. SchUtz, Wiley
InterdisciplinaryReviews: ComputationalMolecular Science, 2011, 2, 242-253.
A. D. Becke, Journalof Chemical Physics, 1992, 98, 5648-5652.
J. A. Montgomery Jr., M. J. Frisch, J. W. Ochterski and G. A. Petersson, Journal of
Chemical Physics, 1998, 110, 2822-2827.
J. A. Montgomery Jr., M. J. Frisch, J. W. Ochterski and G. A. Petersson, Journal of
Chemical Physics, 2000, 112, 6532-6542.
T. B. Adler, G. Knizia and H.-J. Werner, Journal of Chemical Physics, 2007, 127,
221106.
T. B. Adler, H.-J. Werner and F. R. Manby, Journal of Chemical Physics, 2009, 130,
054106.
G. Knizia, T. B. Adler and H.-J. Werner, Journal of Chemical Physics, 2009, 130,
054104.
K. A. Peterson, T. B. Adler and H.-J. Werner, Journalof Chemical Physics, 2008, 128,
084102.
R. J. Bartlett and M. Musial, Reviews of Modern Physics, 2007, 79, 291-352.
K. Raghavachari, G. W. Trucks, J. A. Pople and M. Head-Gordon, Chemical Physics
Letters, 1989, 157, 479-483.
J. D. Watts, J. Gauss and R. J. Bartlett, Journal of Chemical Physics, 1993, 98, 87188733.
J. M. L. Martin, Chemical Physics Letters, 1996, 259, 669-678.
W. Klopper and J. Noga, in Chemistry and Physics, ed. J. Rychlewski, Dordrecht, 2003.
J. Noga and W. Kutzelnigg, Journalof Chemical Physics, 1994, 101, 7738-7762.
J. Aguilera-Iparraguirre, A. D. Boese, W. Klopper and B. Ruscic, Chemical Physics,
2008, 346, 56-68.
J. Aguilera-Iparraguirre, H. J. Curran, W. Klopper and J. M. Simmie, Journalof Physical
Chemistry A, 2008, 112, 7047-7054.
W. Klopper, R. A. Bachorz, D. P. Tew, J. Aguilera-Iparraguirre, Y. Carissan and C.
Hdttig, Journalof Physical ChemistryA, 2009, 113, 11679-11684.
J. Zheng, Y. Zhao and D. G. Truhlar, J. Chem. Theory Comput., 2009, 5, 808-82 1.
A. G. Vandeputte, M.-F. Reyniers and G. B. Marin, Theor Chem Account, 2009, 123,
391-412.
G. A. Petersson, D. K. Malick, W. G. Wilson, J. W. Ochterski, J. A. Montgomery and M.
J. Frisch, Journalof Chemical Physics, 1998, 109, 10570-10579.
Y. Zhao and D. G. Truhlar, Accounts of Chemical Research, 2008, 41, 157-167.
A. D. Boese and J. M. L. Martin, Journalof Chemical Physics, 2004, 121, 3405-3416.
S. Sharma, M. R. Harper and W. H. Green, CanTherm open-source software package,
(2010).
S. W. Benson and J. H. Buss, Journalof Chemical Physics, 1958, 29, 546-572.
S. W. Benson, F. R. Cruickshank, D. M. Golden, G. R. Haugen, H. E. O'Neal, A. S.
Rodgers, R. Shaw and R. Walsh, Chemical Reviews, 1969, 69, 279-324.
I
...
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
69
T. H. Lay, J. W. Bozzelli, A. M. Dean and E. R. Ritter, J. Phys. Chem., 1995, 99, 1451414527.
H. S. Johnston and J. Heicklen, JournalofPhysical Chemistry, 1962, 66, 532-533.
C. Eckart, PhysicalReview, 1930, 35, 1303-1309.
J. A. Miller and S. J. Klippenstein, Journalof Physical Chemistry A, 2003, 107, 77837799.
J. S. Pilgrim, A. McIlroy and C. A. Taatjes, Journal of Physical Chemistry A, 1997, 101,
1873-1880.
F. A. Bischoff, S. Wolfsegger and D. P. Tew, MolecularPhysics, 2009, 107, 963-975.
A. Jalan, 1. M. Alecu, R. Meana-Pafieda, J. Aguilera-Iparraguirre, K. R. Yang, S. S.
Merchant, D. G. Truhlar and W. H. Green, J. Am. Chem. Soc., 2013, 135, 11100-11114.
C. Deng, Q.-G. Li, Y. Ren, N.-B. Wong, S.-Y. Chu and H.-J. Zhu, Journal of
ComputationalChemistry, 2007, 29, 466-480.
J. Li, A. Kazakov and F. L. Dryer, J. Phys. Chem. A, 2004, 108, 7671-7680.
D. R. Kent, S. L. Widicus, G. A. Blake and W. A. Goddard, Journalof Chemical Physics,
2003, 119, 5117-5120.
L. B. Harding, Y. Georgievskii and S. J. Klippenstein, J. Phys. Chem. A, 2010, 114, 765777.
L. B. Harding and S. J. Klippenstein, J. Phys. Chem. Lett., 2010, 1, 3016-3020.
D. H. Ess, S. E. Wheeler, R. G. Iafe, L. Xu, N. Celebi-OlgUm and K. N. Houk, Angew.
Chem. Int. Ed., 2008, 47, 7592-7601.
C. Lafage, J.-F. Pauwels, M. Carlier and P. Devolder, J. Chem. Soc. Faraday Trans. 2,
1987, 83, 731-739.
J. V. Michael, D. F. Nava, W. D. Brobst, R. P. Borkowski and L. J. Stlef, J. Phys. Chem.,
1982, 86, 81-84.
R. A. Perry, R. Atkinson and J. N. Pitts Jr., J. Chem. Phys., 1976, 64, 3237-3239.
G. S. Tyndall and A. R. Ravishankara, InternationalJournalof Chemical Kinetics, 1991,
23, 483-527.
A. A. Westenberg and N. deHaas, J. Chem. Phys., 1973, 59, 6685-6686.
B. A. Ellingson and D. G. Truhlar, J. Am. Chem. Soc., 2007, 129, 12765-12771.
M. W. Chase, Jr., J. Phys. Chem. Ref Data, 1998, Monograph 9, 1-195 1.
G. da Silva, C.-H. Kim and J. W. Bozzelli, J. Phys. Chem. A, 2006, 110, 7925-7934.
D. P. Tabor, M. E. Harding, T. Ichino and J. F. Stanton, J. Phys. Chem. A, 2012, 116,
7668-7676.
W. D. Good, J. L. Lacina and J. P. McCullough, J. Phys. Chem., 1961, 65, 2229-2231.
W. Tsang, Journalof Physical and Chemical Reference Data, 1988, 17, 887-952.
F. Turecek, L. Brabec, T. Vondrak, V. Hanus, J. Hajicek and Z. Havlas, Collect. Czech.
Chem. Commun., 1988, 53, 2140-2158.
W. K. Busfield, H. Mackle and P. A. G. O'Hare, Trans. Faraday Soc., 1961, 57, 10541057.
H. Mackle and D. V. McNally, Trans. FaradaySoc., 1969, 65, 1738-1741.
H. Mackle and P. A. G. O'Hare, Trans. FaradaySoc., 1961, 57, 1070-1074.
H. Mackle and W. V. Steele, Trans. FaradaySoc., 1969, 65, 2053-2059.
B. C. Garrett and D. G. Truhlar, Journalof Physical Chemistry, 1979, 83, 1915-1924.
J. Zheng, T. Yu, E. Papajak, 1. M. Alecu, S. L. Mielke and D. G. Truhlar, Phys Chem
Chem Phys, 2011, 13, 10885-10907.
70
Chapter 3: A kinetic database for organic sulfur and oxygen compounds
3.7 Appendix: Calculations for reactions 21 and 37
Rate parameters were calculated for the two reactions with submerged transition states using the
method described in the chapter, and the component rate coefficients are presented in the
following tables. keg provides the effective rate constant calculated at each temperature, while k/,q
shows the rate constant obtained using the best-fit modified Arrhenius parameters. The ratio of
the fitted rate constants to the rate constants calculated at each temperature show that low fitting
error was obtained for temperatures between 400 and 2000 K, but significantly greater error was
obtained at 300 K. Thus, the specific ke(300 K) should be used at this temperature.
Rate parameters for Reaction 21
T
300
400
500
600
800
1000
1500
2000
k,
1.OOE+13
1.OOE+13
1.OOE+13
1.OOE+13
1.OOE+13
1.OOE+13
1.OOE+13
1.OOE+13
k1
1.27E+11
1.69E+10
5.36E+09
2.58E+09
1.11E+09
7.12E+08
4.41E+08
3.80E+08
KI
1.27E-02
1.69E-03
5.36E-04
2.58E-04
1. 11E-04
7.12E-05
4.41E-05
3.80E-05
1.89E+10
9.28E+10
2.42E+11
4.62E+11
1.05E+12
1. 76E+12
3.60E+12
5.26E+12
keff
kfit
1. 30E+12
8.46E+12
9.78E+12
9.94E+12
9.99E+12
1.OOE+13
1.OOE+13
1.OOE+13
5.58E+12
6.81E+12
7.68E+12
8.34E+12
9.25E+12
9.86E+12
1.08E+13
1. 13E+13
kfit/keff
4.30
0.81
0.79
0.84
0.93
0.99
1.08
1.13
Rate parameters for Reaction 37
_
k-1
K1
k2
keff
kfit
kfit/keff
5.1OE+11
7.20E+10
2.33E+10
1.14E+10
4.91E+09
3.14E+09
1.96E+09
1.71E+09
5.10E-02
7.20E-03
2.33E-03
1.14E-03
4.91E-04
3.14E-04
1.96E-04
1.71E-04
2.82E+11
3.96E+11
4.93E+11
5.77E+11
7.19E+11
8.34E+11
1.04E+12
1.18E+12
3.56E+12
8.46E+12
9.55E+12
9.81E+12
9.93E+12
9.96E+12
9.98E+12
9.99E+12
6.23E+12
7.36E+12
8.13E+12
1.75
0.87
0.85
0.89
0.95
1.00
1.06
1.10
-
T
300
400
500
600
800
1000
1500
2000
1.OOE+13
1.OOE+13
1.OOE+13
1.OOE+13
1.OOE+13
1.OOE+13
1.OOE+13
1.OOE+13
8.69E+12
9.44E+12
9.93E+12
1.06E+13
1.1OE+13
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
71
Chapter 4: Modeling the desulfurization of
hexyl sulfide by supercritical water
4.1
Abstract
A detailed reaction network is proposed for the desulfurization of hexyl sulfide in the presence of
supercritical water (SCW) using the automated Reaction Mechanism Generator (RMG).
Experimental data have shown that pentane, carbon monoxide and carbon dioxide are products
of hexyl sulfide desulfurization in SCW, while none of these are detected in the simple pyrolysis
of hexyl sulfide. The observation of CO and CO 2 in the reaction products is a key result as it
provides evidence that water is acting as a hydrogen source for sulfur reduction. Several
pathways to generate these products from hexyl sulfide are proposed, and kinetic parameters for
the included reactions are calculated using transition state theory and quantum chemical
calculations at the CBS-QB3 level of theory. Using these rate parameters, as well as previously
calculated data from hydrocarbon and sulfur kinetic studies, reaction mechanisms were built
using RMG for the conversion of hexyl sulfide to H 2 S in the presence and absence of SCW.
Predictions from the RMG model agreed with reasonable accuracy with experimental data in the
presence and absence of SCW. Because the model and experiments were in good agreement, flux
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
72
analysis was used to identify the most important reaction steps, and sensitivity analysis was used
to propose reactions that should be studied further to decrease the model's uncertainty.
4.2
Introduction
The ability to generate predictive kinetic models is valuable for many systems, ranging from
process and product design' to engines 2' 3to large-scale atmospheric models. 4 The primary goal
of chemical kinetic models is to make predictions for what ensues in a particular mixture under
specified reaction conditions. The model prediction can then be compared with experimental
data for validation. In many systems of technological importance, the chemistry is complex,
involving hundreds of reactive intermediates. For this reason, systematic construction of
chemical kinetic models by computer algorithms and databases is becoming increasingly
attractive. Several research groups have developed software that automates model generation.'
6
Our own effort in this direction has resulted in the development of the Reaction Mechanism
Generator (RMG), an open source software package for automatic mechanisms generation.7
RMG offers several advanced features, including estimates of pressure dependent rates, thermochemical estimates of cyclic species, solvation effects on rates and the main subject of this thesis,
sulfur chemistry and reaction rates.>
Sulfur, a natural component of crude oil, can range in concentration from 0.1 wt% in "sweet"
samples up to 10 wt% in "sour" samples.'( The majority of the sulfur found in crude oil is
present as over 1000 different molecular structures, ranging from aliphatic sulfides and disulfides
to aromatic thiophenes and larger polycyclic benzothiophenes. 1 Sulfur content in crude oil has a
severe impact on oil production and refinery processes, making "sour" crude undesirable;
moreover, very low sulfur levels are required in most fuels to reduce engine emissions and
prevent poisoning of catalytic converters. However, the sulfur content of remaining crude oil
reserves is trending upward towards heavy "sour" crude, pushing oil companies to seek
alternatives to standard desulfurization techniques, such as hydrodesulfurization (HDS), which
.
requires a heterogeneous catalyst and high pressure H 2
Supercritical water (SCW) treatment of heavy "sour" crude presents advantages over current
desulfurization techniques for several reasons, especially because no catalyst or external
...........
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
73
hydrogen source is necessary. Thermal coker units are used for some heavy, sulfur-rich streams,
but an alternative, SCW treatment, is attractive because it minimizes the formation of difficultto-handle solid coke.12 , 13 Previous research suggests that the desulfurization of sulfur compounds
in SCW has the potential to be a sustainable, commercial-scale process.
4-2
However, large-scale
commercialization of SCW treatment of "sour" crude has been impeded by a lack of mechanistic
details and kinetic information.
To build a valid model for desulfurization in SCW, RMG was extended to include accurate
thermo-chemical and detailed rate information for sulfur compounds. A detailed database for the
thermochemistry and kinetics of sulfur compounds has been developed by Vandeputte et al. 2
This has been extended to include reactions and thermochemistry estimations for compounds
including both sulfur and oxygen, particularly for use in high-temperature models involving
sulfur compounds and water, as discussed in the previous chapter. The sulfur database in RMG
was validated by modeling the decomposition of diethyl sulfide at temperatures between 800 and
1000 K, and it succeeded to predict all of the major compounds of these experiments with
quantitative accuracy.24
The purpose of this paper is to first present the development and implementation of new sulfur
thermochemistry and kinetic calculations, with potential relevance to SCW desulfurization, into
RMG software and databases. Secondly, mechanisms are generated for the SCW desulfurization
of hexyl sulfide, as well as its pyrolysis without water. Predictions from these models are
compared and validated with new and recently published experimental results. 2 5, 26 Sensitivity
analysis is conducted to identify important reactions, and potential strategies for further model
improvement are proposed.
4.3
Methods
4.3.1 Batch Reactor Experiment
The experimental setup has been described in detail elsewhere, 2 6 and only a brief description is
provided here. A stainless steel batch reactor with an internal volume of 24 mL was used for all
experiments. Heating was provided to the reactor by a fluidized sand bath. For the base
experiment, hexyl sulfide and water were added to the reactor in a 1:3 ratio (wt/wt). To
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
74
investigate the concentration effect of water on the production of pentane, the loading of water
was adjusted while other components of the reaction mixture were kept constant. A total of 8
water/sulfide ratios were investigated experimentally, holding the sulfide loading constant, from
a ratio of 0 to 70 by mole. The base experiments were conducted at 400 'C and approximately 28
5 MPa to ensure that the water was in the supercritical state (374 C and 22.1 MPa). Air was
flushed from the reactor with helium, and 20 bar was left in the headspace upon sealing. The
same temperature was used for all experiments with different initial water loadings, and as
expected, the pressure increased with increasing water load. Once the reaction was quenched, the
gas-phase product and liquid phase products were collected and analyzed.
The gas-phase product was analyzed using conventional gas chromatography with either a flame
ionization detector or a thermal conductivity detector (GC-FID or GC-TCD, respectively). The
liquid oil phase was analyzed using two-dimensional GC with both an FID detector and a sulfur
chemiluminescence detector (GCxGC-FID and GCxGC-SCD, respectively). Bulk sulfur content
was measured with X-ray fluorescence (XRF). Representative water samples were analyzed for
sulfur and trace metals content. For more details on the analytical instruments and reactor, see
Kida et al.2 6
4.3.2 Batch Reactor Model
For simulating the batch reactor experiments, the reaction network was implemented in
CHEMKIN-PRO. The batch reactor was simulated using the closed homogenous reactor
model.2 7 The reactor was simulated as closely as possible to the experimental conditions, with a
five-minute linear heat-up from ambient conditions to the experimental temperature of 400 'C.
Initial reactant concentrations were calculated assuming a homogeneous mixture of hexyl
sulfide, water, and inert helium.
4.3.3 Continuous Flow Stirred Reactor (CSTR) Experiment
The experimental CSTR setup has been described in detail elsewhere,2 and as with the batch
reactor, only a brief description is provided here. The main reactor consisted of an Autoclave
Engineers (AE) bolted enclosure housing a 600 mL Inconel-625 vessel, rated for 345 bars at
650 C.28 The stirring mechanism was a magnetically driven AE Magnedrive that could achieve
stirring speeds up to 2500 rpm. Mixing was achieved using a combination of two mixers: a six-
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
75
blade radial turbine mixer attached at the tip of the impeller and a four-blade axial mixer located
-
at approximately 10 cm from the tip of the shaft. Water and hydrocarbon were fed separately
without pre-mixing - to the top of the reactor through two separate tangential feed side ports.
Product was withdrawn from the main bottom port.
The gas-phase product was analyzed in real time with an online conventional
gas
chromatography with a flame ionization detector. The liquid oil phase was analyzed using twodimensional GC with both an FID detector and a sulfur chemiluminescence detector (GCxGCFID and GCxGC-SCD, respectively). Bulk sulfur content was measured with X-ray fluorescence
(XRF).
4.3.4 CSTR Reactor Model
For simulating the CSTR reactor experiments, the reaction network was implemented in
Chemkin-Pro.2
The CSTR reactor was simulated using the perfectly stirred reactor (PSR)
model.
4.3.5 Principles of Automated Mechanism Generation
Reaction Mechanism Generator (RMG-Java, version 4.0.1) employs advanced methods in
thermochemistry and kinetic parameter estimation that allow the construction of complex
reaction networks. 7, 29, 30 RMG has demonstrated the ability to generate accurate kinetic models
ranging from pyrolysis of hydrocarbons to the low and intermediate temperature oxidation of
radicals during autoignition.3 1 ,32
At the heart of RMG is a fast rate-based algorithm that builds chemical kinetic models from an
initial set of reactants and initial conditions; temperature, pressure, and species concentration.3 2
RMG reacts the initial species in all possible ways referencing kinetic and thermodynamic
information stored in RMG's database and integrates the model in time. Each species generated
in the chemical model is classified as either an "edge" or "core" species. RMG tracks the rate or
flux of each new species produced and a species is transferred from the model edge to the model
core once the species flux exceeds a critical value set by the user. The species transferred from
the edge to the core are then reacted with already present core species in the model, to generate a
new set of edge species and reactions. In this iterative process the core and edge are expanded
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
76
and the process is repeated. At each step of model generation the resulting differential equations
are solved for all the species in the core. In this sequential way, a reaction network for the given
initial conditions is formed.
A detailed database for the thermochemistry and kinetics of sulfur compounds has been
developed by Vandeputte et al.21 23 This has been extended to include reactions and
thermochemistry estimations for compounds including both sulfur and oxygen, particularly for
use in high-temperature models involving sulfur compounds and water. ' High accuracy
calculations are already available for many hydrocarbons and small molecules. 34 Cumulatively,
these calculations provide reasonably accurate rate estimations for a very small percentage of all
of possible reactions that can occur. The other rate parameters are estimated by analogy, similar
to what is done in Benson's textbook,3 5 but often using quantum calculations on a small
molecule rather than experimental data as the basis of the analogy. Our experience is that this
approach is not completely reliable. Thus, additional quantum calculations are conducted to
improve the accuracy of rate parameters for reactions that are found to be important via
sensitivity analysis.
4.3.6 Quantum Calculations
Thermochemical and kinetic data were calculated at the CBS-QB3 level of theory using the
Gaussian 03 quantum chemistry package. 3 6 All stable compounds were calculated in their singlet
state, and radical compounds were calculated in their doublet states. Partition functions were
calculated using the CanTherm software package,3 7 using the recommended scaling factor of
0.99 for the frequency analysis. 38 One-dimensional hindered rotations were also included in the
analysis, using scans at the B3LYP/6-3 I IG(2d,p) level for each rotatable bond. Hindered rotor
scans were stepped in 10-degree increments, and all other coordinates were allowed to reoptimize at each step. The effective moment of inertia 1(2,3) for each hindered rotor was
calculated at the equilibrium geometry.3 9
Thermochemistry data for most species were estimated using Benson group additivity.4 ( An
extensive database has been generated for group additivity values for the estimation of sulfur
compounds, and these data are used along with other available estimation data in the RMG
database. 23,
33
However, quantum calculations for specific compounds generally provide more
-----------
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
77
accurate thermochemistry data than group additivity schemes, so these calculations have been
conducted for 30 of the reactants, proposed intermediates, and products of the hexyl sulfide
decomposition mechanism, in addition to data for compounds that were calculated in previous
work.
23 , 33
Thermochemical parameters were calculated in CanTherm. Enthalpy and entropy of formation
were calculated at 298 K, and heat capacities were calculated at 300, 400, 500, 600, 800, 1000,
1500, 2000, and 2400 K. Bond additivity corrections were applied to obtain the enthalpy values,
although no correction is available for the C=S double-bond, due to lack of experimental data.41
This topic will be explored further in Chapter 7.
Tunneling corrections were applied to the rate constant calculations using the asymmetric Eckart
method, which has been shown to provide accurate corrections for this type of chemistry.42 43
Rate constants were calculated in CanTherm using conventional transition state theory at 59
temperatures between 300 and 2000 K. These were fitted to the modified Arrhenius expression,
=
A-T"-exp
J
,
k(T)
R-T
where T is the temperature and R is the gas constant. The three coefficients to be fitted are the
Arrhenius constant A, the temperature factor n, and the activation energy E.
4.4
Results: Quantum Calculations
4.4.1 Water-Catalyzed Elimination of H 2 S
Kida et al. have recently discussed how water catalyzes the dehydration of germinal
mercaptoalcohols, reporting rate parameters for the reactions shown in Table 1.26 The presence
of the additional water molecule is calculated to increase the first-order decomposition rate of the
mercaptoalcohol by an order of magnitude, which can significantly affect product branching in
the resulting reaction mechanisms.
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
78
Table 1. Modified Arrhenius coefficients computed for the elimination of H 2 S to form acetaldehyde. A [s-' for
26
reaction 1, cm3 /(mol*s) for reaction 2], n (unitless), and Ea (kJ/mol). From Kida et al.
logA
OH
H
-s
O
HSH
[o'
OH
SH+ H 2 0
Ea
0-H
1.
2.
Rate Parameters
n
+ H 2S
0.13
12.62
k(673K) = 0.041 1/s
185.09
3.05
0.54
k(673K)*[H20] = 0.74 1/s
93.20
'o'H
-
+ H 2 S + H 20
Due to the comparative speed of the water-catalyzed reaction in Table 1, a brief study was
conducted to determine kinetic parameters for three additional reactions, which are presented in
Table 2. The first reaction in this table is the elimination of water from the mercaptoalcohol
compound to form a thioaldehyde. A slightly lower activation energy was calculated for this
reaction as compared with the H 2S elimination reaction in Table 1, but under the conditions of
this study they will occur at similar rates. Thus, we expect a significant amount of pentane
production from the water-catalyzed elimination of H 2 S from the mercaptohexanol. We expect
this channel to have some competition from the water elimination reaction to reform the
thioaldehyde.
In Table 2, reactions 4 and 5 are analogous to reactions 2 and 3, respectively, but catalyzed by
hydrogen sulfide instead of water. In both cases, the transition state is stabilized by the presence
of the additional H 2 S molecule to form a six-member ring. This stabilization effect is about 30
kJ/mol less than in the water-catalyzed version of both reactions. With the lower rate constants
and significantly lower concentration of H2 S in the reaction mixture in this study, the H 2 Scatalyzed reactions should be negligible compared with the water-catalyzed reactions. However,
this type of reaction could be important in situations where large amounts of H 2 S are being
produced, including pyrolysis conditions where water is not present. The analogous hexyl sulfide
reactions for those presented in the two tables have been added to the RIG database in the
direction presented, but the termolecular reverse reactions are also calculated by RMG using
thermodynamic consistency.
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
79
Table 2. Modified Arrhenius coefficients for three reactions catalyzed by H20 or H2 S. A [cm 3 /(mol*s)], n
(unitless), and E,, (kJ/mol).
Rate Parameters
n
logA
OHH
3.
SH
/ +H
2S+H2S -0.79 3.44 115.06 4.
---H+ H
20 1.08 2.59 86.02
20 + H
H
'
OH
S
'
H
a
+ H 20
Ea
SH+ H 2 S
H
SHH
OHH''
5.
H+ H 2S
H-S-',
-1
)'SH
-
H
S+H20+H2S0.32
2.94 119.75
S
4.4.2 Hydrogen Migration
Four intramolecular hydrogen abstraction (or hydrogen migration) reactions were chosen for this
work, based on possible relevance to the hexyl sulfide decomposition mechanism. Due to the
stabilization from the neighboring sulfur atom, the most stable product should be the hexyl
sulfide alpha-radical produced from hydrogen abstraction of the initial reactant. However, the
proposed decomposition mechanisms suggest that other hexyl sulfide radicals-particularly the
beta radical-could also be important intermediates in some of the pathways. Thus, rate
parameters were calculated for reactions with five- or six-membered cyclic transition states to
convert between different hexyl sulfide radicals.
Rate parameters for four hydrogen migration reactions are presented in Table 3, and the
transition state geometries can be seen in Figure 1. The ring size of the transition state had the
greatest effect on the rate constant, as reactions 7 and 9 had greater A-factors but lower n-factors,
as well as activation energies approximately 22 kJ/mol less than the similar reactions (6 and 8,
respectively) with one fewer carbon in the cyclic transition state. Comparing reactions with the
same number of atoms in the transition state ring, we see that the reaction rates at 400 'C will be
similar (within a factor of 10) for reactions 6 and 8, and likewise for reactions 7 and 9.
.
...
. ..
...
....
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
80
7.
6.
1. 6
3
1.
1.403
9
1. 181.
1.357
1.361
1,355'
8.
9.
1.
5
1.
1.365
1.65
1.344-
71376
Figure 1. Transition state geometries for the four hydrogen migration reactions. Distances (Angstroms).
Table 3. Calculated rate constants for hydrogen migration reactions. A (s-'), n (unitless), and E, (kJ/mol).
Rate Parameters
Reactions
6. -
SH
7.
lo
SH
8.
SS
9.
S
,
--
-
1W
SH
SH
logA
n
Ea
-4.75
4.50
49.91
-1.29
3.24
29.04
-2.94
3.95
46.73
-1.55
3.28
24.73
4.4.3 Radical Addition to Multiple Bond
Six radical addition reactions-beta scission reactions in the reverse direction-were considered
for this work. These proceed via the pathway shown in Figure 2, and the calculated modified
Arrhenius parameters are presented in Table 4. Reactions 10 through 14 were chosen as possible
consumption reactions for the thioaldehyde formed from hexyl sulfide decomposition. Radical
addition to the sulfur atom in the C-S double bond were not calculated in this work, as rate
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
81
parameters for many of these reactions have previously been calculated and added to the RMG
database.2 Reactions 15 through 17 are relevant to hexyl sulfide decomposition in the beta
scission direction, as they are possible final steps in the production of thiophenic compounds
from hexyl sulfide.
R3
R2
1
R4
Figure 2. A radical addition reaction.
Activation energies for the four reactions involving addition of a radical to the thiocarbonyl
group were all below 5 kJ/mol, and negative activation energies were fit to reactions 11 and 13
(although the overall rate constants exhibit the usual positive temperature dependence). This is
due to the instability of thiocarbonyl compounds, which are known to polymerize at room
temperature. 44 The energetics of reaction 14 reveal a submerged transition state, and a complex
was optimized near this geometry. A rate constant was estimated for this reaction using the
method discussed in the previous chapter.
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
82
3
Table 4. Calculated rate constants for radical addition and beta scission reactions. A [cm /(mol*s)I, n
(unitless), and E, (kJ/mol).
Reactions
Rate Parameters
logA
10.
As+ 14
11.
Xs
9.07
n
1.46
Ea
4.92
0.19
3.16
-6.73
3.24
2.50
2.26
1.58
2.82
-5.14
12.2
0.03
2.10
)W
7.40
1.76
9.89
W
1.99
3.25
24.33
4.06
2.56
11.09
Ag
13.
-
14. N-&
+ 'C 2 H5
No
+
V-
SH
S
S
15.
+ q
16.
+ 'C2 H 5
S
17./
'HS
+
+
+ C2 H 3
S
- OW
/
M
,
12.
&
SH
+ -MSH
Greater activation energies were calculated for the addition of a radical to a stable thiophenic
compound (reactions 15-17). Of more interest to this work is the reverse direction for reactions
15 and 16, which can be estimated using thermodynamic consistency. Beta scission of the ethyl
radical from the initial cyclic radical occurs via a significantly lower energy pathway than beta
scission of hydrogen. Thus, if beta scission of this radical were the primary method of generated
thiophenic compounds from hexyl sulfide, the production of thiophene over ethyl thiophene
would be expected, in disagreement with experimental data. However, disproportionation
reactions also promote the generation of ethyl thiophene, and this could help explain the
experimental results. The beta scission of the vinyl radical to form thiophene, the reverse of
reaction 17, has a high activation energy. This species would be more likely to undergo radicalmediated tautomerization to eventually form ethyl-thiophene, an experimentally observed
pyrolysis product of hexyl sulfide. This mechanism is discussed further below.
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
83
4.4.4 Cyclic Sulfide Formation
A cyclic sulfide formation reaction is a unimolecular radical addition, where a radical attacks the
sulfur atom of a thiocarbonyl group to form a cyclic sulfur compound, as shown in Figure 3.
Three of these reactions, depicted in table 5, were selected for this work, as possible intermediate
steps leading to the generation of the experimentally observed ethyl-tetrahydrothiophene and
ethyl-thiophene.
R2R2
-<>
-Ri
Figure 3. A cyclic sulfide formation reaction.
Modified Arrhenius parameters are presented in Table 5, and the transition state geometries are
presented in Figure 4. Reaction 18 is nearly barrierless in the forward direction, while reactions
19 and 20 have similar calculated activation energies. All three of these reactions are exothermic
in the cyclization direction. Reaction 20 has the greatest barrier in the reverse direction, as it
requires the breaking of a stable thiophenic ring to form an unstable thioketene.
Table 5. Calculated rate constants for cyclic sulfide formation reactions. A (s-1), n (unitless), and E, (kJ/mol).
Reactions
logA
Rate Parameters
Ea
n
3
6.65
1.17
1.15
*
11.90
0.10
44.78
-
S
S
18.
11.85
-0.14
45.71
0.
S
-
19.
20C-'
S
I f
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
ivia,
ty
19.
18.
84
2.374
_374
1.2
LA
20.
2.739
%1.
Figure 4. Transition state geometries for the three cyclic sulfide formation reactions. Distances (Angstroms).
4.4.5 Thermochemistry Calculations
In addition to thermochemical parameters calculated in previous work33 , parameters were
calculated here for 30 additional species. These include the reactants and products of the
reactions studied in the quantum chemistry work, as well as calculations that were used to
estimate thermochemical parameters for reactants, products, and likely intermediates for the
hexyl sulfide mechanisms. These parameters are further refined and discussed in Chapter 7.
4.5
Results: RMG Model Performance
4.5.1 Reaction Path Analysis
To investigate the effect of SCW on the high-temperature reactivity of hexyl sulfide,
mechanisms were generated by RMG in the presence and absence of water to simulate batch
reactor experiments that were previously conducted at 400 OC. 2 6 The SCW mechanism includes
273 species and 5971 reactions. Reaction path analysis of the RMG mechanisms suggests that
the formation of the major products in the SCW treatment of hexyl sulfide occurs via the reaction
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
85
pathways proposed by Kida et al., as presented in Figure 5.26 The reactant and observed products
are encased in boxes, and predicted products that are not observed experimentally are encased by
dashed boxes. This figure neglects reactions with fluxes below one percent for clarity.
The highest flux channel for the cracking of hexyl sulfide (species 1 in Figure 5), which is
responsible for 61% of the total consumption, starts with a hydrogen abstraction from the parent
molecule. The primary H-abstraction occurs from the alpha-carbon (carbon adjacent to the sulfur
atom), as the resulting alpha-radical is stabilized by its proximity to the sulfur atom. Small
amounts of gamma-, delta-, and epsilon-radicals are predicted, and these all are expected to
convert to the alpha-radical via intramolecular hydrogen migration reactions, but mainly react
back to form the initial reactant. The alpha-radical then undergoes a beta scission reaction (65%
flux) to form hexane (2) and hexanethial, the latter of which will be discussed in more detail.
The second major reaction pathway (27% flux) involves a hydrogen abstraction from hexyl
sulfide to form a beta radical, which quickly undergoes beta scission to form 1-hexene (4) and
hexanethiyl radical. Hexanethiyl radical further reacts to form hexanethiol (3). The 1 -hexene is
predicted to convert to 2- and 3-hexene (5 and 6, respectively) via hydrogen abstraction
reactions, with the majority predicted to exist as 2-hexene after 30 minutes.
Hexanethial produced by the beta scission of the alpha radical is predicted to be consumed by
two pathways. The main one (47%) is the pentane (5) and CO production pathway proposed by
Kida el al. (Path 1 in Figure 5). Hexanethiyl radical is then predicted to react with some CO to
form carbonyl sulfide (OCS), which would likely react further to form CO2.45 RMG also predicts
a variety of radical addition reactions, mainly attacking the carbon side of the C=S double-bond,
leading to the production of larger minor products, some of which continue to react. Additional
hexanethial consumption pathways are also predicted, but with much smaller fluxes. While none
of these products is observed experimentally, these pathways could be early steps in the
formation of large aromatic compounds, which may explain the dark color of the product
solution observed in some of the experiments. 2 6 The pathways will be discussed further in the
analysis of the anhydrous decomposition mechanism.
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
R
RH
C5H 13SH
3
R
C 6H 13s
.
24%
4
27
5
RH
CH1
S2%
RHCH1
R
R
R
R
S'CSH13
'RH
RH
24%
29%
C7H13
RH
86
CH
H13
2%
'61%
RH
C 6H 13
86%
2
C 5H
R
6
CGH 13
86%
R
(
1
%
H
R
S
7%
H2 0
47%
Y
S
OHH
C6 H13
CSH 11
SH
KR
6%
47%1
RH
H2S
SH
C 5H11
1CS
47%
C6H 13
6*
RH
2H2
RH
46
RH
6
Figure 5. Predicted reaction fluxes for the SCW desulfurization of hexyl sulfide at 400 *C
The mechanism for hexyl sulfide pyrolysis without water includes 231 species and 2773
reactions. The major reaction pathways for pyrolysis without water are largely similar to the
SCW case, and the total flux analysis for this mechanism is presented in Figure 6. The hydrolysis
of the thioaldehyde shown in Figure 5 is not possible in the absence of water, so there is greater
flux toward the diene pathway than in the SCW mechanism. In addition, a lengthy reaction
sequence to produce thiophenic compounds is predicted, and this mechanism is presented in
Figure 7.
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
87
In both models, dienes have been predicted as significant products although they are not
observed experimentally. This may correspond to the amount of coke-non-volatile species with
a high aromatic content-that was observed in both experiments. Although this content was not
quantified, it was clear that more coke was produced in the absence of water, due to the darker
color of the pyrolysis product and substantial loss of carbon that could be quantified by GC
analysis of gas- and liquid-phase products. 2 6 Thus, we propose that the production of dienes and
cyclic species from the thioaldehyde is an early step in the production of these large compounds.
Unfortunately, it is extremely difficult to track these species in detailed mechanisms, as there is
no single "coke molecule," but many different large species that will individually have very
small concentrations. Using the current version of RMG with available computational resources,
it is difficult to build a comprehensive mechanism including species with more than 30 heavy
atoms. Improving the ability of mechanism generation software to capture large-molecule
chemistry should be a topic of future research.
88
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
SH
RH
D
SH
1%10%
RH
6
1<1%
-10%
RH
R
R
R
3S
6H27%6C
1
3
RH
RH
RH
27%
27%
R
27%
4
R
%
5
S'C-6H 13
RH
RH
N,
S'C6H13
27R.
R~
RHR
~
C13
RH
*
RA RH
1%
C6H13
%C6H13
1
73%R
S C6)K13
C 6H 13
RH
1%
R'
C 6H 14
2
RH
RRH
71%
C6H13
71%
31%
SH
R
R*
20%
20%
R
11%
20%
SH
RH
31%
R'
i See fig. 15
SH
31 %H
RH
7
31%
28%
RH
28%
R
.
RH
Figure 6. Predicted reaction fluxes for the pyrolysis of hexyl sulfide at 400 *C in the absence of water
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
89
C6 H11
RH
R
SH
S SH
SH
RH
s
C 6H,1
RH
RH
S
S
S
RH
RH
-s
S
C 2H5
S
&
R
R
k' _)
R
yRH
RH
F
F
RH
A
RH
R
irS
Figure 7. Speculative reaction pathway for the formation of thiophenic compounds from the pyrolysis of
hexyl sulfide at 400 *C in the absence of water
4.5.2 Model validation for hexyl sulfide conversion in a CSTR
An additional mechanism was generated using RMG for the reaction mixture of hexyl sulfide,
hexadecane, and SCW, simulating the reaction conditions used by Patwardhan et al.2 ' The CSTR
mechanism was generated to be valid for temperatures between 400 and 450 'C, and it contains
140 species and 1957 reactions. The predicted reaction pathways for hexyl sulfide decomposition
are largely similar to the hexyl sulfide + SCW case, although the radical pool is different in this
case. C-S bond scission remains the dominant initiation reaction, but as hexadecane was
present in a much higher concentration than hexyl sulfide in the CSTR experiments, radicals
produced by the hexadecane solvent via bond-breaking or hydrogen abstraction exist in a higher
concentration.
The model predictions are compared with experimental data for hexyl sulfide conversion in
Figure 8. Overall, RMG predicts sulfide conversion accurately, with a slight underprediction at
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
90
the lowest temperature and a more significant overprediction at the highest. This may reflect the
uncertainty in the overall activation energy for hexyl sulfide decomposition predicted by RMG.
Our model is predicted for reactions that occur at a lower conversion (up to 25%), so it is likely
to be inaccurate at very high conversions. It is also important to remember the experimental
uncertainty in measuring the conversion of a relatively small amount of sulfide in a bulk of
hexadecane.
100
....................... ....................
80
-
-
400C, 500ppm
400C, 1000ppm
0-
60
-425C,
o5
---- 450C,
40
~Expt 400C, lOO0ppm
A
t
X
20
K
1000ppm
1000ppm
p4
Expt 400C, 500ppm
A
Expt 425C, 1000ppm
0
Expt 450C, 1000ppm
0
0
10
20
40
50
30
Residence Time (min)
60
70
Figure 8. Comparison of RMG model with CSTR data for hexyl sulfide conversion in SCW.
4.5.3 Model validation for product distributions in a batch reactor
SCW Mechanism
All of the major products are predicted reasonably accurately for SCW treatment, in comparison
with previously conducted batch reactor experiments.
6
Comparisons are presented in Figure 9
for hexyl sulfide, hexane, hexenes, hexanethiol, and pentane (note that the first 5 minutes, the
reactor heat-up time, are not presented). The conversion of hexyl sulfide and concentrations of
hexane and hexanethiol are predicted with excellent accuracy in comparison to experimental
data. The RMG model also captures the reactivity of hexanethiol in this case, although the
consumption rate of the thiol is overpredicted in comparison with experimental
data. The
concentration of this sulfur compound reaches a maximum at some point in the reaction
timescale, before converting to other products-this is discussed further in the discussion of
reaction pathways. In addition, pentane, a key product of the SCW experiments, is predicted
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
91
accurately by the RMG mechanism, suggesting that the SCW-hexanethial pathway is indeed
leading to the formation of this product.
1 00
0.9
0
0
S
75
0
0.6
50
0
0
0.3
25
C5H11
0
0
5
12
10
is
S
20
C 5H11
25
C6H14
5
30
10
Time (minutes)
15
20
25
30
Time (minutes)
0.6
0.4
C 6 H 13 SH
E
0.3
0.4
0.2
0
0.2
0
*
0.1
5
----10
,--15
0.0
-
0.0
20
25
5
30
10
Time (minutes)
15
20
Time (minutes)
25
30
0.6
0
C5H12 1
0.0
5
10
15
20
25
30
Time (minutes)
Figure 9. Comparison of RMG model with batch reactor data for hexyl sulfide in SCW at 400 *C. Full
mechanism (solid), experimental data (o)
Normalized sensitivity coefficients for hexyl sulfide concentration were calculated at a time of 6
minutes, and these are presented in Figure 10. At this time, the most sensitive reaction for hexyl
sulfide decomposition is the beta-scission of the alpha-radical to form hexanethial and hexyl
radical, which are intermediates in the main predicted product pathway to form pentane and
hexane, respectively. Hydrogen transfer reactions involving H2S and hexanethial are also
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
92
important to our overall prediction of the sulfide decomposition rate. Although they are involved
in minor pathways, our prediction is also sensitive to disproportionation reactions involving
hexanethial and subsequent intermediates. The rate estimates for these reactions have large
uncertainty in comparison with the other most sensitive reactions. However, it is important to
note that the other rate constants also have uncertainties of a factor of 2 or more, and these
uncertainties will propagate to a significantly greater uncertainty in the overall model
predictions. Decreasing this uncertainty through improved thermochemistry and rate estimates
will be a topic of further study.
-0.05
0
0.0.5
Ce.,1 1
'
C Hl<
S
'CrH 11
S + CbH,1
C 4H9
+
C5HII-
C5H 11 '
'S
s
C4H
4+
S
C5H 1
-
+
.0.1
SS
S
C,Hi
C4HR
S
SH
+ C4H9
H2S
+
3
CrH-1
SH
+
CH
SH
+
CSH
SH
*
S
H 2S
C 4H9
+
S
bHib
HS CH
SCH
C4HqC
C 4 H9 IC4H,
C4H9
5
4 H+
SHH
+
CsH11
S'CH
-
--
SH
C4H]
1
:
+
SH + C5H,1
S
S
CA
Figure 10. Normalized sensitivity coefficients for hexyl sulfide concentration, calculated at t=360 s
Non-SCW Mechanism
As discussed in the reaction path analysis section, the computer-generated non-SCW mechanism
predicted thiophene production as a major pathway, although this was not observed
experimentally. A speculative mechanism was also generated using the same input parameters
but disallowing the cyclicization reaction ultimately leading to the production of thiophene,
which is presented as the red arrow in Figure 7 (this was done by setting the activation energy of
this reaction arbitrarily high). The results of both simulations are presented in this section.
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
93
The predicted conversion of hexyl sulfide, and production of hexane, hexenes, and hexanethiol
are compared with batch reactor experiments in Figure 11. The pyrolysis model with the fast
channel to thiophene blocked shows excellent agreement with experimental data for hexyl
sulfide conversion, although the full mechanism disagrees with experimental conversion by
about a factor of two. This difference is primarily due to the disproportionation reaction
(presented in Figure 7) that forms ethyl-thiophene. This reaction is predicted to decrease the
overall radical pool significantly. Major products are predicted with greater accuracy by the
mechanism neglecting thiophene formation
--
100
X4-- - -------
In
0.9
za
C
.2
0
E
4,
=
(A
4,
25
C51-H1
5
10
is
20
0
C5HI1
S
0.3
C6H114
30
25
5
10
15
20
25
30
Time (minutes)
Time (minutes)
0.6
04
In
x
x
C6H 13 SH
0
X
0.3
.
0.4
0.2
/
I-
02
0
/
0.0
5
10
C 6H 12
15
20
25
Time (minutes)
30
/
0.1
Is
0
0.0
5
10
15
20
25
30
Time (minutes)
Figure 11. Comparison of RMG model with batch reactor data for hexyl sulfide pyrolysis at 400 *C. Full
mechanism (solid), mechanism blocking fast cyclization (dashed), experimental data (x)
4.5.4 Effect of Water Concentration
Additional batch reactor experiments were conducted for this work to investigate the dependence
of pentane production on the initial concentration of water, and mechanisms were generated
using RMG to model these experiments. The comparison of the model prediction and
experimental data is presented in Figure 12. We see that the RMG model significantly
underpredicts the production of pentane at low concentrations of water, and slightly overpredicts
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
94
it at high concentrations. This disagreement underscores the complexity of the chemistry
occurring in this system. Experiments below a water/sulfide ratio of 24 (the first 5 experimental
points in Figure 12) did not reach the critical pressure. These experiments were therefore
conducted in superheated steam. At these low-water conditions, reaction 1 is more important
than reaction 2. As the water loading increases in the supercritical range, the production of
pentane is predicted to increase but then level off at a pentane yield greater than what is observed
experimentally. This overprediction could be due to the absence of important competing
pathways for thioaldehyde consumption. This is expected to be a major source of the
experimentally observed "coke" (colored, nonvolatile material), which was produced in both the
water and no-water cases (but decreased when supercritical water was present). This molecularweight growth chemistry is extremely complex and currently very difficult to predict, but it will
be possible to explore these pathways more extensively as computational methods improve.
0.80
0.60
S0.400
0
0.00
0
20
40
60
80
100
Initial Moles Water/Hexyl Sulfide
Figure 12. Comparison of experimental data and RMG simulations for production of pentane in 30 minutes
in a 400 *C batch reactor, from different initial water/sulfide ratios. Model (solid,+), experimental data (o).
4.6
Conclusions
Reaction mechanisms have been generated using the automated Reaction Mechanism Generator
to model the reactivity of hexyl sulfide, with and without the presence of supercritical water.
Quantum chemistry calculations have been completed to improve the rate constant and
thermochemistry estimations used in mechanism generation. Calculations on the formation of
carbon monoxide from a thioaldehyde confirm the previous hypothesis that a water-catalyzed
pathway is likely to contribute to CO formation at supercritical conditions. Rate constants and
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
95
thermochemistry parameters have also been calculated for intramolecular hydrogen migration
reactions, as well as other reactions potentially relevant to the formation of thiophenic
compounds from sulfide pyrolysis.
CSTR models using the RMG mechanism provide excellent agreement with experimental data
for the decomposition of hexyl sulfide in the presence of hexadecane and SCW. Good agreement
is also observed between experimental measurements and model predictions for almost all of the
products generated in the batch reactor study. Sensitivity analysis for hexyl sulfide conversion
shows that reasonable rate constant estimations are employed for the most important reactions in
the simulation, but as always, more accurate and computationally expensive calculations could
possibly improve agreement with the data.
Sensitivity analysis is a useful tool for identifying areas of improvement in a kinetic mechanism,
but it alone is not sufficient in confirming the accuracy of a mechanism. A rate constant that is
misestimated by multiple orders of magnitude can be identified as sensitive, only to disappear
from this analysis when a more reasonable estimate is applied; more worryingly, a poorly
estimated (or missing) reaction may not appear in the sensitivity analysis at all even though it is
actually quite important. Improvements in the rate estimation method, via improvements in the
RMG algorithm or automated transition state searches and rate calculation methods, will provide
more useful mechanisms on the first attempt and allow one to focus on refining the reactions that
are truly important to the model predictions.
4.7
References
1.
S. Brarendregt, P. J. M. Valkenburg, E. S. Wagner, M. Dente and E. Ranzi, in 14th
Ethylene Producers'Confirence, AIChE, New Orleans, LA, 2002, pp. 2497-2537.
F. Battin-Leclerc, J. M. Simmie and E. Blurock, eds., Cleaner Combustion: Developing
DetailedChemical Kinetic Models, Springer, London, 2013.
Y. Shi, H.-W. Ge and R. D. Reitz, Computational Optimization of Internal Combustion
Engines, Springer, London, 2011.
Rethinking the Ozone Problem in Urban and Regional Air Pollution, National Academy
of Press, Washington, DC, 1991.
W. H. Green, in Advances in Chemical Engineering, ed. G. B. Marin, Academic Press,
vol. 32, pp. 1-50.
2.
3.
4.
5.
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
96
A. S. Tomlin, T. Turinyi and M. J. Pilling, in Comprehensive Chemical Kinetics, ed. M.
J. Pilling, Elsevier, 1997, vol. 35, pp. 293-437.
W. H. Green, J. W. Allen, R. W. Ashcraft, G. J. Beran, C. A. Class, C. Gao, C. F.
Goldsmith, M. R. Harper, A. Jalan, G. R. Magoon, D. M. Matheu, S. S. Merchant, J. D.
Mo, S. Petway, S. Raman, S. Sharma, K. M. Van Geem, J. Song, J. Wen, R. H. West, A.
Wong, H.-W. Wong, P. E. Yelvington and J. Yu, Reaction Mechanism Generator
(RMG), v. 4.0.1 (2013). http://rmg.sourceforge.net.
J. W. Allen, C. F. Goldsmith and W. H. Green, Phys Chem Chem Phys, 2012, 14, 11311155.
A. Jalan, R. H. West and W. H. Green, J. Phys. Chem. B, 2013, 117, 2955-2970.
R. Hua, Y. Li, W. Liu, J. Zheng, H. Wei, J. Wang, X. Lu, H. Kong and G. Xu, J.
Chromatogr. A, 2003, 1019, 101-109.
J. Beens and R. Thijssen, J. High Resolut. Chromatogr., 1997, 20, 131-137.
A. Kruse and E. Dinjus, Journalof SupercriticalFluids, 2007, 39, 362-380.
N. Akiya and P. E. Savage, Chemical Reviews, 2002, 102, 2725-2750.
A. R. Katritzky, R. A. Barcock, M. Balasubramanian, J. V. Greenhill, M. Siskin and W.
N. Olmstead, Energy & Fuels, 1994, 8.
0. M. Ogunsola and N. Berkowitz, Fuel, 1995, 74, 1485.
T. Adschiri, R. Shibata, T. Sato, M. Watanabe and K. Arai, Ind Eng. Chem. Res., 1998,
37, 2634-2638.
B. M. Vogelaar, M. Makkee and J. A. Moulijn, Fuel Processing Technology, 1999, 61,
265-277.
A. R. Katritzky, M. Balasubramanian and M. Siskin, Energy & Fuels, 1992, 6, 431-438.
T. Sato, T. Adschiri, K. Arai, G. L. Rempel and F. T. T. Ng, Fuel, 2003, 82, 123 1-1239.
M. Watanabe, S. Kato, S. Ishizeki, H. HInomata and R. L. Smith, Jr., J. Supercrit. Fluids,
2010, 52, 48-52.
A. G. Vandeputte. PhD dissertation, University of Ghent, 2012.
A. G. Vandeputte, M. K. Sabbe, M.-F. Reyniers and G. B. Marin, Phys Chem Chem
Phys, 2012, 14, 12773-12793.
A. G. Vandeputte, M. K. Sabbe, M.-F. Reyniers and G. B. Marin, Chemistry-A European
Journal, 2011, 17, 7656-7673.
A. G. Vandeputte, C. A. Class, M.-F. Reyniers, W. H. Green and G. B. Marin, Submitted,
2014.
P. R. Patwardhan, M. T. Timko, C. A. Class, R. E. Bonomi, Y. Kida, H. H. Hernandez, J.
W. Tester and W. H. Green, Energy & Fuels, 2013, 27, 6108-6117.
Y. Kida, C. A. Class, A. J. Concepcion, M. T. Timko and W. H. Green, Phys Chem Chem
Phys, 2014, 16, 9220-9228.
CHEMKIN-PRO 10131, (2013) Reaction Design, San Diego.
P. A. Marrone. PhD Thesis, Massachusetts Institute of Technology, 1998.
J. Song, S. Raman, J. Yu, C. D. Wijaya, G. Stephanopoulos and W. H. Green, Abstr. Pap.
Am. Chem. Soc., 2003, 226, U530-U531.
J. Song, R. Sumathi, J. Yu and W. H. Green, Abstr. Pap. Am. Chem. Soc., 2004, 228,
U233.
K. M. Van Geem, M.-F. Reyniers, G. B. Marin, J. Song, W. H. Green and D. M. Matheu,
AIChE J., 2006, 52, 718-730.
Chapter 4: Modeling the desulfurization of hexyl sulfide by supercritical water
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
97
M. R. Harper, K. M. Van Geem, S. P. Pyl, G. B. Marin and W. H. Green, Combust.
Flame, 2011, 158, 16-41.
C. A. Class, J. Aguilera-Iparraguirre and W. H. Green, Submitted, 2014.
C. F. Goldsmith, G. R. Magoon and W. H. Green, Journal of Physical Chemistry A,
2012, 116, 9033-9057.
S. W. Benson, The Foundationsof Chemical Kinetics, McGraw-Hill, New York, 1960.
M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman,
J. Montgomery, J. A., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar,
J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H.
Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T.
Nakajima, Y. Honda, 0. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J.
B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, 0. Yazyev,
A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A.
Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C.
Strain, 0. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V.
Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A.
Liashenko, P. Piskorz, 1. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham,
C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M.
W. Wong, C. Gonzalez and J. A. Pople, Gaussian 03, (2004) Gaussian, Inc., Wallingford
CT.
S. Sharma, M. R. Harper and W. H. Green, CanTherm open-source software package,
(2010). http://github.com/GreenGroup/CanTherm
A. P. Scott and L. Radom, Journalof.Physical Chemistry, 1996, 100, 16502-16513.
A. L. L. East and L. Radom, Journalof ChemicalPhysics, 1997, 106, 6655-6674.
S. W. Benson, F. R. Cruickshank, D. M. Golden, G. R. Haugen, H. E. O'Neal, A. S.
Rodgers, R. Shaw and R. Walsh, Chemical Reviews, 1969, 69, 279-324.
G. A. Petersson, D. K. Malick, W. G. Wilson, J. W. Ochterski, J. A. Montgomery and M.
J. Frisch, Journalof Chemical Physics, 1998, 109, 10570-10579.
A. G. Vandeputte, M. K. Sabbe, M.-F. Reyniers, V. Van Speybroeck, M. Waroquier and
G. B. Marin, J. Phys. Chem. A, 2007, 111, 11771-11786.
C. Eckart, Physical Review, 1930, 35, 1303-1309.
N. J. Cooper, Compr. Org. Funct. Group Transform. H, 2005, 3, 355-396.
C. Deng, Q.-G. Li, Y. Ren, N.-B. Wong, S.-Y. Chu and H.-J. Zhu, Journal of
Computational Chemistry, 2007, 29, 466-480.
Chapter 5: Modeling the decomposition of alkylaromatic compounds
98
Chapter 5: Modeling the decomposition of
alkylaromatic compounds
5.1 Abstract
The Reaction Mechanism Generator (RMG) is used to model the dealkylation of aromatic
compounds in pyrolysis and supercritical water (SCW) treatment. Experiments on whole crude
oil have shown that heavy hydrocarbons are cracked to form lighter compounds in SCW
treatment, increasing the value of the product. Model compound studies were used to investigate
the mechanism for this upgrading process, and this chapter presents our efforts to explain the
results using ab initio methods and automated mechanism generation. Thermochemical
calculations were able to explain the differing reaction rates of 2-hexylthiophene,
3-
hexylthiophene, and hexylbenzene. RMG was then used to build detailed reaction mechanisms
for the decomposition of hexylbenzene. The key result was that SCW does not have any direct
impact on the decomposition of this model compound, as effectively identical product
distributions were observed experimentally and predicted by RMG with and without water. This
result suggests that SCW is not directly reacting with these compounds in crude oil, but
facilitating upgrading by preventing other reactions-such as addition to a thioaldehyde,
discussed in Chapter 4-that lead to the formation of large compounds.
Chapter 5: Modeling the decomposition of alkylaromatic compounds
5.2
99
Introduction
A key finding in the experimental work on the SCW desulfurization of hexyl sulfide was the
formation of heavy compounds in the anhydrous pyrolysis experiments, evidenced by the darker
product liquid and lack of mole balance closure. The SCW experiment provided a clearer
product with better mole balance, suggesting not only that intermediate sulfur products are a
possible cause of coking in the high-temperature chemistry of crude oils, but also that SCW
could prevent this coking.' We were able to explore some of the mechanistic reasons for that in
the previous chapter. Experimentation on heavy crude oil provided a similar result, that SCW
treatment resulted in lighter and therefore more valuable product distributions than simple
pyrolysis.2 In particular, this was exhibited by the cleavage of alkyl chains connected to aromatic
rings, producing a much smaller alkylaromatic compound. Model compound experiments and
detailed mechanistic studies would allow us to investigate this process on a fundamental level.
In this chapter, a model for hexylbenzene decomposition is generated using RMG. Ab initio
calculations allow us to explain the relative reaction rates of three alkylarenes thermochemically,
suggesting that
the decomposition
occurs via free-radical
mechanism.
The
detailed
decomposition mechanisms for hexylbenzene with and without SCW are generated and then
validated with experimental data, agreeing within the uncertainty of the calculations. The
pyrolysis case can then be compared with the SCW decomposition to help explain the role of
SCW in the overall crude oil upgrading process.
5.3
Methods
5.3.1 Reaction Simulation
The Reaction Mechanism Generator (RMG) was used to build kinetic mechanisms for this
work.' The RMG algorithm has been discussed extensively in past work,4 5 and only a brief
introduction will be provided here. The key feature is a flux-based algorithm for model
generation, pursuing reaction pathways in directions with greater flux while omitting those with
lower predictions. RMG
produces a file containing NASA
polynomials to estimate
thermochemical parameters for each of the species in the mechanism, as well as the modified
Chapter 5: Modeling the decomposition of alkylaromatic compounds
100
Arrhenius parameters for each reaction. This file can be imported into CHEMKIN-PRO to
simulate the kinetic experiments, and conduct flux and sensitivity analyses. 6
For this work, mechanism generation and simulation conditions were chosen to match the
experimental conditions as closely as possible. Mechanisms were generated in RMG using the
experimental temperature, pressure, and starting concentrations (one with and one without
SCW). All reactions were assumed to be in the high-pressure limit. Reactor simulations were
conducted using the "Closed Homogeneous Batch Reactor" model in CHEMKIN-PRO,
including a 1 0-minute linear heat up time to reach the experimental temperature of 450 'C.
5.3.2 Quantum Calculations
Previously calculated Arrhenius and thermochemical are available in literature for the
decomposition mechanisms of alkylaromatic 7 compounds, and some of these were added to the
RMG database to assist in parameter estimation for this work. In addition, the RMG database is
able to estimate reasonably accurate rate parameters for many other relevant reactions using
previously calculated data. However, when generating mechanisms containing thousands of
reactions, it is not possible to find extremely accurate rate parameters for each of the predicted
reactions. Thus, some of them must be roughly estimated, and these estimations can be uncertain
by multiple orders of magnitude. These uncertainties are mitigated by calculating more accurate
rate constants using ab initio techniques. These calculations can still be uncertain by up to a
factor of ten, so it is important to consider these uncertainties when analyzing these reaction
mechanisms.
In this work, some reactions were identified as having a significant effect on the rate of
phenyldodecane decomposition, and the initially estimated parameters were improved using
quantum mechanics and transition state theory. Gaussian 038 was used to determine the
geometries and vibrational frequencies of stable molecules and reaction transition states at the
B3LYP/6-31 1G(2d,p) level of theory, and single point energies were calculated using CBS-QB3.
Parameters for particularly important reactions, including the retroene reaction depicted in Figure
1, were refined using CBS-QB3 9' 10 and CCSD(T)-F12/cc-pVDZ-F12"" single point energies.
Calculations using the second, coupled-cluster method were conducted using Molpro.1 4
Chapter 5: Modeling the decomposition of alkylaromatic compounds
The open-source CanTherm
5
101
software package was used to calculate rate constants and
thermochemical parameters using transition state theory. A scaling factor of 0.99 was used for
the frequency analysis. One-dimensional hindered rotations were also included in this analysis,
using scans at the B3LYP/6-3 1 G(2d,p) level of theory in 10 degree increments. The effective
moment of inertia
1(2,3)
was calculated for each hindered rotor. Modified Arrhenius constants
were then derived for each of these reactions, and these parameters were added to the RMG
database to improve model generation.
ph
[Ph
1401
H
1
+
Ph
1.927
2.323
Figure 1. Proposed pathway (top) and optimized transition state (bottom) for the retroene reaction to
produce toluene and 1-pentene. Interatomic distances (X).
5.4
Results and Discussion
5.4.1 Quantum Calculations
While similar product distributions were obtained in experiments on the pyrolysis of 2hexylthiophene, 3-hexylthiophene, and hexylbenzene, the rate of decomposition was measured to
be significantly different. 2 We theorized that this was due to the stability of the radicals arising
from each compound, which can be illustrated by resonance in Figure 2. Thermochemistry
calculations were conducted at the CBS-QB3 level of theory, for the analogous propylarenes
instead of hexylarenes, to save computational time. The calculated enthalpies, presented in Table
, show that the stability order of the three compounds is 3-propylthiophene > propylbenzene >
Chapter 5: Modeling the decomposition of alkylaromatic compounds
102
2-propylthiophene. This agrees qualitatively with the decomposition rates for the analogous
hexylarenes.
Rj
R
C Nr-r
U-
R*
S
S
R
Figure 2. Dealkylation of three alkylarenes, and resonance structures of the resulting radicals.
Table 1. Calculated enthalpies of reaction for C-C bond breaking in alkylarenes, and experimentally
observed conversions2 in pyrolysis at 450 *C for 30 min.
Reaction
AH rxn
(kcal/mol)
Observed
Conversion (%)
77.2
93
74.9
9
R
R
R
S/1
R
R
70
78.1
ZR
5.4.2 RMG Model Performance
The reaction mechanism generated to model the pyrolysis experiment contains 279 species and
583 reactions, while the SCW model contains 275 species and 578 reactions. The major reaction
pathways predicted for the two cases are effectively identical, in agreement with the
experimental results. The main reaction steps, including fluxes, are presented in Figure 11.
Hexylbenzene pyrolysis is predicted to proceed mainly via normal cracking, in accordance with
previous modeling research on alkyl aromatics.7 The "retroene reaction," the molecular reaction
forming pentene and toluene from the reactants, is predicted to provide appreciable toluene
production, although only about 10% of the overall decomposition was via this pathway. This is
in agreement with previous studies, which predicted 80% of toluene production via this
molecular pathway at 330 'C, but only 20% at 400
oC.
6
,'17
This reaction, as well as the free-
Chapter 5: Modeling the decomposition of alkylaromatic compounds
103
radical toluene production pathway, also account for the foriation of 1-pentene, which then can
go on to react further via decomposition or addition reactions.
Styrene is predicted as a major product via beta-scission pathway. Experimental results from this
and previous work7 '
18
suggest that this compound is an intermediate in the formation of
ethylbenzene, although our experiments at 450 'C show some styrene remaining as a product
after 40 minutes. RMG predicts that the reverse disproportionation pathway leads to some
production of ethylbenzene, but this reaction is less dominant than at lower temperatures.
A comparison of experimental vs. model hexylbenzene conversion is plotted in Figure 12, and
the comparison for the main aromatic products is presented in Figure 13. Excellent agreement is
observed for the reactant conversion, suggesting that the rate constants and thermochemistry for
the main reaction steps are reasonably accurate. Toluene is overpredicted, but well within the
model's uncertainty. The sum of ethylbenzene and styrene production predicted by RMG also
agrees with what was observed experimentally, which also suggests that the first few reaction
steps are predicted well. However, ethylbenzene is underpredicted by the model while styrene is
significantly
overpredicted,
likely
due to two
factors.
First,
the expected
reverse
disproportionation pathway, which seems to be supported by the experimental results, has been
predicted to be slower than necessary to account for the observed amount of ethylbenzene. This
disagreement is likely due to the uncertainty in the calculation of rate constants for
disproportionation reactions, which is about a factor of 10. Second, RIG predicts a number of
large molecules (two or more benzene rings) to be produced from these experiments, and each of
these compounds is predicted in very small amounts. The low concentration of each of these
species is reasonable, as there are many different possible addition and recombination products;
no single large compound is likely to be produced in great quantity. On the other hand, we
propose that many different large molecules are produced in these experiments, each in rather
small quantities. This hypothesis is supported by the small peaks observed on the heavier end of
the GC chromatographs, as well as the lack of benzene balance closure, which is also observed
for hexyl thiophene. 2
104
Chapter 5: Modeling the decomposition of alkylaromatic compounds
20/At
4%
--,C2H
7
7%
28%
10%
3%
~
C 3H6
-
C5HIe
42%
6%
649 23%
2%
C4H
C 4H
40%
CH 3 +o CH 4
8
A
2%
5%
2%
Figure 3. Reaction fluxes for the pyrolysis of hexylbenzene at 450 *C.
0
75
-
100
40
C
50
25
-
a
-
45%
I~~~
0
0
10
20
30
I
40
Time (minutes)
Figure 4. Hexylbenzene conversion vs. time at 450 *C. (can add more plots if needed)
(x) pyrolysis experiment, (o) SCW experiment, (-) pyrolysis model, (--) SCW model.
7%
0.6
-
--
105
-
Chapter 5: Modeling the decomposition of alkylaromatic compounds
0.
-
'X
E
X
ra
0.2
0.0
0
10
20
30
40
lime (minutes)
Figure 5. Model (solid lines) comparison with experiments (pyrolysis x, SCW o) for production of toluene
(blue), styrene (black), and ethylbenzene (red). Aromatic molar yield is defined here as fraction of total
aromatic compounds present as the given product. Only the pyrolysis model is presented here
5.5
Conclusions
In this chapter, a reaction mechanism was presented for the decomposition of hexylbenzene at
450 'C. This compound was predicted to react primarily via normal cracking, with some amount
of the conversion occurring via the molecular retroene reaction to form 1 -pentene and toluene
directly from the reactant. The predictions for major products agreed reasonably well with the
experimental results, but an extended reaction mechanism would likely generate better
predictions of minor products.
In the RMG mechanism and experimental product distribution, no change was predicted with the
addition of SCW. Thus, it is probable that water does not react directly with an alkylbenzene or
its intermediates during treatment, but with other reactive species that would lead to the
formation of larger compounds. One example of this would be the reactive thioaldehyde
discussed in Chapter 4, which leads to the formation of thiophenes in pyrolysis but lighter
compounds in SCW treatment. However, there are thousands of other species in these fuels, and
additional investigations may uncover additional benefits of this relatively simple process.
Chapter 5: Modeling the decomposition of alkylaromatic compounds
106
5.6
References
1.
Y. Kida, C. A. Class, A. J. Concepcion, M. T. Timko and W. H. Green, Phys Chem Chem
Phys, 2014, 16, 9220-9228.
Y. Kida, A. G. Carr and W. H. Green, Energy & Fuels, 2014, 28, 6589-6595.
W. H. Green, J. W. Allen, R. W. Ashcraft, G. J. Beran, C. A. Class, C. Gao, C. F.
Goldsmith, M. R. Harper, A. Jalan, G. R. Magoon, D. M. Matheu, S. S. Merchant, J. D.
Mo, S. Petway, S. Raman, S. Sharma, K. M. Van Geem, J. Song, J. Wen, R. H. West, A.
Wong, H.-W. Wong, P. E. Yelvington and J. Yu, Reaction Mechanism Generator
(RMG), (2013).
J. W. Allen, C. F. Goldsmith and W. H. Green, Phys Chem Chem Phys, 2012, 14, 11311155.
M. R. Harper, K. M. Van Geem, S. P. Pyl, G. B. Marin and W. H. Green, Combust.
Flame, 2011, 158, 16-41.
CHEMKIN-Pro, (2011) Reaction Design Inc., San Diego.
V. Burkle'-Vitzthum, R. Michels, G. Scacchi and P.-M. Marquaire, Ind. Eng. Chem. Res.,
2003, 42, 5791-5808.
M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman,
J. Montgomery, J. A., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar,
J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H.
Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T.
Nakajima, Y. Honda, 0. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J.
B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, 0. Yazyev,
A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A.
Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C.
Strain, 0. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V.
Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A.
Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham,
C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M.
W. Wong, C. Gonzalez and J. A. Pople, Gaussian 03, (2004) Gaussian, Inc., Wallingford
CT.
J. A. Montgomery Jr., M. J. Frisch, J. W. Ochterski and G. A. Petersson, Journal of
ChemicalPhysics, 1998, 110, 2822-2827.
J. A. Montgomery Jr., M. J. Frisch, J. W. Ochterski and G. A. Petersson, Journal of
ChemicalPhysics, 2000, 112, 6532-6542.
J. Aguilera-Iparraguirre, A. D. Boese, W. Klopper and B. Ruscic, Chemical Physics,
2008, 346, 56-68.
J. Aguilera-Iparraguirre, H. J. Curran, W. Klopper and J. M. Simmie, JournalofPhysical
Chemistry A, 2008, 112, 7047-7054.
W. Klopper, R. A. Bachorz, D. P. Tew, J. Aguilera-Iparraguirre, Y. Carissan and C.
Hattig, JournalofPhysical ChemistryA, 2009, 113, 11679-11684.
H.-J. Werner, P. J. Knowles, G. Knizia, F. R. Manby and M. Schitz, Wiley
InterdisciplinaryReviews: ComputationalMolecular Science, 2011, 2, 242-253.
S. Sharma, M. R. Harper and W. H. Green, CanTherm open-source sofiware package,
(2010).
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
Chapter 5: Modeling the decomposition of alkylaromatic compounds
17.
18.
V. Burkkd-Vitzthum, R. Michels, G. Scacchi and P.-M. Marquaire, Industrial
Engineering Chemistry Research, 2003, 42, 5791-5808.
P. E. Savage and M. T. Klein, Industrial & Engineering Chemistry Research, 1987, 26,
374-376.
P. E. Savage and M. T. Klein, Ind. Eng. Chem. Res., 1987, 26, 374-376.
&
16.
107
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
108
Chapter 6: Modeling the effect of sulfur on
phenyldodecane decomposition
6.1
Abstract
The automated Reaction Mechanism Generator (RMG) is used to simulate the decomposition of
phenyldodecane in the presence of diethyl disulfide. Recent experiments have shown that the
presence of diethyl disulfide has a statistically insignificant effect on conversion of the
alkylaromatic over 72 hours at 350 'C, suggesting that increased concentration of this radicalproducing compound might not have the effect that was proposed by previous researchers.
Detailed kinetic modeling allows us to investigate these results mechanistically, by modeling
each elementary step of the process. Mechanisms were first built to model the individual
pyrolysis mechanisms for phenyldodecane and diethyl disulfide. Both mechanisms were
validated with the available experimental data. Diethyl disulfide is predicted to react within
minutes, forming mostly ethane and hydrogen sulfide, as well as methane and carbon dioxide in
the presence of water. An RMG mechanism including both compounds was then generated, and
reactor simulation showed that accelerated phenyldodecane conversion only occurred during the
intial minutes of the experiment, while disulfide conversion was in progress.
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
6.2
109
Introduction
Fossil fuels are produced by the decomposition of plant and animal remains deep beneath the
surface of the earth. In particular, kerogen is formed by the decomposition of algae, plankton,
and woody plants. 1 This heavy, waxy material can decompose further to form petroleum and
natural gas, the former of which is the basis for one of the world's most important industries.
Unfortunately, despite advances in geochemistry and engineering, petroleum exploration remains
a highly uncertain field, as the success of multibillion dollar drilling decisions is largely
governed by chance. Improved knowledge of the chemical processes linking kerogen to oil to
gas would be one significant step to better utilization of the planet's energy resources.
Currently, decisions in oil exploration are assisted by the predictions of lumped kinetic models,
which consist of simple mechanisms linking proposed reactants to products of interest. Rate
parameters in these mechanisms are fit to match data from experiments conducted at conditions
drastically different from geological conditions. This is necessary because useful experimental
results at 200 'C could take millions of years to achieve, while the decomposition of similar
compounds at 350 'C can be observed in hours. However, the extrapolation of parameters
collected at these elevated temperatures to those of interest geologically introduces massive
uncertainty into the model, as the relevant chemistry can differ greatly at 200 and 350 'C.
Detailed reaction mechanisms, which model a process of interest as a series of elementary
reaction steps, have the potential to provide more relevant models for use in conditions relevant
in the kerogen maturation process. The open-source Reaction Mechanism Generator2 (RMG)
allows a user to automatically generate a reaction mechanism for hydrocarbons containing
oxygen, nitrogen, or sulfur. This software is able to propose new reactions that fit templates for
known reaction types, and then estimate the necessary reaction rate constants using available
experimental or ab initio data. The resulting mechanisms are completely free from bulk fitting to
experiments, but new experimental data are extremely useful in the validation of these models.
For this work, phenyldodecane was chosen as a representative model compound in crude oil, and
diethyl disulfide was used as the sulfur contributor. The resulting high-temperature mechanisms
were compared with experimental work to better understand the effects of sulfur at
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
Ito
experimentally relevant conditions, and an additional reaction mechanism was generated at
geologically relevant conditions to investigate whether sulfur had the same effect.
6.3
Methods
6.3.1 Reaction Simulation
RMG 2 has previously been used to model a variety of processes relevant in fuel chemistry,
including the combustion and pyrolysis of various candidate biofuels. 3 ,
4
RMG builds these
mechanisms using a rate-based algorithm, automatically proposing any possible reaction (fitting
within the templates in the RMG database) for a given set of reactants. Arrhenius and
thermochemical parameters are estimated for every reaction, and these data are used to estimate
the rate of each reaction in the model. These rate estimations allow us to continue to follow
reaction pathways that are predicted to proceed more quickly.
For our initial investigation, we have neglected solvent effects. A reaction mechanism was first
built for phenyldodecane conversion at 350 'C and 80 bar for a reaction time of 72 hours, and a
separate mechanism was built for diethyl disulfide pyrolysis to 'ensure that the major pathways
for each compound were followed. These phenyldodecane and diethyl disulfide mechanisms
were combined, and then RMG was employed to predict the cross-reactions between the species
contained in the two separate mechanisms, as well as the additional products of these reactions.
Mechanisms were also generated for the reaction in the presence of water. For these runs, a "seed
mechanism" was included to predict the formation of CO2 from the hydration of thioaldehyde
compounds, which has previously been demonstrated to be important in the desulfurization of
sulfides by supercritical water. 5 After mechanism generation was complete, reactor simulations
were conducted using the "Closed Homogeneous Batch Reactor" model in CHEMKIN-PRO. 6
6.3.2 Quantum Calculations
Previously calculated Arrhenius and thermochemical are available in literature for the
decomposition reactions of organosulfur 740 and alkylaromatic 1 compounds, and these were
available for model generation in this work. However, the RMG database still does not have
accurate data available for 100% of the thousands or millions of reactions that could possibly be
involved in oil-to-gas mechanism generation. Thus, an estimation algorithm is implemented to
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
III
provide as accurate parameters as possible for a given reaction using the data available to RMG,
even when specific rate parameters for a reaction are not available. Depending on the data
available, these parameters can still be uncertain by multiple orders of magnitude. Reactions that
are found to be important for overall mechanism accuracy can be refined using ab initio
methods.
Quantum calculations were conducted using the Gaussian 03 software package,
using
optimizations and frequencies computed at the B3LYP/6-31 1G(2d,p) level of theory. Single
point energies were calculated using CBS-QB3 for all reactions, and CCSD(T)-F12/cc-pVTZF12 for reactions found to be particularly important to the mechanism predictions. This second
method has been found to provide more accurate energies, 13-15 and these coupled-cluster
calculations were conducted in Molpro.1 6 The CanTherm software package' 7 was then used to
calculate reaction rate constants using transition state theory, including treatment of onedimensional hindered rotations and Eckart tunneling.' 8
6.4
Results and Discussion
6.4.1 Diethyl Disulfide Decomposition Mechanism
To better understand the role of sulfur linkages in hydrocarbon cracking, it is first necessary to
identify the important reactions in the decomposition mechanisms of these sulfur compounds. To
this end, RMG was first used to build a small reaction mechanism for diethyl disulfide
decomposition, at conditions applicable to the experiments done by Lewan 19 and in this work,
but in the absence of phenyldodecane. Work has previously been conducted on the pyrolysis of
dimethyl disulfide,
0
and many of the reaction rate constants and thermochemical parameters
from that work are applicable to the current study. These data, as well as other calculated data for
the reaction rates and thermochemical parameters of sulfur compounds,7- 9 were used to generate
a pyrolysis mechanism for diethyl disulfide that is valid between 200 and 350 'C, but with
phenyldodecane replaced by inert argon. The mechanism generation was set to terminate when
60% conversion of the disulfide was achieved, and the core tolerance was set at 10% of the
characteristic flux.
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
112
The RMG mechanism contains 101 species and 3261 reactions, and the major reaction pathways
at a reaction time of 60 seconds are presented in Figure 1. Scission of the S-S bond is an
important radical generator within the first seconds of the simulation, and C-S bond scission
becomes more important after this initiation phase (due to recombination reactions decreasing
the net flux of the S-S scission pathway). At 60 seconds, 90% of the reacting disulfide is
predicted to be consumed by a hydrogen abstraction, followed by a beta scission reaction to
produce the ethanethiyl radical and thioacetaldehyde. Much of the ethanethiyl quickly
decomposes to ethylene and a mercapto radical, which abstracts a hydrogen atom to form
hydrogen sulfide. Thioacetaldehyde, which contains a C-S double-bond, is not as easily
desulfurized, and it mostly participates in addition reactions to form stable intermediates,
including a cyclic compound containing two sulfur atoms. These intermediates are predicted to
react over a longer timescale to form more H 2 S, as well as stable thiophene.
The conversion of diethyl disulfide over time is plotted for 200 to 350 'C in Figure 2. At the
temperature studied experimentally, the reactant is 50% consumed within the first 15 seconds.
This half-life increases to seven minutes, 11 hours, and 10 months at 300, 250, and 200 'C,
respectively. H2 S generation occurs somewhat more slowly than reactant conversion, as
presented in Figure 3. At temperatures above 300 'C, very fast H 2 S generation is predicted at the
initial stages of the reaction time, with significantly slower generation over the remainder of the
72 hours. H 2 S is likely to behave as a chain-transfer agent and have some effect on hydrocarbon
cracking products, but the presence of this compound itself is not likely to have a great effect on
the total concentration of radical species in the reaction mixture, as the S-H bond is
approximately as stable as a C-C bond. Finally, the concentration of various radicals formed in
pyrolysis is presented in Figure 4. Reactive ethylthiyl (C 21 5S) and thioformylmethyl (C 2 H3 S)
radicals reach a maximum within the first second of the experimental timescale and then
decrease for the remaining 72 hours, while relatively stable S 2 (a diradical) and HS 2, which have
low hydrogen affinity, are formed later on.
HS
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
5%
0-
2H 5
+
S
+
<1%
113
-
RH
5%R
1*R'
H
RH
90%
4%
RH
4%
R'
SH:
+
C
HS
S
R'
90%
)*RH
4%
RH
R'
77%~
SH77
77%
:C2 H4 |
RH
20%R
R'
S
42%
42%
RH
-
+
zS
42%
S
SH
S
|- SH
SRH
77%
.
4%
4%
RH
42%
RF
S
IS R'
-SH
4
RH
4%
R
Figure 1. Decomposition pathway of neat DEDS at a temperature of 350 *C and a time of 60 seconds.
Percentages are presented relative to total decomposition rate of DEDS at this time. Products in dashed boxes
are predicted to be consumed through subsequent reactions, while ethane and hydrogen sulfide are predicted
to be stable products.
114
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
100
100
IA
0
wj
0
C
75
75
LI
.-
-
--
0
50
50
25
25
Of
0
...
.
....
0
0
48
24
100
0
0.001
72
0.1
0.01
1
10
100
Time (Hours)
Time (hours)
Figure 2. Predicted conversion of diethyl disulfide vs. time. 350 *C (solid), 300 *C (dashed), 250 *C, (double),
and 200 *C (dotted)
1.00
Ifl
0
LU
0 0.75-ii
4..
0.50I
0.250.00
0
24
72
48
Time (hours)
Figure 3. Predicted production of H 2 S from diethyl disulfide pyrolysis. 350 *C (solid), 300 *C (dashed), 250
*C, (double), and 200 *C (dotted)
1.E-0 2
-
-
1.E+u U
1.E-0 4
0
1.E-0
1.E-0
8
i.E-1
i.E-1 4
1.E-07
1.E-)4
1.E-Ol
-
1.E+02
-
.E- 2
i.E+i-5
Time (s)
Figure 4. Concentration vs. time for representative radicals formed by DEDS pyrolysis at 350 *C. S2 diradical
(black solid), HS 2 radical (black dashed), thioformylmethyl radical (gray solid), ethylthiyl radical (gray
dashed)
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
115
Experimental data for disulfide pyrolysis are scarce, but the data collected by Lewan and in a
previous study of dimethyl disulfide pyrolysis provide evidence that our predicted mechanism is
reasonable. This model predicts approximately 90% conversion of diethyl disulfide in two hours
at 590 K and 10 kPa, which is in good agreement with the 85% conversion observed
experimentally by Coope & Bryce for dimethyl disulfide pyrolysis.2 1 This is evidence that the
main reaction rate constants have been estimated with reasonable accuracy; therefore, we can
continue explore the effect of diethyl disulfide on hydrocarbon cracking by modeling the
pyrolysis of a mixture of this compound and phenyldodecane.
To explore the role of water in disulfide decomposition, an additional reaction mechanism was
generated to simulate the gold bag experiments conducted with phenyldodecane, diethyl
disulfide, and water, but with phenyldodecane replaced by inert argon to simplify the initial
simulations. Using the same termination criteria as the pyrolysis case, the resulting mechanism
contains 248 species and 26,641 reactions. This is more than double the size of the pyrolysis
mechanism in the absence of water, mainly due to the inclusion of the seed mechanism.
The predicted decomposition pathway for diethyl disulfide at 350 'C and a time of 60 seconds is
presented in Figure 4. The primary reaction steps are largely the same as in the predictions
without water, with the most significant difference in the decomposition of the thioaldehyde
intermediate. At 60 seconds, the thioaldehyde consumption pathway is similar to the case
without water, but the hydration pathway will take priority over the full timescale. The hydration
of this compound leads to the production of a mercapto-alohol, which then loses hydrogen
sulfide to form acetaldehyde. This pathway can be catalyzed by the presence of additional water
molecules.5 The aldehyde can then decompose to form carbon monoxide, which can then react to
the experimentally observed carbon dioxide either directly via water gas shift, or by a free radical
pathway that proceeds with a carbonyl sulfide (OCS) intermediate. The production of carbon
dioxide from carbonyl sulfide has previously been studied,2 2 and can also be catalyzed in the
presence of water.
Overall, the generated mechanism predicts that approximately one mole each of ethane, methane,
and other gas-phase carbon (in the form of carbon dioxide or carbonyl sulfide) to be formed from
each mole of diethyl disulfide reacted. This agrees with experimental results, with only slight
disagreement in the amount of carbon dioxide vs. carbonyl sulfide predicted, as no carbonyl
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
116
sulfide was observed experimentally. This is likely due to modeling uncertainties, as the water
gas shift reaction kinetics were roughly estimated, and rate constants in the carbonyl sulfide to
-~s*
+
<1%
-~s*
9%
C2H5
-S-8
+
carbon dioxide pathway have uncertainties of approximately a factor of two.
RH
RH
9%
6%
RH
87%
R'
R'
3%
C 2H6
-'S'
SH
-
86%
H 20
+
2%
5
2
4%
53%
R-
C 2H 4 '
RH
r
SH
~7
SRH
35%
R'
20%
38%
RH
35%5
."SH
+
-3SH
H 2S
OH
S --- 4\s
2%
--- ----S
<1%
H2 0
<1 O
<1%
RH
2%
R'
H2*-
if
0 2H 5
-1% d+ 2%
OH 3
---
F
+__
RH
OH
H20
<1% <1{
R*r
I
OH
C 2 H6
O
SH
<1%
002 +
HR 2S
Figure 5. DEDS decomposition pathway at a temperature of 350 *C and a time of 60 seconds, in an equimolar
mixture with water. Percentages are presented relative to total decomposition rate of DEDS at this time.
Products in dashed boxes are predicted to be consumed later in the reaction timescale, while ethane and
hydrogen sulfide are predicted to be stable products.
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
117
6.4.2 Phenyldodecane Decomposition Mechanism
The kinetics of phenyldodecane (PDD) pyrolysis in the absence of any other additional reactants
has been previously studied,2 3'
24
and thermodynamic data for liquid PDD at elevated
temperatures and pressures are also available, allowing for a thorough interpretation of the
behavior of this oil surrogate at experimental and reservoir conditions with additional reactants
present. The short and long chain products of PDD cracking are also more easily recoverable
than shorter chain alkyl substituted aromatics.
A kinetic study of the similar phenyldecane has been previously conducted, including
experimental and modeling work." The mechanism generated was largely similar to our initial
RMG-predicted mechanism for phenyldodecane, and it included a calculated rate constant for the
key reverse disproportionation reaction connecting styrene to ethylbenzene formation. In
addition, rate parameters for a molecular elimination reaction to form toluene and an alkene (the
retroene reaction presented in Figure 5) have been calculated prior to this work. Thus, the
parameters for these reactions were added to RMG to generate an improved mechanism, and this
improved mechanism was used as a starting point for investigating the effect of diethyl disulfide
on conversion.
H
H
Ph
R
I
Ph
R
Ph- +
R
Figure 6. Retroene reaction of an alkyl benzene to form toluene and an alkene
An RMG mechanism containing 365 species and 15,755 reactions was generated to model the
experiment of Savage & Klein at 400 *C, 23 and the comparisons between model and
experimental data for the reactant and major products are presented in Figure 6. Reasonable
agreement with the experimental data can be seen in this figure, as the rate of PDD pyrolysis is
predicted within a factor of two.
Major product predictions from the RMG model have been compared with experimental data,
and these comparisons along with RMG-predicted reaction mechanisms for these products are
presented in Figure 6. The model qualitatively captures the concentration profiles of undecene
and toluene. In both the experiment and model, undecene and toluene are produced at the same
118
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
rate for early time points, as they arise from the same reaction mechanism (shown in Figure 6).
As the experiment progresses, the mole fraction of toluene continues to increase while the mole
fraction of undecene stabilizes (and decreases slightly in the experiment). RMG predicts
undecene to be consumed primarily by radical addition reactions, forming larger hydrocarbon
species.
1
X
0.75
Xx
0.5
IL
X
0
-
PDD Data
PDD Model
0.25
0
240
180
120
60
0
Time (min)
0.3
x
.2 0.2
-Undecene
U-
Model
Toluene Model
.9
A
Undecene Data
x Toluene Data
&A
A
0
X
0
80
60
40
20
% Conversion PDD
q
Ph.-MC
9
H1 9
RH
C1H22
RH
Ph
CH19
4
Ph-
f
K
2 -1400Ph-
Figure 7. Comparison plot (top) and predicted pathway (bottom) for production of undecene and toluene
from neat PDD pyrolysis at 400 *C. Experimental data from Savage & Klein.
Figure 7 shows the RMG predictions of three major products-decane, styrene, and
ethylbenzene-produced by another pathway from PDD. This proceeds by abstraction of
119
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
hydrogen neighboring the phenyl group, which then undergoes a beta-scission reaction to form
the expected cracking products, decane and styrene. Experimental data suggests that the styrene
proceeds to form ethylbenzene. This was automatically predicted by RMG: styrene can abstract a
hydrogen atom from PDD in a reverse disproportionation reaction to form two stable radical
species, leading to the formation of stable ethylbenzene. This pathway is underpredicted by the
RMG mechanism, but ethylbenzene is still predicted within an order of magnitude of the
experimental data.
0.09
C6
. 0.06
Decane Data
X
0
X
Styrene Data
Ethylbenzene Data
X
0.03
Decane Model
-
1X
-Styrene
Model
Ethylbenzene Model
0
0
80
60
40
20
% Conversion PDD
RH
q
ClOHH
5
PhM-
9C
H 19
ClOH22
RH
RH
Ph"" > C9H19
Ph-
q
RH
q
Ph-Ph
styrene, and
Figure 8. Comparison plot (top) and predicted pathway (bottom) for production of decane,
ethylbenzene from PDD pyrolysis at 400 *C. Experimental data from Savage & Klein.
at
Comparison with the product distributions collected experimentally by Reeves after treatment
350 'C for 72 hours can be seen in Figure 8. Good agreement is observed for all major products
expected
except for undecane, which is significantly underpredicted by RMG. There are two
to form
pathways for formation of this product. The first is bond scission of the initial radical
undecane.
benzyl and undecyl radicals, the latter of which would abstract hydrogen to form
Undecene produced in the major toluene production pathway in Figure 6 could also abstract
for these
hydrogen in a reverse disproportionation reaction to form undecyl radical. Parameters
reactions are reasonably estimated by RMG, but these could be investigated further using more
accurate ab initio methods to improve model performance.
120
Chapter 6: Modeling the effect pf sulfur on phenyldodecane decomposition
1000
100
10-
%
0
-6
E
0.1
0.01
Figure 9. Comparison of experimentally observed (dashed, x) and RMG predicted (solid, A) products from
pyrolysis of neat phenyldodecane at 350 *C for 72 hours.
6.4.3 Phenyldodecane Decomposition in the Presence of Diethyl Disulfide
Pyrolysis mechanisms containing 95 and 70 species for phenyldodecane and diethyl disulfide,
respectively, were found to provide similar product predictions as the full mechanisms analyzed
previously, so these were used as seed mechanisms to build the full mechanism for the
decomposition of phenyldodecane in the presence of diethyl disulfide, as studied experimentally
by Lewan 19 and by Reeves. RMG was used to determine the additional hydrogen abstraction,
addition, and substitution reactions that are possible between species in each mechanism, and
model expansion continued using a core tolerance of 0.5. This mechanism was built to be
applicable
for the conditions and concentrations investigated by Lewan,19 from pure
phenyldodecane to a 10:1 mixture (by mole) of phenyldodecane and water. The final mechanism
includes 155 species and 2840 reactions. It was also built to be valid at temperatures between
200 and 350 *C, so that the resulting mechanism could be used to make predictions at geological
conditions.
The decomposition of phenyldodecane is predicted to occur by largely the same mechanism in
the presence and absence of the disulfide, via the free radical and molecular pathways that
mainly produce toluene, styrene, ethylbenzene, decane, and undedecane. However, as seen in
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
121
Figure 8, very little acceleration is predicted in the phenyldodecane decomposition rate when the
concentration of disulfide is increased from two to eight percent in the initial reaction mixture,
although some increase is predicted when increasing from no sulfur to two percent. The small
predicted effect in decomposition rate could be due to the fast decomposition of diethyl disulfide,
which was discussed previously, producing a large amount of relatively stable hydrogen sulfide
(and possibly some very stable thiophene compounds). These products are not likely to impact
the phenyldodecane decomposition rate after they are produced, although they might affect
product selectivities.
100
0
0
00
5
o0
U
0
0
0.02
0.04
0.06
0.08
0.1
Initial Mole Fraction DEDS
&
Figure 10. Comparison of RMG predictions with experimental data for the conversion of phenyldodecane
after 72 hours at 350 *C, in the presence of varying amounts of diethyl disulfide (RMG solid line
rectangles, Lewan experiment white circles, Reeves experiment yellow circles). Error bars on Reeves data
indicate 95% confidence intervals.
The same reaction mechanism was used to simulate conditions relevant to the geological oil-togas process. The predicted effect of diethyl disulfide on phenyldodecane conversion at 200 'C is
presented in Figure 10. Nearly the same sulfur effect is predicted at this temperature, as the
disulfide conversion and H 2S production are largely complete within the first one percent of the
overall reaction time. An increase in phenyldodecane conversion is predicted within this short
period, which is illustrated in Figure 12, as the concentration of reactive radicals is greater. After
the majority of disulfide has reacted to form stable products, phenyldodecane is predicted to
react at same rate in all four simulations. Thus, the presence of other sulfides that produce
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
122
radicals more slowly (such as sulfides and thiols) may provide a more sustained increase in the
radical concentration of the mixture.
-
100
0
0
75
-
(A
-
50
0
0
a. 25
'
0-
0
0.02
0.04
0.06
0.08
0.1
initial Mole Fraction DEDS
Figure 11. RMG predictions for the conversion of phenyldodecane after 1000 years at 200 *C, in the presence
of varying amounts of diethyl disulfide.
1001
1
75
50
-
.2
C
0
-
25
0
I
0
5000
10000
15000
20000
Time (years)
Figure 12. Predicted conversion of phenyldodecane (single lines) and diethyl disulfide (double line) vs. time at
9
200 *C with four different initial disulfide concentrations. Initial disulfide loading as described in Lewan' : 0
(black solid), 0.982 (black dotted), 1.767 (gray solid), 3.370 (gray dotted). Predicted disulfide conversion is
nearly identical for the different runs, so only 0.982 case is presented.
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
6.5
123
Conclusions
The automated Reaction Mechanism Generator has been used to model the cracking of a model
oil component, phenyldodecane, in the presence and absence of diethyl disulfide. Individual
decomposition mechanisms were generated for each of the two components, and these
mechanisms were validated with experimental data before modeling the pyrolysis of a mixture of
the two compounds. RMG predicts that diethyl disulfide reacts to form ethane, hydrogen sulfide,
and some thiophene in the absence of water; in the presence of water, single-carbon species are
predicted in addition to ethane and hydrogen sulfide. Much of this process occurs within the first
minutes of the 72 hour experiment, accelerating the conversion of phenyldodecane for this short
time due to the increased radical presence. However, as products of disulfide pyrolysis are
formed, the radicals recombine with other small radicals and large phenyldodecane-derived
radicals. After this time, much of the sulfur is present as H 2S, which does not have a significant
effect on phenyldodecane conversion.
6.6
References
1.
H. H. Schobert, in Chemistry of Fossil Fuels and Biofuels, Cambridge University Press,
New York, 2013.
W. H. Green, J. W. Allen, R. W. Ashcraft, G. J. Beran, C. A. Class, C. Gao, C. F.
Goldsmith, M. R. Harper, A. Jalan, G. R. Magoon, D. M. Matheu, S. S. Merchant, J. D.
Mo, S. Petway, S. Raman, S. Sharma, K. M. Van Geem, J. Song, J. Wen, R. H. West, A.
Wong, H.-W. Wong, P. E. Yelvington and J. Yu, Reaction Mechanism Generator
(RMG), (2013).
J. W. Allen, A. M. Scheer, C. W. Gao, S. S. Merchant, S. S. Vasu, 0. Welz, J. D. Savee,
D. L. Osborn, C. Lee, S. Vranckx, Z. Wang, F. Qi, R. X. Fernandes, W. H. Green, M. Z.
Hadi and C. A. Taatjes, Combust. Flame, 2014, 161, 711-724.
M. R. Harper, K. M. Van Geem, S. P. Pyl, G. B. Marin and W. H. Green, Combust.
Flame, 2011, 158, 16-41.
Y. Kida, C. A. Class, A. J. Concepcion, M. T. Timko and W. H. Green, Phys Chem
Chem Phys, 2014, 16, 9220-9228.
CHEMKIN-PRO 10131, (2013) Reaction Design, San Diego.
A. G. Vandeputte, M.-F. Reyniers and G. B. Marin, ChemPhysChem, 2013, 14, 17031722.
A. G. Vandeputte, M.-F. Reyniers and G. B. Marin, ChemPhysChem, 2013, 14, 37513771.
A. G. Vandeputte, M. K. Sabbe, M.-F. Reyniers and G. B. Marin, Phys Chem Chem
Phys, 2012, 14, 12773-12793.
2.
3.
4.
5.
6.
7.
8.
9.
Chapter 6: Modeling the effect of sulfur on phenyldodecane decomposition
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
124
A. G. Vandeputte, M. K. Sabbe, M.-F. Reyniers and G. B. Marin, Chemistry-A European
Journal, 2011, 17, 7656-7673.
V. Burkle'-Vitzthum, R. Michels, G. Scacchi and P.-M. Marquaire, Ind. Eng. Chem.
Res., 2003, 42, 5791-5808.
M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman,
J. Montgomery, J. A., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar,
J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H.
Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T.
Nakajima, Y. Honda, 0. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J.
B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, 0. Yazyev,
A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A.
Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C.
Strain, 0. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V.
Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A.
Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham,
C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M.
W. Wong, C. Gonzalez and J. A. Pople, Gaussian 03, (2004) Gaussian, Inc., Wallingford
CT.
J. Aguilera-Iparraguirre, A. D. Boese, W. Klopper and B. Ruscic, Chemical Physics,
2008, 346, 56-68.
J. Aguilera-Iparraguirre, H. J. Curran, W. Klopper and J. M. Simmie, Journal of Physical
Chemistry A, 2008, 112, 7047-7054.
W. Klopper, R. A. Bachorz, D. P. Tew, J. Aguilera-Iparraguirre, Y. Carissan and C.
Hattig, Journal of Physical Chemistry A, 2009, 113, 11679-11684.
H.-J. Werner, P. J. Knowles, G. Knizia, F. R. Manby and M. Schutz, Wiley
Interdisciplinary Reviews: Computational Molecular Science, 2011, 2, 242-253.
S. Sharma, M. R. Harper and W. H. Green, CanTherm open-source software package,
(2010).
C. Eckart, Physical Review, 1930, 35, 1303-1309.
M. D. Lewan, Nature, 1998, 391, 164-166.
A. G. Vandeputte, M.-F. Reyniers and G. B. Marin, J. Phys. Chem. A, 2010, 114, 1053 110549.
J. A. R. Coope and W. A. Bryce, Can. J. Chem., 1954, 32, 768-779.
C. Deng, Q.-G. Li, Y. Ren, N.-B. Wong, S.-Y. Chu and H.-J. Zhu, Journal of
Computational Chemistry, 2007, 29, 466-480.
P. E. Savage and M. T. Klein, Ind. Eng. Chem. Res., 1987, 26, 374-376.
F. Behar, F. Lorant, H. Budzinski and E. Desavis, Energy & Fuels, 2002, 16, 831-841.
Chapter 7: A thermochemical database for organosulfur compounds
125
Chapter 7: A thermochemical database for
organosulfur compounds
7.1
Abstract
A database of calculated gas-phase thermochemical parameters has been built for organic
compounds containing sulfur. This includes sulfides, thiols, and thiophenes, as well as some
oxidized sulfur compounds. Energy calculations were conducted using the high-accuracy
CCSD(T)-F12a/cc-pVXZ-F12 method, where double- (X=D) and triple-zeta (X=T) basis sets
were used depending on molecule size. A training set was chosen containing compounds for
which reasonably accurate experimental enthalpy measurements were available. Calculated
enthalpies were compared with these experimental values to regress a set of bond additivity
corrections, which are then applied to the calculated enthalpies of the remaining compounds.
Entropies and heat capacities were also calculated and validated against the available
experimental data. The resulting parameter database can be used to generate new group
additivity estimation schemes for studying the pyrolysis and oxidation of organosulfur
compounds.
...................
Chapter 7: A thermochemical database for organosulfur compounds
7.2
.126
Introduction
Thermochemical parameters are critical to our ability to model reacting systems accurately.
Gibbs free energies are required to calculate the equilibrium constant of a reaction, which is used
to estimate the rate of a reverse reaction when forward rate parameters are available. Heat
capacities are necessary to capture temperature changes in nonisothermal systems.
Enthalpies, entropies, and heat capacities are generally determined most accurately via
experiments, but this is not always feasible (particularly for unstable compounds). Because of
this, quantum calculations have been a reasonable way to generate thermochemical data for a
large number of species in a relatively small amount of time. Recent advancements, including
the use of coupled-cluster calculations with F 12 corrections, have greatly improved the accuracy
of the parameters estimated, particularly the enthalpies. When compared with accurate
experimental data, such as the Active Thermochemical Tables (ATcT), 2, a systematic error is
observed.
This error generally scales linearly with molecule size, and it suggests that some
This has frequently been accomplished
correction can be added to decrease the uncertainty.
using either atomic or bond additivity corrections (AAC's and BAC's, respectively), both of
which have been shown to substantially decrease deviations between experiments and quantum
calculations.
3-6
Substantial data are available for the estimation of thermochemical parameters for species
containing carbon, hydrogen, and oxygen, 4'
6-8
and these have been used extensively in model
generation for combustion and pyrolysis chemistry. 9' 10
This database was subsequently
expanded to include organosulfur compounds 3 and some organosulfur compounds with
oxygen." However, many of these early calculations were conducted using DFT methods that
generally provided uncertainties greater than 1 kcal/mol for enthalpies of formation, even after
additivity corrections were applied. The uncertainties for the parameter estimations of sulfur
compounds are even greater, as accurate experimental data for these species are rare.
Advances in coupled-cluster calculation methods12 '
13
have made it possible to calculate single-
point energies for molecules to much greater accuracy with reasonable computational cost.
These have been validated for a variety of kinetic and thermochemical systems,14-1 6 and the
implementation
of additivity corrections improves
the accuracy
further for enthalpy
Chapter 7: A thermochemical database for organosulfur compounds
127
calculations.1 7 Application of this technique to sulfur compounds, with additional experimental
enthalpies for use as a training set, will allow one to calculate accurate thermochemical
parameters for sulfur compounds in a reasonable amount of time.
However, when building mechanisms considering millions of possible species, it is still not'
reasonable to calculate thermochemical data for each compound using these methods. Methods
for automatic quantum calculations without user intervention have made it feasible to perform
these calculations for a greater number of species, but group additivity methods 7'
8
remain the
best bet for estimating these parameters in the shortest possible time. Expansive databases of
group additivity values (GAV's) are available for C/H/O and C/H/S compounds, but there is
relatively little available for C/H/O/S species.
These are particularly important when studying
the oxidation of sulfur compounds, which occurs in combustion and oxidative desulfurization
mechanisms
18-20
In this work, a database of enthalpies, entropies, and heat capacities for organosulfur compounds
is presented. A set of 60 compounds for which reasonably accurate enthalpies are available are
selected from literature, and calculations are conducted for these molecules using CCSD(T)-F 12
with the cc-pVDZ-F12 and cc-pVTZ-F12 basis sets to determine bond additivity corrections
(BAC's).
Thermochemical parameters are calculated for an additional set of molecules to
generate a database of accurate calculations that can be used to regress GAV's and cyclic
corrections, as well as hydrogen bond increments (HBI's) for radical species.
7.3
Methods
Optimizations and frequency calculations were conducted using density functional theory in
Gaussian 032. Molecules were optimized in their singlet state, with the exception of S2, which
was found to be at its ground state as a triplet. Radical species were calculated as doublets. The
CBSB7 basis set has been found to provide accurate geometries and force constants, so
B3LYP/CBSB7 was used for most molecules in this work. Optimizations using CCSD(T)/63 l+G(d') were also performed when the converged CBSB7 geometry was in doubt.
22
Single-point energies were calculated using the Molpro software package . Based on the work
by Buesser et al,17 where CCSD(T)-F12a/cc-pVDZ-F12 and CCSD(T)-F12a/cc-pVTZ-F12 were
Chapter 7: A thermochemical database for organosulfur compounds
128
shown to provide highly accurate enthalpies of formation when BAC's were applied, these two
basis sets were chosen for our work to achieve a balance between accuracy and computation
time. Quadruple-zeta basis sets provide even greater accuracy, and these should be considered
when the resources are available.
Partition functions and thermochemical parameters were calculated using the open-source
CanTherm software package23.
A scaling factor of 0.99 was used for the frequency analysis.
One-dimensional hindered rotations were also included, using scans in 10 degree increments at
the B3LYP/6-31 lG(2d,p) level of theory. The effective moment of inertia
](2,3)
was calculated
for each rotor at the equilibrium geometry. In the final analysis, BAC's were applied to estimate
the enthalpy of formation, and the entropy of formation and heat capacities between 300 and
2000 K were also calculated.
7.3.1 Regression of Bond Additivity Corrections (BAC's)
BAC's are applied to minimize the error, c, between the calculated and experimental enthalpy of
formation:
AH2 98 ,expt = A H2 9 8,calc +
nIc
+
4
where ci is the correction for bond type i (such as a C=C double-bond) and ni is the number of
that bond type present in the molecule.
Experimental thermochemistry data for sulfur
compounds were obtained from the NIST WebBook2 4 . Because these data were obtained by a
large variety of research groups using different experimental methods, reported uncertainties
vary widely among the dataset. As such, the enthalpies reported in the ATcT tables are known to
much greater uncertainty, so the BAC's calculated for C/H/O compounds using these data are
likely to have less uncertainty than those regressed using the available data for sulfur
compounds. For this work, we will hold the BAC's constant at the previously regressed values
for C/H/O compounds, and new BAC's will only be regressed for the bonds including a sulfur
atom. 60 acyclic species with reported uncertainties below 0.7 kcal/mol were chosen as the
training set for BAC regression. Three additional species containing a C=S double-bond were
also included in this analysis, even though they have uncertainties greater than 0.7 kcal/mol. This
Chapter 7: A thermochemical database for organosulfur compounds
129
was done due to the lack of experimental data for this type of compound, but it is important to
remember the uncertainty introduced due to this decision.
7.4
Results and Discussion
7.4.1 BAC Regression and Validation
60 acyclic species for which gas-phase enthalpy measurements have been reported with
uncertainties below 0.70 kcal/mol were selected for the training set from the NIST Chemistry
WebBook.2 4 The cumulative density function (CDF) for the deviations of each calculation from
the experiment is presented in Figure 1. The mean absolute deviation (MAD) for the original
calculations is 1.62 kcal/mol, with an average deviation of 1.41 kcal/mol. This suggests a
systematic overprediction of the enthalpy of formation, which should be partially resolved with
the implementation of BAC's.
Corrections for the S-H, C-S, C=S, S-S, 0-S, and O=S bonds were derived using leastsquares regression, decreasing the mean absolute deviation of the training set enthalpies from
1.62 to 0.72 kcal/mol. 80% of the species in the training set agreed within I kcal/mol of the
experimental data, but compounds containing O=S double-bonds were found to have the greatest
deviation from experiments, as shown in Figure 2. These deviations were on both ends of the
error range, suggesting that a simple O=S correction would not correctly treat the calculation
errors for sulfates, sulfites, sulfones, and sulfoxides. Thus, a modified bond additivity correction
scheme (MBAC) was proposed, including an additional O=S=O correction to replace the O=S
correction for species where the sulfur atom is double-bonded to two oxygen atoms. This
additional fitting parameter decreases the mean absolute deviation further to 0.55 kcal/mol,
providing the CDF in Figure 3. Note that while the highlighted compounds are much closer to
zero error than in Figure 2, they still represent many of the compounds with the greatest
disagreement with experiment. Thus, further work is necessary in accurately modeling the
thermochemistry of these compounds.
Chapter 7: A thermochemical database for organosulfur compounds
00
00
-
1
130
0.75
Q 0.5
0.25
0
-4
4
2
0
-2
Error (Calc - Expt) [kcal/mol]
6
Figure 1. CDF for AHf*(298 K) deviations between experiment and calculation, not including BAC's.
1
a P0
0.75
0.5
0.25
0 0
0
-4
00
2
0
-2
Error (Calc - Expt) [kcal/mol]
4
Figure 2. CDF for AHf*(298 K) deviations between experiment and calculation, including standard BAC
treatment. Gray: sulfoxides and sulfites, Black: sulfones and sulfates, White: others.
131
Chapter 7: A thermochemical database for organosulfur compounds
10
V
0.75
Q 0.5
0.25
0
-4
2
0
-2
Error (Calc - Expt) [kcal/mol]
4
Figure 3. CDF for AHr*(298 K) deviations between experiment and calculation, including MBAC treatment.
Gray: sulfoxides and sulfites, Black: sulfones and sulfates, White: others.
The derived BAC's are presented in Table I for the double- and triple-zeta cc-pVXZ-F12 basis
sets. The fitting statistics for the training set are presented in Table 2, and enthalpies and
entropies
of compounds
thermochemistry
modeled in the
of additional
training set are presented
in Table
species for which experimental thermochemical
3. The
data were
available, but were not included in the training set for MBAC regression, are presented in Table
4. Sulfides, disulfides, thiols, and cyclic sulfides (including thiophenes) are modeled with
excellent accuracy with the MBAC scheme. In addition, entropies are reproduced well by these
calculations, deviating by an average of 0.15 cal/(mol*K).
Much larger errors are observed for enthalpy calculations of cyclic sulfones, as well as vinyl
sulfide and the most Sn cyclic species. These calculations roughly agreed with those that had
been reported previously in CBS-QB3 calculations,; so it is possible that errors in the geometry
optimization are the cause of the deviations in enthalpy calculations. Thus, care should be taken
when building models using thermochemistry estimations for these types of molecules.
7.4.2 Thermochemical Database
The same calculations were calculated for an additional set of organosulfur compounds, for
which no accurate experimental data were available. These compounds were selected for their
potential relevance in the pyrolysis, SCW desulfurization, and oxidation of organosulfur
Chapter 7: A thermochemical database for organosulfur compounds
132
compounds. The MBAC's regressed in the previous section were applied to the initially
calculated enthalpies of formation, and these refined values are available at rmg.mit.edu.
Standard entropies and heat capacities between 300 and 2000 K are also provided. The full
database of 282 species is provided in the appendix of this chapter.
Table 1. Regressed MBAC's (kcal/mol) regressed to model organosulfur thermochemistry. C-H, C-C, and
C-O bonds were held fixed to previously regressed BAC from ATcT data.
Regressed MBAC's
C-H
VDZ
-0.49
C-C
S-H
C-S
C=S
S-S
C-O
0-S
O=S
O=S=0
-0.71
0.87
0.42
0.51
0.86
-0.29
0.23
-0.53
1.95
VTZ
-0.10
-0.35
0.52
0.13
-0.12
0.30
0.06
0.15
-2.61
0.27
Table 2. Fitting statistics for CCSD(T)-F12/cc-pVXZ-F12 enthalpy calculations (kcal/mol), where X=D or T.
Method
Mean Absolute Deviation
Average Deviation
Root Mean Square Deviation
Number of Molecules
Number of Fitting Parameters
VDZ
4.73
4.47
5.63
59
0
VDZ-MBAC
0.66
0.02
0.89
59
7
VTZ
1.62
1.41
2.13
59
0
VTZ-MBAC
0.55
0.02
0.77
59
7
Chapter 7: A thermochemical database for organosulfur compounds
133
Table 3. Comparison of calculated (triple-zeta) and experimental enthalpies of formation and standard
entropies for 59 gas-phase species selected for training set.
Af1 0 -98 [kcal/mol]
34.0
MBAC
34.5
Experiment
46.76
This Work
45.9
-5.6
-4.6
49.18
67.3
68.6
28.4
28.7
-5.7
67.2
68.4
49.1
50.3
18.2
-15.5
17.3
-9.1
-11.3
-16.3
50.32
56.38
56.45
57.6
60.55
65.91
0.15
-17.8
-14.41 i 0.26
-25.99
Experiment
SH
H 2S
-4.9
67
68
28.3
29.88
-5.46
:C=s
HC'=S
H 2C=S
H 3 C-S'
H 3 C-SH
0.1
2
2
HS
0.48
0.14
16.41 2
-8.96 i 0.48
-11.03
-16.39 t 0.15
HS
-18.39
S
S
HS
S
HS-
HS
HS
S' 298 [cal/(mol K)]
No BAC
28.1
28.5
-5.3
56.4
55.1
57.1
61.1
65.4
69.68
68.8
69.63
70.9
80.1
-18.6
77.9
78.0
-13.5
-14.4
79.71
79.8
0.21
-23.6
-25.0
83.68
82.8
-22.97
0.19
-21.9
-23.3
87.3
-23.06
0.21
-21.9
-23.2
87.1
-18.8
-20.3
88.0
-20
0.55
-8.8
-11.1
S
86.6
-21.43
0.18
-20.2
-21.7
HS
-26.49
0.42
-23.9
-25.8
98.3
HS-k
-30.33
0.22
-30.9
93.1
HS
-28.91
0.23
-27.5
-29.3
83.9
-27.4
0.28
-25.7
-27.6
95.7
0.23
-25.4
-27.3
96.1
HS
-30.76 i 0.21
-29.8
-31.6
90.2
HS
-24.42
-22.2
-22.9
-24.2
98.6
-25.03
0.26
0.19
-24.9
98.7
-28.93
0.18
-27.4
-29.4
0.6
-25.0
-27.0
95.4
0.23
-28.1
-30.5
107.7
-35.34 + 0.25
-33.9
-36.3
99.3
0.24
-33.5
-35.9
100.8
-27.42
HS
\
-28
-30.9
HS
-35.37
85.25
90.14
90.2
-1-1- ..........
Chapter 7: A thermochemical database for organosulfur compounds
134
Table 3. Comparison of calculated (triple-zeta) and experimental enthalpies of formation and standard
entropies for 59 gas-phase species selected for training set (Continued).
ArH
Experiment
S
298
So 2 9 8 [caV(mol K)]
[kcaVmo]
No BAC
MBAC
Experiment
This Work
-29.1
-30.3
0.6
0.6
-26.5
-26.8
-29.1
-29.4
107.6
107.7
-35.3
0.6
-31.9
-34.4
98.7
-33.93
0.3
-31.4
-34.0
100.3
1.2
1.3
S
1.2
^'
S
6
S
0
53.1
-35.97
0.36
-36.0
-38.9
74.6
-49.1
0.4
-46.2
-50.3
92.7
-60.9
0.4
-54.5
-59.7
111.4
-65.5
0.4
-59.6
-64.8
102.2
-70.939
0.012
-71.2
-70.9
-102.61
0.62
-100.3
-101.5
91.4
-103.6
0.7
-101.2
-102.4
94.2
-115.5
0.5
-110.8
-113.6
90.1
-125.2
0.5
-121.0
-124.4
99.2
-131.9
0.46
-128.4
-132.3
106.9
-164.2
0.5
-165.2
-165.1
93.4
0.0024
-181.0
-182.1
108.6
30.4
30.7
27.6
27.3
59.327
59.4
0
o5 o
d '-b
6
*
s*
53.045
o
o
-180.7
SC=
SoS
HS-
30.74
27.95 0.26
SH
HS
-SH
HS-'
SH
HS
SH
54.54
56.88
54.6
56.9
-2.23
0.55
-3
.0
-2.4
-5.76
-5.8
-5.8
-7.13
0.28
0.44
-7.5
-7.5
80.7
91.3
-12.08
0.25
-11.4
-11.9
102.9
-17.85
-16.8
-17.9
98.0
-15.6
-14.8
-16.7
-16.2
111.3
-19.83
0.36
0.36
0.35
0.35
-17.5
-19.5
105.8
118.9
-28.04
0.26
-25.6
-27.9
116.3
-16.97
-15.58
83.8
80.48
Chapter 7: A thermochemical database for organosulfur compounds
135
Table 4. Comparison of calculated (triple-zeta) and experimental enthalpies of formation and standard
entropies for 58 additional species.
S
AfH029 8 [kcaVmol]
Experiment
S
rI
SS
No BAC
MBAC
2 98
[caV(mol K)]
Experiment
This Work
19.7
0.24
19.1
18.6
61.0
14.6
0.3
15.8
14.7
68.7
2.7
0.3
11.6
10.5
70.0
0.24
29.5
26.6
66.6
25.33 t 0.96
42.0
39.2
21.67
0.30
21.3
19.0
71.3
20.86
0.28
21.8
19.5
70.9
-8.02
0.28
-6.4
-8.0
72.0
20.16
0.22
22.9
19.4
76.8
19.74
0.22
22.8
19.4
76.7
-15.18+ 0.25
-13.1
-15.2
77.5
-15.27
0.18
-12.3
-14.4
80.2
-14.47
0.20
-12.3
-14.4
80.6
-11.42
0.18
-9.0
-11.0
85.6
0.7
7.0
4.4
93.1
S
27.49
S
S
S
S
S
1/
S
S
S
81.43
81.6
OH
S^
4.3
S
SH
26.86
0.21
30.8
26.3
80.5
SH
-15.38
0.23
-11.6
-14.3
93.4
-22.88
0.19
-19.9
-22.5
87.0
-42.5
1.2
-41.0
-42.1
81.7
-32.33
0.47
-30.5
-32.3
78.0
-46.89
0.47
-38.9
-40.8
76.3
19.3
13.9
84.8
-52.9
-54.6
90.2
oSH
S
0;
o
S
5.98
0.72
0
-54.45
0.22
Chapter 7: A thermochemical database for organosulfur compounds
136
Table 4. Comparison of calculated (triple-zeta) and experimental enthalpies of formation and standard
entropies for 58 additional species (Continued).
AfH
Experiment
0
S
29 8
S 0 2 9 8 [caV(mol K)]
[kcaVmol]
No BAC
MBAC
Experiment
-31.79
0.61
-36.4
-38.8
82.2
-37.10
0.65
-37.1
-39.5
82.4
-89.1
0.8
-89.6
-89.7
76.6
-29.7
0.7
-29.7
-31.1
74.6
-97.6
0.8
-94.8
-95.4
87.5
-61.1
0.4
-54.5
-56.5
81.7
-62.61
0.73
-53.3
-55.2
81.0
-31.2
-33.7
86.4
S
0
S
9=0
Sx.
This Work
0
'0
sO
,
S
-36 + 0.9
0
-71.5
0.48
-60.8
-63.3
88.8
S OH
-69.7
0.52
-60.8
-63.3
88.8
\
-77.03
0.65
-73.0
-75.4
100.6
-109.87
0.63
-103.1
-104.8
106.4
-111.92
0.63
-108.1
-110.4
113.7
-113.1
0.9
-108.8
-110.5
97.5
-94.591
-95.3
-97.6
61.37
61.5
-175.7
-175.0
-174.2
71.41
74.6
25.23
26.0
17.9
60.88
17.3
25.2
17.7
60.9
70.9
-11.8
-11.2
-11.8
91.0
-18.1
-19.2
97.8
-26.8
-26.3
-28.0
101.4
-32._8
-31.6
-3.9
109.0
dl \
S-S
S
..os
-
S
-18.9
S
<
1.0
70.82
Chapter 7: A thermochemical database for organosulfur compounds
137
Table 4. Comparison of calculated (triple-zeta) and experimental enthalpies of formation and standard
entropies for 58 additional species (Continued).
AfH%98 [kcaVmol]
0-S S
S
OI '>=OS
S0 29 8 [cal/(mol K)]
This Work
Experiment
Experiment
No BAC
MBAC _
-13.5
-13.0
-15.6
-5.3
-6.6
74.7
-29.3
76.9
39.8
40.7
63.2
1.6
50.7
49.7
77.1
0.48
28.6
28.2
79.5
15.8
15.6
92.9
28.2
27.3
87.7
34.84
52.6
53.8
71.8
26.14
29.4
30.9
77.9
24.36
36.0
37.8
85.5
27.17
26.3
28.4
94.0
24
32.7
35.1
109.8
-3.6
1.2
-30.1
1.2
63.788
63.9
S
S
33.81
S
S
S
S
QS
S
60.5
22.41
IfS
13.8
S
S
18.53
S-S
S-s
5S
S
S
0.64
S s
S-S
S-S'
7.5 Conclusions
Thermochemical parameters have been calculated for an extensive set of organic compounds
containing sulfur. A new set of bond additivity corrections were regressed to improve the
agreement of calculated enthalpies with experimental data, and entropies and heat capacities
were validated with available measurements. Calculations of thermochemical parameters were
conducted for a set of additional sulfur compounds, to allow for the regression of a more
accurate group additivity scheme, which can be used in thermochemical parameter estimation for
pyrolysis and oxidation models involving sulfur.
138
Chapter 7: A thermochemical database for organosulfur compounds
The best thermochemical estimation schemes combine high-level ab initio methods with
accurate experimental data. In some cases, such as organosulfur chemistry, experimental data are
deficient. More accurate experiments, and the inclusion of sulfur compounds in the Active
Thermochemical Tables, 2 would allow for the regression of additivity corrections with much
lower uncertainty. And while improved quantum chemistry methods, including multireference
coupled cluster methods2 5 2 7 and improved treatments of anharmonicity,28,
29
may make the
regression of bond additivity corrections redundant in the future, experimental data will remain
necessary to confirm that organosulfur thermochemistry is correctly modeled in the future.
7.6
References
1.
B. Ruscic, R. E. Pinzon, M. L. Morton, N. K. Srinivasan, M.-C. Su, J. W. Sutherland and
J. V. Michael, J. Phys. Chem. A, 2006, 110, 6592-6601.
B. Ruscic, R. E. Pinzon, M. L. Morton, G. von Laszevski, S. J. Bittner, S. G. Nijsure, K.
A. Amin, M. Minkoff and A. F. Wagner, J. Phys. Chem. A, 2004, 108, 9979-9997.
A. G. Vandeputte, M. K. Sabbe, M.-F. Reyniers and G. B. Marin, Chemistry-A European
Journal, 2011, 17, 7656-7673.
M. K. Sabbe, A. G. Vandeputte, M.-F. Reyniers, V. V. Speybroeck, M. Waroquier and G.
B. Marin, Journal of Physical Chemistry A, 2007, 111, 8416-8428.
G. A. Petersson, D. K. Malick, W. G. Wilson, J. W. Ochterski, J. A. Montgomery and M.
J. Frisch, Journal of Chemical Physics, 1998, 109, 10570-10579.
C. F. Goldsmith, G. R. Magoon and W. H. Green, Journal of Physical Chemistry A,
2012, 116, 9033-9057.
S. W. Benson and J. H. Buss, Journal of Chemical Physics, 1958, 29, 546-572.
S. W. Benson, F. R. Cruickshank, D. M. Golden, G. R. Haugen, H. E. O'Neal, A. S.
Rodgers, R. Shaw and R. Walsh, Chemical Reviews, 1969, 69, 279-324.
M. R. Harper, K. M. Van Geem, S. P. Pyl, G. B. Marin and W. H. Green, Combust.
Flame, 2011, 158, 16-41.
K. M. Van Geem, M.-F. Reyniers, G. B. Marin, J. Song, W. H. Green and D. M. Matheu,
AIChE J., 2006, 52, 718-730.
C. A. Class, J. Aguilera-Iparraguirre and W. H. Green, Submitted, 2014.
W. Klopper and J. Noga, in Chemistry and Physics, ed. J. Rychlewski, Dordrecht, 2003.
J. Noga and W. Kutzelnigg, Journal of Chemical Physics, 1994, 101, 7738-7762.
J. Aguilera-Iparraguirre, A. D. Boese, W. Klopper and B. Ruscic, Chemical Physics,
2008, 346, 56-68.
J. Aguilera-Iparraguirre, H. J. Curran, W. Klopper and J. M. Simmie, Journal of Physical
Chemistry A, 2008, 112, 7047-7054.
W. Klopper, R. A. Bachorz, D. P. Tew, J. Aguilera-Iparraguirre, Y. Carissan and C.
Hdttig, Journal of Physical Chemistry A, 2009, 113, 11679-11684.
B. Buesser, J. Aguilera-Iparraguirre and W. H. Green, In preparation.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
..................
Chapter 7: A thermochemical database for organosulfur compounds
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
139
M. T. Timko, E. Schmois, P. R. Patwardhan, Y. Kida, C. A. Class, W. H. Green, R. K.
Nelson and C. M. Reddy, Energy & Fuels, 2014, 28, 2977-2983.
X. Zheng, J. W. Bozzelli, E. M. Fisher, F. C. Gouldin and L. Zhu, Proceedings of the
Combustion Institute, 2011, 33, 467-475.
X. Zheng, E. M. Fisher, F. C. Gouldin and J. W. Bozzelli, Combust. Flame, 2011, 158,
1049-1058.
M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman,
J. Montgomery, J. A., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar,
J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H.
Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T.
Nakajima, Y. Honda, 0. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J.
B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, 0. Yazyev,
A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A.
Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C.
Strain, 0. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V.
Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A.
Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham,
C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M.
W. Wong, C. Gonzalez and J. A. Pople, Gaussian 03, (2004) Gaussian, Inc., Wallingford
CT.
H.-J. Werner, P. J. Knowles, G. Knizia, F. R. Manby and M. Schiltz, Wiley
Interdisciplinary Reviews: Computational Molecular Science, 2011, 2, 242-253.
S. Sharma, M. R. Harper and W. H. Green, CanTherm open-source software package,
(2010). http://github.com/GreenGroup/CanTherm
P. J. Linstrom and W. G. Mallard, eds., NIST Chemistry WebBook, NIST Standard
Reference Database Number 69, National Institute of Standards and Technology,
Gaithersburg, MD.
F. A. Evangelista, W. D. Allen and H. F. Schaefer, J. Chem. Phys., 2006, 125, 1-16.
F. A. Evangelista, W. D. Allen and H. F. Schaefer, J. Chem. Phys., 2007, 127, 1-17.
J. Pittner, J. Chem. Phys., 2003, 118, 10876-10889.
J. Zheng, T. Yu, E. Papajak, I. M. Alecu, S. L. Mielke and D. G. Truhlar, Phys Chem
Chem Phys, 2011, 13, 10885-10907.
B. C. Garrett and D. G. Truhlar, Journal of Physical Chemistry, 1979, 83, 1915-1924.
140
Chapter 7: A thermochemical database for organosulfur compounds
Appendix: Calculated Thermochemical Parameters
7.7
P
AffFI
so
[kcaVmnol]
298 K
-6.5
65.6
28.0
-5.7
[ca/(nol K)]
298 K
49.1
50.3
55.1
61.1
300 K
8.2
7.2
9.2
12.1
400 K
8.5
7.4
10.3
13.9
500 K
8.9
7.7
11.4
15.7
HS
18.8
-7.9
-10.1
-13.6
65.4
68.8
70.9
80.1
14.1
17.9
17.6
23.3
16.6
20.9
20.9
27.8
19.0
23.7
23.9
32.0
600 K
9.3
7.9
12.4
17.4
21.1
26.3
26.7
35.9
HS
-15.9
78.0
23.5
28.3
32.6
36.4
42.5
-11.6
43.8
79.8
81.6
23.0
25.4
27.4
29.9
31.6
33.8
35.4
37.0
HS
-20.8
82.8
31.0
37.4
42.8
HS
HS
-19.0
-18.9
87.3
87.1
29.3
35.2
29.2
-16.0
88.0
-17.3
[caY(nol K)]
800 K
1000 K
10.1
10.8
8.2
8.5
14.1
15.3
20.2
22.3
24.6
27.2
34.2
30.8
31.2
34.6
42.2
47.0
2000 K
12.1
8.7
17.1
25.8
31.1
39.6
40.0
54.2
12.8
8.8
18.1
27.6
33.2
42.5
42.8
58.0
13.1
8.8
18.6
28.5
34.2
43.9
44.1
59.8
47.1
54.3
58.0
59.8
41.7
42.1
46.4
45.8
53.8
51.3
57.7
54.2
59.6
55.6
47.3
54.4
59.7
68.2
72.8
75.1
40.8
45.7
53.7
59.6
68.6
73.3
75.5
35.3
40.8
45.7
53.7
59.6
68.6
73.3
75.5
28.7
34.3
39.7
44.7
52.8
58.9
68.1
72.9
75.2
86.6
29.4
35.3
40.6
45.4
53.1
59.0
68.1
72.9
75.3
-SH
5.3
85.6
27.1
34.9
42.1
48.5
58.5
65.6
76.1
81.3
83.8
10.6
-19.9
93.1
98.3
34.6
35.9
41.3
42.4
47.0
48.8
51.8
54.6
59.5
64.3
65.2
71.6
74.1
82.7
78.9
88.5
81.2
91.1
HS
-25.2
93.1
35.3
43.0
49.9
55.8
65.4
72.4
83.0
88.6
91.3
HS
-23.6
93.9
35.2
42.5
49.3
55.2
64.9
72.0
82.9
88.5
91.2
HS
HS
-21.7
95.7
35.3
42.7
49.4
55.3
64.9
72.0
82.9
88.5
91.2
-21.5
96.1
34.9
42.1
48.8
54.8
64.6
71.9
82.8
88.5
91.2
HS
-25.9
90.2
35.5
43.3
50.2
56.2
65.7
72.7
83.2
88.7
91.3
90.9
90.9
:C=s
H 2C=S
H 3C-SH
S
HS
S
S
S
2400
K
1500 K
H2 S
S
-18.3
-19.0
98.6
98.7
35.1
34.3
41.9
41.3
48.4
47.9
54.2
54.0
63.9
63.9
71.2
71.2
82.3
82.4
88.2
88.2
S
7.6
90.2
35.6
43.3
50.0
55.7
65.0
71.9
82.5
88.2
91.0
-21.1
95.4
35.5
42.5
49.0
54.8
64.4
71.5
82.5
88.2
90.9
80.5
25.4
32.2
38.2
43.2
50.6
55.7
62.8
66.4
68.2
-7.1
93.4
33.1
42.1
50.4
57.7
69.3
77.7
90.1
96.3
99.3
-SH
15.3
87.0
31.6
41.2
49.9
57.6
69.6
78.3
91.0
97.4
100.4
HS
-23.0
107.7
42.1
49.7
57.1
64.0
75.4
84.0
97.0
103.7
106.8
HS
-29.0
99.3
41.4
50.3
58.4
65.4
76.7
85.0
97.5
103.9
107.0
HS
-28.6
100.8
41.6
50.9
59.0
66.0
77.2
85.3
97.6
104.1
107.1
-21.6
-21.9
107.6
107.7
41.4
40.8
49.3
48.8
56.8
56.5
63.7
63.5
75.0
74.9
83.5
83.5
96.6
96.6
103.4
103.4
106.6
106.6
S
'
SH
I H34.4
S
o
S
141
Chapter 7: A thermochemical database for organosulfur compounds
S
ArH
[kcalmoll
298 K
So
[cal(nol K)]
298 K
300 K
400 K
500 K
-27.1
98.7
41.4
50.3
58.2
65.1
76.2
-26.6
100.3
42.1
50.7
58.4
65.1
76.0
C0
[caV(mol K)]
800 K
1000 K
600 K
1500 K
2000 K
2400 K
84.4
97.0
103.6
106.8
84.2
96.8
103.4
106.6
S
-s --
-0.5
53.1
7.3
7.6
7.8
8.1
8.4
8.6
8.8
8.8
8.9
0* S
-35.8
74.6
22.0
26.0
29.2
32.0
36.6
40.0
45.5
48.5
49.9
, SS
-39.8
81.7
24.3
29.1
33.2
36.8
42.2
46.1
51.7
54.5
55.7
20.4
84.8
29.3
34.9
39.5
43.1
48.6
52.4
57.8
60.5
61.7
-50.7
90.2
28.0
34.2
39.9
44.9
52.8
58.3
66.1
69.9
71.5
-44.1
92.7
33.6
40.0
45.7
50.7
58.8
64.8
74.1
78.9
81.2
-50.4
111.4
45.4
54.0
62.1
69.3
80.9
89.6
102.6
109.3
112.5
-55.6
102.2
46.3
55.9
64.0
70.9
82.1
90.3
103.0
109.6
112.8
-74.5
59.4
9.6
10.4
11.1
11.7
12.5
12.9
13.4
13.6
13.7
-90.0
76.6
24.5
29.0
32.8
36.1
41.2
45.0
51.0
54.2
55.6
-94.1
87.5
29.8
35.7
40.8
45.2
52.2
57.4
65.2
69.3
71.3
-30.6
86.4
32.5
38.6
43.4
47.2
52.8
56.8
62.7
65.9
67.4
-63.4
91.2
33.2
39.9
45.5
50.3
57.8
63.1
71.1
75.1
77.1
-98.7
91.4
34.1
42.0
49.0
55.2
65.0
72.1
82.6
88.0
90.5
-99.7
94.2
36.0
43.3
49.5
54.9
63.5
69.8
79.4
84.5
86.9
-70.8
100.6
41.4
49.1
55.6
61.0
69.5
75.7
85.3
90.4
92.9
-100.4
106.4
41.9
49.9
57.3
63.8
74.3
82.1
93.8
99.8
102.6
-104.4
113.7
47.1
56.4
65.1
72.8
85.3
94.4
108.0
114.9
118.2
-106.3
97.5
42.4
51.4
58.9
65.3
75.3
82.7
94.0
99.9
102.7
-98.8
61.5
12.2
13.8
15.2
16.2
17.6
18.3
19.1
19.4
19.6
-111.0
90.1
27.3
32.2
36.7
40.5
46.3
50.2
55.8
58.4
59.6
-120.3
99.2
33.4
39.6
45.3
50.2
57.7
62.9
70.3
73.8
75.4
-126.6
106.9
39.4
47.1
54.1
60.2
69.5
76.0
85.1
89.4
91.4
-177.2
74.6
22.1
24.7
26.4
27.5
29.2
30.4
32.4
33.6
34.1
-165.6
93.4
31.2
35.8
40.0
43.7
49.7
54.1
60.6
63.9
65.4
-179.4
108.6
42.9
50.5
57.4
63.5
73.2
80.1
90.3
95.3
97.6
28.9
25.1
-3.6
54.6
56.9
83.8
7.8
10.8
23.8
8.1
11.8
27.2
8.3
12.5
30.2
8.5
13.0
32.9
8.7
13.7
37.2
8.8
14.0
40.5
8.9
14.5
45.6
8.9
14.7
48.2
8.9
14.7
49.4
-5.8
80.7
22.7
26.1
29.2
32.1
36.7
40.1
45.4
48.1
49.4
8
0
0^ S 0
d/
S
d/ \
O.,.
Ns
LO
\b
1\0
Od/ \OS'b
S <
' \b
0
~S
HO SOH
HO O
o a
& 'b
S'-S*
S=C=S
HS--
SH
142
Chapter 7: A thermochemical database for organosulfur compounds
ArH-i1
SO
[kcaVmnol]
298 K
[ca/(nol K)]
298 K
300 K
400 K
500 K
-23.5
101.4
40.6
48.6
55.6
61.6
71.1
-14.3
118.9
45.6
54.0
62.0
69.2
-21.7
116.3
46.2
54.9
63.0
-27.9
109.0
47.2
56.5
-16.7
63.9
10.6
14.0
92.9
17.3
Co*
[caV(mol K)J
1000 K
600 K
800 K
1500 K
2000 K
2400 K
78.1
88.6
94.1
96.6
80.9
89.4
102.1
108.6
111.7
70.1
81.7
90.1
102.7
109.0
112.0
64.7
71.7
82.9
91.1
103.2
109.5
112.4
11.4
12.1
12.5
13.1
13.3
13.6
13.8
13.8
31.4
36.1
40.1
43.4
48.5
52.0
57.1
59.5
60.7
77.0
22.9
26.2
29.1
31.7
36.2
39.8
45.5
48.5
49.9
11.4
82.5
29.3
34.7
39.1
42.7
48.5
52.9
60.0
63.9
65.7
59.9
74.3
19.6
22.6
25.1
27.2
30.5
33.0
37.0
39.1
40.1
54.4
81.4
25.4
29.9
33.7
36.8
41.8
45.5
51.2
54.3
55.7
20.0
67.6
15.3
18.1
20.5
22.5
25.6
27.9
31.5
33.4
34.3
12.5
75.3
21.0
25.2
28.8
31.8
36.5
40.0
45.6
48.6
50.0
63.6
14.8
63.0
84.2
14.2
29.3
16.0
34.4
17.4
38.6
18.4
42.1
20.0
47.9
21.1
52.3
22.9
59.6
24.0
63.6
24.5
65.4
9.0
89.8
35.0
42.1
48.0
52.8
60.2
65.7
74.2
78.8
81.2
57.0
82.6
25.6
29.8
33.3
36.4
41.3
45.0
50.9
54.0
55.5
51.8
89.9
31.8
37.2
41.8
45.8
52.4
57.3
65.0
69.2
71.2
16.7
75.4
22.5
26.2
29.4
32.1
36.6
40.0
45.5
48.5
49.8
9.6
82.5
28.4
33.6
37.9
41.5
47.4
52.0
59.5
63.5
65.4
60.6
72.7
19.6
22.6
25.0
27.0
30.2
32.7
36.7
38.9
39.9
32.0
88.5
30.9
39.0
46.1
52.2
61.4
67.7
76.9
81.4
83.5
86.8
79.3
23.2
26.7
29.6
32.0
35.8
38.5
42.6
44.7
45.7
57.5
93.6
34.4
43.4
51.1
57.6
67.4
74.1
83.5
88.1
90.3
133.7
73.8
21.2
24.3
26.4
27.9
29.9
31.3
33.8
35.3
36.0
101.4
91.6
33.0
40.8
47.3
52.7
60.6
66.1
73.9
77.7
79.5
HS- S
HS-SH
1.0
-2.6
61.8
72.7
11.7
17.4
13.0
19.8
14.1
21.9
15.0
23.6
16.2
26.5
16.9
28.5
17.8
31.6
18.3
33.2
18.5
33.9
HS-S
20.3
87.7
26.7
30.7
34.3
37.4
42.3
45.9
51.3
53.9
55.2
64.6
84.3
24.9
28.3
31.0
33.2
36.5
38.8
42.4
44.3
45.2
34.0
101.0
36.2
44.5
51.7
57.9
67.2
73.6
82.5
86.8
88.8
3.1
73.5
17.1
18.9
20.3
21.2
22.4
23.1
23.8
24.0
24.0
25.6
27.9
29.8
34.8
37.6
38.9
39.4
36.5
17.7
39.2
18.9
32.8
43.6
20.6
46.7
21.8
51.4
23.5
53.8
24.3
54.8
24.6
'
S
S
HS
HS>
HS>
HS
HS
-
HS
HS
S
K
-
-
S
S
S
/
K>
S-S
S-S
SS
HS
HS
SH
S/
-5-/HS
S
-0.9
-4.9
24.0
83.9
90.5
66.8
22.9
29.7
14.3
33.4
16.2
143
Chapter 7: A thermochemical database for organosulfur compounds
ArfI
298 K
so
[caV(nol K)]
298 K
300 K
400 K
500 K
1500 K
2000 K
2400 K
18.2
46.4
75.5
80.7
19.1
23.3
22.5
27.4
25.4
30.7
27.8
33.4
31.5
37.2
34.1
39.8
37.7
43.4
39.4
45.1
40.1
45.9
93.0
78.2
21.3
24.2
26.3
28.0
30.4
32.1
34.4
35.5
36.0
61.5
96.2
32.8
40.8
47.6
53.3
61.5
67.1
74.6
78.0
79.5
54.3
81.7
20.8
24.0
26.7
28.9
31.9
33.6
35.3
35.7
35.8
HS---SH
28.3
1.9
78.1
73.1
19.7
19.8
21.9
21.4
23.6
22.7
24.9
23.9
26.7
26.0
27.8
28.0
29.2
31.2
29.7
33.0
29.8
33.7
HS '1,SH
-4.1
81.8
25.4
28.5
31.1
33.4
37.3
40.4
45.5
48.1
49.3
25.8
75.7
21.9
25.3
27.7
29.5
32.0
33.8
36.9
38.7
39.6
19.7
82.6
27.9
32.9
36.6
39.4
43.5
46.4
51.2
54.0
55.2
31.1
89.9
31.4
39.4
46.5
52.4
61.5
67.9
77.0
81.5
83.6
33.8
100.5
38.5
47.9
56.1
62.9
73.4
80.7
91.2
96.5
99.0
15.1
76.0
19.4
23.2
26.9
30.2
35.5
39.4
45.3
48.3
49.7
10.3
81.8
25.7
30.9
35.7
39.9
46.7
51.7
59.4
63.5
65.4
S
3.7
85.6
31.7
38.8
44.8
50.1
58.4
64.5
73.9
78.8
81.1
S
42.1
68.6
17.2
21.1
24.4
27.2
31.4
34.1
37.8
39.6
40.6
S
88.5
68.1
15.9
18.4
20.4
22.0
24.3
25.8
28.2
29.6
30.2
56.2
82.0
27.5
35.2
41.9
47.5
55.9
61.4
68.9
72.6
74.4
54.1
68.4
15.6
18.6
21.1
23.1
25.9
27.5
29.1
29.7
30.0
9.9
73.0
19.0
22.8
26.4
29.6
35.0
39.0
45.1
48.2
49.7
34.5
77.8
23.6
28.4
32.5
36.0
41.5
45.4
51.3
54.3
55.8
79.5
76.7
21.4
25.0
28.1
30.7
34.8
37.7
42.2
44.6
45.7
49.3
90.9
33.2
42.2
50.0
56.4
66.2
73.0
82.6
87.2
89.4
45.3
79.3
21.4
25.4
28.8
31.7
35.9
38.7
42.6
44.4
45.2
15.6
75.9
19.3
22.5
25.1
27.4
31.0
33.6
37.4
39.2
40.0
13.3
85.2
27.7
32.5
37.0
40.9
47.5
52.4
59.9
63.8
65.6
8.4
92.1
33.4
39.9
45.6
50.6
58.6
64.6
73.9
78.8
81.1
0.9
98.6
38.9
47.0
54.1
60.4
70.3
77.6
88.6
94.4
97.1
78.5
91.3
28.3
32.9
37.1
40.9
46.8
51.1
57.5
60.7
62.3
63.2
84.9
24.7
27.8
30.4
32.6
36.0
38.5
42.3
44.3
45.2
29.9
96.6
37.2
44.4
50.5
55.8
64.0
70.1
79.5
84.5
86.9
46.2
102.1
35.7
41.2
45.6
49.1
54.4
58.0
63.1
65.6
66.7
[kcaYmol]
HS, S.-S
HS
SH
SH
-
HS
C0
[ca/(mol K)]
600 K
1000 K
800 K
S
HS
S
S
SH
S
S
AN 5
NS
5
SH
S
ASH
S
NS---- SH
'-S r
S
'
zzS S
144
Chapter 7: A thermochemical database for organosulfur compounds
S
ArHP
So
[kcalrnol]
298 K
[caV(nol K)]
298 K
300 K
400 K
500 K
1500 K
2000 K
2400 K
14.7
85.1
24.5
29.2
33.4
37.0
42.6
46.4
52.0
54.7
55.9
6.0
92.9
28.8
35.1
40.8
45.7
53.3
58.6
66.2
69.9
71.6
11.5
85.5
24.8
29.4
33.3
36.8
42.1
45.9
51.5
54.3
55.6
26.9
93.8
34.2
42.0
48.9
55.0
64.3
70.9
80.4
85.1
87.3
28.1
85.2
29.7
35.9
41.1
45.5
52.4
57.5
65.2
69.2
71.2
34.6
76.7
22.7
28.1
32.7
36.5
42.4
46.4
52.1
54.8
56.1
75.5
86.4
26.9
31.9
36.2
39.9
45.7
50.0
56.4
59.7
61.3
61.2
109.3
39.4
48.0
55.7
62.2
72.1
78.8
88.0
92.4
94.4
96.6
89.5
27.5
30.8
33.0
34.6
36.8
38.2
39.9
40.8
41.1
38.2
89.1
26.1
31.0
35.5
39.3
45.5
49.8
56.1
59.2
60.7
-20.3
92.1
35.4
42.5
48.3
53.2
60.8
66.3
74.8
79.1
81.1
9.2
86.0
25.6
27.5
29.0
30.2
32.3
34.0
36.8
38.2
38.9
1.9
91.2
32.8
36.5
39.0
41.0
44.3
46.9
51.4
53.7
54.8
-10.9
86.3
31.6
36.5
40.4
43.8
49.2
53.3
60.0
63.5
65.1
-14.2
97.1
36.6
42.9
48.2
52.8
60.1
65.6
74.2
78.7
80.8
-1.2
99.3
38.1
42.9
46.8
50.1
55.3
59.3
65.4
68.7
70.2
71.9
108.2
43.5
56.7
68.0
77.3
91.1
100.5
113.5
120.1
123.2
128.8
119.6
47.3
56.9
65.1
71.9
82.2
89.2
99.0
103.7
105.8
43.9
126.7
47.3
55.7
63.0
69.1
78.3
84.6
93.1
97.2
99.0
-43.1
68.6
18.4
21.1
22.9
24.4
26.4
28.0
30.9
32.7
33.5
-50.7
75.8
24.6
28.7
31.7
34.1
37.6
40.3
45.1
47.9
49.2
-53.9
84.9
31.0
36.0
40.0
43.2
48.5
52.5
59.3
63.0
64.8
-58.8
82.5
31.0
36.5
40.7
44.0
49.0
52.8
59.4
63.1
64.8
-30.9
-27.8
64.5
62.3
13.5
11.7
15.2
13.9
16.7
15.8
18.0
17.5
19.9
20.0
21.3
21.6
23.2
23.7
24.0
24.5
24.3
24.8
-41.8
73.3
18.7
22.0
24.8
27.1
30.7
33.2
37.1
38.9
39.7
C0
[caV(mol K)]
800 K
600 K
1000 K
SS
HS
S
S
S
S
1 S" S
S
S
HS
SH
HS rS H
SH
SH
SH
SH
SH
HSSH
H
S
/
SH
S-SH
S <
SH
HO__SH
SH
SH
H
H
,TSH
OH
O'SH
HO-'S
0
ASH
145
Chapter 7: A thermochemical database for organosulfur compounds
C
ArHI
S
[kcaVmoll
298K
[caV(mol K)]
298K
300K
400K
500K
-38.0
71.0
16.7
20.1
23.3
26.1
30.4
45.3
82.7
25.1
29.3
33.0
36.2
-41.8
81.0
22.1
26.9
31.3
-89.6
69.5
16.8
19.7
-75.1
66.8
17.0
-101.2
14.2
94.5
76.4
-88.3
[cal(mol K)]
1000K
800K
600K
1500K
2000K
2400K
33.5
37.8
39.6
40.4
41.5
45.3
51.1
54.0
55.2
35.2
41.3
45.6
51.7
54.7
56.0
22.1
23.9
26.4
27.8
29.2
29.6
29.7
20.3
22.9
24.8
27.2
28.3
29.2
29.8
30.0
32.4
20.8
37.8
24.5
42.4
46.4
52.6
57.2
64.1
67.7
69.4
27.9
31.0
36.0
39.7
45.5
48.5
49.9
77.2
21.1
23.9
26.2
28.0
30.8
32.8
35.9
37.5
38.3
-45.7
-36.9
76.0
76.9
24.1
23.6
28.6
27.2
32.0
30.2
34.5
32.7
38.0
36.8
-39.7
-32.7
83.5
71.2
27.6
17.4
32.6
20.6
37.1
23.4
41.1
25.9
47.5
29.8
44.9
45.2
59.7
37.1
47.7
47.9
63.6
39.2
49.0
49.2
65.5
0
40.5
40.0
52.3
32.8
S
-16.7
70.5
16.0
19.3
22.3
25.0
29.3
32.4
36.9
39.1
40.0
-46.6
80.2
22.3
27.3
31.8
35.7
41.7
45.8
51.6
54.3
55.5
79.6
21.9
26.2
30.4
34.3
40.7
45.4
52.0
54.9
56.1
50.7
90.2
28.0
34.2
319.9
44.9
52.8
58.3
66.1
69.9
71.5
-22.7
81.1
28.4
33.3
36.8
39.4
43.2
46.1
50.9
53.7
55.0
-8.6
93.9
38.1
47.0
54.1
59.7
67.9
73.6
82.2
86.7
88.8
OH
S
0
SH
OH
0
OH
HS
S
OH
SH
HO
HOH
SH
SH
OH
HO
S
OH
0
SH
.
..
T
f-31.4
TSH
H
H
/
H
'-
H
40.0
146
Chapter 7: A thermochemical database for organosulfur compounds
Cyclic Species
Af11P
[kcal/moll
298 K
so
[caV(mol K)]
298 K
300 K
400 K
500 K
C0
P
[caV(mol K)]
1000 K
800 K
600.K
1500 K
2000 K
2400 K
28.0
31.9
34.0
35.0
19.9
61.0
12.9
16.2
19.1
21.6
25.4
[
17.3
68.7
17.2
22.1
26.6
30.4
36.4
40.7
46.9
50.0
51.6
S
13.4
70.0
18.9
23.5
27.5
31.0
36.4
40.3
46.1
49.2
50.8
31.0
66.6
17.5
22.7
27.0
30.4
35.3
38.5
43.1
45.5
46.6
23.3
71.3
20.0
25.8
30.8
35.1
41.5
46.0
52.5
55.8
57.4
24.0
70.9
20.0
25.7
30.8
35.0
41.5
46.0
52.5
55.8
57.4
72.0
21.6
28.2
34.2
39.4
47.5
53.3
61.8
66.0
68.1
25.3
76.8
23.0
29.1
34.4
38.8
45.6
50.3
57.0
60.6
62.3
25.3
76.7
23.2
29.2
34.5
39.0
45.7
50.4
57.1
60.6
62.3
-9.5
77.5
26.3
34.6
42.1
48.5
58.7
66.0
76.7
82.1
84.7
-8.7
80.2
27.6
35.7
42.9
49.0
58.7
65.6
75.9
81.3
83.8
-8.7
80.6
27.8
35.7
42.8
49.0
58.6
65.5
75.9
81.3
83.8
-28.4
78.0
23.6
29.7
35.1
39.7
46.5
51.3
58.1
61.7
63.3
-37.1
76.3
23.1
29.3
34.7
39.2
46.2
51.1
58.0
61.6
63.3
-33.2
82.2
28.1
36.0
42.8
48.6
57.5
63.8
73.0
77.6
79.8
-33.9
82.4
28.2
36.0
42.9
48.6
57.6
63.8
73.0
77.6
79.8
-30.1
74.6
22.2
27.5
32.0
35.6
40.9
44.3
48.9
51.5
52.7
-53.6
81.7
26.6
33.7
39.6
44.6
51.9
56.8
63.8
67.5
69.2
-52.5
81.0
26.6
33.6
39.6
44.5
51.8
56.7
63.8
67.4
69.2
88.8
32.6
40.8
47.8
53.7
62.6
68.9
78.0
82.5
84.7
88.8
32.6
40.8
47.8
53.7
62.6
68.9
78.0
82.5
84.7
-7.1
74.7
21.5
25.8
29.1
31.6
34.9
36.9
39.8
41.3
42.0
-29.7
76.9
23.2
28.3
32.5
35.9
40.9
44.2
49.1
51.6
52.7
S
S
37.3
63.2
11.6
12.4
13.0
13.3
13.6
13.7
13.8
13.9
13.9
47.7
77.1
22.4
26.7
30.0
32.4
35.5
37.4
40.1
41.5
42.1
cs> :s
S
/
S
S
S
-3.8
01/
S
S
S
S
0
0
0
9=0
S'
-59.0
S
IS -- O
S
CS-O
147
Chapter 7: A thermochemical database for organosulfur compounds
So
Cp*
[kcaVnol]
298 K
[caV(inol K)]
298 K
300 K
400 K
500 K
[caV(nol K)]
1000 K
800 K
600 K
26.3
79.5
24.3
29.4
33.5
36.8
41.5
27.1
87.7
29.3
35.8
41.3
45.9
48.8
71.8
16.9
18.2
18.9
24.0
77.9
20.1
21.8
29.7
85.5
25.6
18.6
94.0
23.8
ArH-
S
CS
S
1500 K
2000 K
2400 K
44.7
49.4
51.7
52.8
52.8
57.5
64.4
67.7
69.3
19.2
19.4
19.5
19.7
19.8
19.8
22.6
23.0
23.3
23.4
23.7
23.8
23.8
27.4
28.3
28.7
29.1
29.3
29.6
29.7
29.7
30.8
32.9
33.9
34.4
34.9
35.2
35.6
35.7
35.7
109.8
37.9
40.2
41.4
42.1
42.8
43.1
43.4
43.6
43.6
-17.2
78.1
25.1
31.3
36.5
40.7
47.0
51.2
57.5
60.8
62.4
-16.6
77.4
25.4
31.7
37.0
41.3
47.4
51.4
57.3
60.7
62.4
S
S-S
S-SS
S
S / SSI
s-Ss
S/
S-S
\S
\S-S'
S
OH
S
OH
148
Chapter 7: A thermochemical database for organosulfur compounds
Radical Species
H 3C-S'
SH
HS-S'
S
HCOS
S
S S
HS
S'
H 2C-SH
CHSH
AfH0
So
[kcaVmnol]
298 K
29.1
33.5
23.6
[caV(nmol K)]
298 K
57.1
45.9
60.9
300 K
9.6
6.9
9.5
400 K
10.9
7.0
10.2
500 K
12.3
7.0
10.7
C0
[caV(mol K)]
1000 K
600 K
800 K
13.6
15.7
17.3
7.6
7.1
7.4
11.9
12.3
11.2
1500 K
19.9
8.1
13.0
2000 K
21.3
8.4
13.3
2400 K
22.0
8.6
13.5
17.0
70.9
14.4
16.3
18.0
19.5
22.0
23.9
26.7
28.2
28.9
67.9
51.8
89.7
56.4
64.5
58.9
8.8
13.2
12.8
9.4
15.7
14.3
10.0
17.8
15.4
10.5
19.6
16.2
11.2
22.4
17.2
11.8
24.4
17.9
12.6
27.4
19.0
13.1
29.0
19.7
13.3
29.7
20.0
63.0
76.6
23.7
30.2
35.9
40.7
47.8
52.5
58.7
62.0
63.7
53.8
64.6
11.5
12.9
14.1
15.1
16.4
17.2
18.3
18.9
19.2
24.8
37.4
73.8
62.5
14.4
13.0
15.1
14.2
15.7
15.5
16.1
16.6
16.8
18.5
17.3
19.8
17.9
21.8
18.3
22.9
18.4
23.3
31.6
73.2
16.9
19.6
22.2
24.6
28.4
31.3
35.6
37.8
38.8
44.3
45.3
CHSH
47.3
76.5
20.4
24.1
27.4
30.2
34.5
37.6
42.1
CHSH
91.7
74.4
20.0
22.6
24.6
26.2
28.5
30.2
32.9
34.3
35.0
63.2
88.3
31.1
39.0
45.8
51.5
59.9
65.6
73.5
77.3
79.0
45.1
75.0
19.2
22.7
25.3
27.3
30.0
31.7
34.0
35.3
35.8
39.5
77.4
16.5
18.0
19.4
20.8
23.0
24.5
26.7
27.8
28.3
25.0
80.6
21.9
25.6
29.4
32.9
38.7
43.0
49.5
52.8
54.4
40.0
85.6
26.0
30.8
35.2
39.0
45.0
49.5
56.1
59.4
61.0
85.1
84.1
24.7
28.4
31.6
34.4
38.7
41.9
46.7
49.2
50.4
37.0
82.6
23.9
28.6
32.6
35.8
40.6
43.9
48.4
50.4
51.3
33.6
86.5
22.0
24.4
26.9
29.3
33.2
36.2
40.6
42.8
43.9
72.5
67.9
15.4
17.7
19.4
20.8
23.0
24.5
27.1
28.5
29.1
16.7
18.4
21.3
23.4
26.5
28.1
28.8
CH
S
CHH
HS'CHSH
SH
SH
C. SH
O
H
SH
SH
'CSH
CIs
S
S
S
S
S
58.3
66.5
12.8
14.7
25.8
68.0
16.0
19.1
21.9
24.4
28.3
31.3
36.0
38.5
39.6
20.9
75.0
21.7
26.4
30.4
33.9
39.5
43.7
50.2
53.6
5 5. 3
14.5
79.4
28.0
34.2
39.4
43.9
51.0
56.3
64.6
68.9
71.0
33.9
79.9
20.8
24.3
27.4
30.1
34.3
37.2
41.5
43.6
44.6
53.3
74.4
18.9
22.8
26.1
28.9
33.4
36.6
41.5
44.2
45.4
62.7
76.0
21.8
26.9
31.3
34.8
40.1
43.6
48.5
50.9
52.0
45.2
72.8
18.7
22.5
25.7
28.5
33.0
36.3
41.4
44.1
45.3
41.9
73.3
18.5
22.1
25.3
28.2
32.9
36.3
41.4
44.1
45.3
42.1
73.0
18.0
21.7
25.1
28.1
32.8
36.2
41.4
44.1
45.3
20.3
84.8
19.1
21.1
22.9
24.5
27.0
28.9
31.7
33.1
33.8
S
S
S S
149
Chapter 7: A thermochemical database for organosulfur compounds
ArH
[kcaVnol]
298 K
so
[caI/(nol K)]
298 K
300 K
400 K
500 K
56.9
86.7
29.1
37.0
43.8
49.5
58.1
S
56.4
85.4
29.0
36.9
43.7
49.5
l-SCH2
34.1
71.2
18.2
21.0
23.4
37.8
84.2
23.6
26.4
41.8
84.5
24.8
28.9
80.7
22.2
25.0
91.0
18.6
S' d
O
CH 2
HSO 'CH
OH
2
Cp0
[caV(mol K)]
800 K
600 K
1000 K
1500 K
2000 K
2400 K.
64.1
72.8
77.1
79.1
58.1
64.1
72.8
77.1
79.1
25.5
29.0
31.7
36.1
38.5
39.6
28.9
31.1
34.5
37.1
40.9
42.9
43.8
27.5
29.7
31.7
34.9
37.5
41.5
43.7
44.8
26.1
29.8
33.2
38.8
42.9
49.3
52.7
54.3
28.1
33.2
38.2
42.7
50.0
55.4
63.7
67.9
69.9
99.6
33.5
39.6
45.6
51.1
60.3
67.1
77.6
82.9
85.5
73.7
103.2
34.9
40.4
45.7
50.6
58.7
64.8
74.1
79.0
81.3
54.6
77.5
18.4
21.5
24.6
27.5
32.1
35.6
40.6
43.2
44.4
-7.2
67.9
13.8
16.2
18.2
19.9
22.4
24.1
26.7
28.1
28.9
-14.2
75.1
20.0
23.8
26.9
29.5
33.6
36.5
41.0
43.3
44.5
-17.8
83.9
25.8
30.8
35.2
39.0
44.8
49.1
55.4
58.6
60.2
-14.4
86.7
28.3
33.5
37.5
40.8
45.7
49.2
54.8
57.8
59.4
-6.3
88.1
29.4
34.1
37.8
40.9
45.7
49.2
54.8
57.9
59.5
7.8
64.2
11.2
12.4
13.5
14.4
15.8
16.7
18.1
18.8
19.1
-4.2
72.8
15.9
18.6
21.0
23.1
26.3
28.6
32.0
33.8
34.6
-7.6
82.6
22.0
25.6
28.9
32.0
36.9
40.6
46.1
48.9
50.2
HO
OH
-
$
S.>
-
S
OH
OH
"SH
OH
SH
O
S'
*
S
18.4
110.2
40.0
47.6
54.9
61.4
72.2
80.1
92.1
98.2
101.2
41.9
73.3
18.5
22.1
25.3
28.2
32.9
36.3
41.4
44.1
45.3
-10.8
78.4
22.8
26.1
28.9
31.1
34.7
37.3
41.2
43.1
44.1
2.0
74.5
20.7
24.6
27.7
30.4
34.4
37.3
41.6
43.9
44.9
OH
SH
6
',SH
Chapter 8: Conclusions
Chapter 8: Conclusions
150
150
Chapter 8: Conclusions
8.1 Summary
This thesis has provided an important step toward the detailed modeling of complicated fuel
mixtures that include sulfur compounds. Where it was previously only possible to build
speculative mechanisms manually, reasonably accurate kinetic mechanisms can automatically be
generated for a variety of systems involving organic sulfur compounds. The already-extensive
database for organosulfur chemistry has been expanded to include the kinetics of reactions
involving both sulfur and water, and the thermochemical database has been expanded to make it
possible to model oxidative sulfur chemistry. As more accurate rate estimation schemes continue
to take hold in the chemistry community, the scope and accuracy of these methods is sure to
increase.
The sulfur database in RMG has been validated by a variety of experimental studies. Accurate
pyrolysis models have been generated for t-butyl sulfide, making this the largest sulfur
compound for which quantitatively accurate pyrolysis models are available. RMG was able to
identify the pyrolysis and supercritical water desulfurization mechanisms of hexyl sulfide,
confirming water's role as a desulfurization agent as well as an inhibitor in the formation of
Chapter 8: Conclusions
151
aromatics in the high-temperature treatment of crude oils. Automatic mechanism generation has
also proven useful in geochemical modeling, allowing us to study the effects of sulfur
compounds on phenyldodecane decomposition, at timescales far beyond those that can be
studied experimentally.
8.2
Recommendations for future work
As automatic mechanism generation continues to expand to modeling new, industrially relevant
systems, accurate kinetic models will be desired for larger and larger compounds. These
molecules provide many more opportunities for bonds to be broken, hydrogen atoms to be
abstracted, and cycles to be formed; all of this can be quite taxing on the available memory.
Mixtures of multiple large species compound the problem further, causing RMG's memory
requirement to greatly exceed what is available. Even models for relatively light compounds in
crude oil, such as hexyl sulfide, should predict the formation of larger aromatic species in certain
conditions: as there is no single "coke" molecule, a pyrolysis system may produce millions of
different large molecules in concentrations too small to be considered individually. However, the
sum of these species can affect the overall chemistry greatly, and improvements in RMG are
necessary to model these systems accurately.
"Pruning" (removing some edge products from a model when a flux does not reach a certain
cutoff point) should allow RMG to include more important products, but changes in how the
structure and reactivity of large species are represented in RMG will let us build models that
capture the most information possible. One representation, which has previously been used in
coal modeling, consists of modeling a large organic molecule as a group of unreactive nodes
(consisting of fused aromatic rings) linked by reactive alkyl chains that may also contain oxygen
or sulfur. In this treatment, the exact geometry of the nodes is unimportant, and reactions only
need to be considered for the linking chains. Additional lumping can be done if necessary, and
the memory saved should allow for the quantitative modeling of more complicated chemistries
than what has been investigated previously.
As the modeling of larger and larger species becomes feasible, these compounds are likely to
remain in the liquid phase, even at the high temperatures frequently studied in pyrolysis and
.
....
Chapter 8: Conclusions
152
combustion chemistry. Significant efforts have already been undertaken to facilitate mechanism
generation for these systems,' and the addition of sulfur groups to these solvation models will
allow us to generate reasonable reaction mechanisms for processes involving large sulfur
molecules in the liquid phase.
As future improvements in automated reaction mechanism generation schemes and ab initio
methods make it possible to model the kinetics of larger and larger molecules, the role of
uncertainty in these models will continue to increase. Errors in thermochemical property
calculations tend to be approximately proportional to molecule size, 2 and as these uncertainties
can be mitigated with additivity schemes, the importance of accurate experimental data for
benchmarking purposes becomes paramount. This problem has largely been solved for
compounds containing carbon, hydrogen, oxygen, and nitrogen (among others) by the Active
Thermochemical Tables, 3 but the problem of high experimental uncertainties remains for sulfur
compounds. An accurate experimental database for sulfur chemistry will greatly broaden the
horizons for the accurate modeling of organosulfur systems, whether they are petrochemical,
geological, environmental, or biological.
8.3
References
1.
2.
A. Jalan, R. H. West and W. H. Green, J. Phys. Chem. B, 2013, 117, 2955-2970.
G. A. Petersson, D. K. Malick, W. G. Wilson, J. W. Ochterski, J. A. Montgomery and M.
J. Frisch, Journalof Chemical Physics, 1998, 109, 10570-10579.
B. Ruscic, R. E. Pinzon, M. L. Morton, G. von Laszevski, S. J. Bittner, S. G. Nijsure, K.
A. Amin, M. Minkoff and A. F. Wagner, J. Phys. Chem. A, 2004, 108, 9979-9997.
3.
Download