Selective Oxidation of Methane in Sulfuric Acid: Understanding and Improving Catalyst Activity, Stability, and Selectivity in the Periana System Dissertation zur Erlangung des Grades eines Doktors der Naturwissenschaften Vorgelegt der Fakultät der Chemie und Biochemie der Ruhr-Universität Bochum von Tobias Zimmermann geboren am 01.02.1986 in Pforzheim Bochum, 2015
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Selective Oxidation of Methane in
Sulfuric Acid: Understanding and
Improving Catalyst Activity, Stability,
and Selectivity in the Periana System
Dissertation
zur Erlangung des Grades eines Doktors der Naturwissenschaften
Vorgelegt der Fakultät der Chemie und Biochemie der
Ruhr-Universität Bochum
von
Tobias Zimmermann
geboren am 01.02.1986 in Pforzheim
Bochum, 2015
Die vorliegende Arbeit wurde von August 2011 bis Oktober 2015 am
Max-Planck-Institut für Kohlenforschung in Mülheim an der Ruhr unter der Leitung
von Prof. Dr. Ferdi Schüth angefertigt.
Referent: Prof. Dr. Ferdi Schüth
Koreferent: Prof. Dr. Roland A. Fischer
Das ist leider geil – muss ich sagen!
Strietzel, 1990
I
ACKNOWLEDGEMENTS
Up to now I have been very lucky and always been supported, developed and
promoted by senior persons who were wiser and more experienced than I. These
acknowledgements are for these mentors, friends and persons who influenced me and
this thesis.
Vielen Dank an meine Eltern Bernd und Friedlinde sowie meine Geschwister Conny
und Matze für die Unterstützung von klein auf. Danke Mama und Papa, dass ihr mich
auch unterstützt, wenn ihr selbst eine andere Entscheidung treffen würdet, wie zum
Beispiel mit der Chemie, die aber nicht so gefährlich ist wie ihr denkt.
Vielen Dank an meine Freunde aus Schule und Sport, insbesondere an Cle, Lolle, Juli,
Schimpanze Jonas und eure Familien, für die gemeinsame Zeit und die damit
verbundene Erholung von den Strapazen der Wissenschaft (wenn sie mal nicht so
will/wollte wie ich ).
Thanks to the TU München and its teaching staff for the excellent education.
Especially thanks to the groups of Ueli Heiz and Matthias Arenz. Thanks Ueli for
your versatile support during my studies, the master thesis, and the time after.
Thanks a lot Sebastian for getting me into AK Heiz and your scientific influence on
me during my master studies. Also, it has always been a pleasure to spent leisure with
you and to learn from you. The same goes to Tschurli, one of the persons why you
should love Austria.
Great influence for the choice of my PhD topic had Mirza Cokoja, as well as
Professor Herrmann and Professor Kühn. In their lectures I heard for the first time
from the Periana system which was also the topic of the oral examination. Also thanks
for your support in getting a stipend.
Thanks to my friend from university, Lennart, for the legendary days in Munic and
Mülheim, the good team work in our undergraduate lab courses, and the knowledge
that I can rely on you.
Danke Wolfgang und Uschi, meine garchinger “Adoptiveltern”, dass ihr euch immer
liebevoll um mich gekümmert habt und ich immer wusste, dass jemand da ist, den ich
um Hilfe fragen kann.
Coming closer to people who were directly involved in this thesis, I would like to
thank a long list of persons: Danke an alle, die in irgendeiner Form Verbindung zum
Max Planck Institut für Kohlenforschung haben. Insbesondere danke an diejenigen,
denen ich während meiner Zeit hier begegnet bin und die hier arbeiten.
I am especially grateful to my Doktorvater (supervisor) Ferdi for the scientific
freedom I had, the trust in me, and all the opportunities I was offered. I am glad that I
was allowed to do mistakes, and I always enjoyed your stories. To make a long story
short: I always had the feeling that I could come with any problem to you.
Thanks to the examination board, especially to Professor Roland Fischer, for agreeing
on being my second examinator.
Another important person of my methane oxidation adventure is Mario: I think Team
Methan is just awesome! Also thanks for the friendship and the good evenings we
had. In this context I also have to mention Kommune MH and Caro Columbina: If I
will ever share a flat it will be with you (or Schimpanze).
Thanks Teo that I could learn from you in the acetylene project. It is good to know
that I have friends like you, I can rely on.
Thank you, Wolfgang and Claudia that you are there in case something goes terribly
wrong, your reassuring advice, and the nice evenings.
II
Danke an meine Mädels Kirsten, Annette und Sarah: Mütterliche Fürsorge und etwas
Gegacker ist durch nichts zu ersetzen!
Thanks to all group members for the pleasant time. Also thanks to the former group
members who gave me a warm welcome, wherever I met them (Joker you are the
kindest person on earth). It is nice to be part of the big Schüth family. I always felt at
home and the right place!
Thanks to JP, Tim, Stefano, Felix, and Andre for going climbing, sailing, running, and
playing squash with me, and to Niki for the hard party experiences.
Thanks Gonzalo that you are never tired of answering my questions! It was a pleasure
to meet such an excellent scientist.
Thanks to my office mates Jorro aka JoJo der Reifendealer, ManoGangsta, Ivy, Dr.
Pille, Julia, Hequing, boss of the entire agent Wang clan, and Nico, an expert on
pseudo plastic extrusion.
Vielen Dank allen Technikern und Mitarbeitern der analytischen Abteilungen für alle
Currently, petrochemistry uses predominantly oil and natural gas as raw materials to
produce aromatics, lower olefins, and synthesis gas (Syngas), i.e., a mixture of
hydrogen, H2, and carbon oxides, COx, important building blocks for many base
chemicals, e.g., methanol. An important route for the production of olefins is steam
cracking where a feed of longer chain saturated hydrocarbons is thermally cracked in
the presence of steam into a stream, containing shorter chain unsaturated
hydrocarbons. Steam reforming (SMR) on the other hand is one method for the
production of Syngas where steam reacts with hydrocarbons on a solid catalyst to
form a mixture of H2 and CO. Although it is possible to run the same type of process
with different raw materials, natural gas, respectively its main component methane, is
mainly used for producing Syngas, whereas natural gas liquids (NGLs), i.e., a mixture
of ethane, propane, butane and pentane, and naphtha, a fraction of crude oil, is used
for steam cracking.1
However, the raw material basis for industrial organic chemistry has not always been
like that and this basis changed several times in history. At the beginning of organic
chemical industry, coal and tar distillation were the entry into organic synthesis until
acetylene chemistry, still based on coal, took over, after the turn of the last century.
With the end of the Second World War petroleum resources became cheap and
abundant, which paved the way for modern chemical industry, and olefins to replace
acetylene.1,2
Syngas based production of chemicals, e.g., methanol, was available
already at an earlier time in history, but usually based on coal gasification, and SMR
was introduced in 1930 in the US, and became important after 1962 with the
introduction of ICI´s tubular reforming under pressure.3
Selective oxidation of methane in sulfuric acid
2
Oil and gas majors4 and the International Energy Agency (IEA)
5 expect that even
though the share of renewable energy will continue to increase, fossil fuels will
remain the main source of energy until 2035 (figure 1a). More importantly, the share
of natural gas as primary energy source is anticipated to grow, in contrast to oil and
coal (figure 1b). Consumption in absolute numbers will increase, due to economic
growth and increasing population on earth.
Since the chemical industry is tied to the energy sector, which is by volume
significantly larger, and both use the same raw materials, organic chemical industry
will likely undergo another, albeit slower and smoother, transition of the basic
resources used for chemical production and follow the energy sector. As a result,
growing importance of natural gas and methane for production of chemicals can be
foreseen.
Figure 1: a): Energy production by source. b): Share of primary energy sources of total energy
consumption. Based on reference 4.
Indeed, exploitation of shale gas can be seen as one driver for the increasing
importance of natural gas, and first consequences of the shale gas boom in North
America on the chemical industry are already visible.6-8
Major chemical companies
announced or started to install new steam cracking capacities in the US to make use of
cheap NGLs produced by hydraulic fracturing of shale deposits.9,10
The increasing
share of NGLs as feed for steam crackers and the different product spectrum
compared to naphtha feeds will, as a consequence, lead to less production of
propylene by steam crackers. Thus, additional capacity for “on-purpose” propylene
production, e.g., by the methanol-to-propylene (MTP) process, which was already
tested in a demonstration plant in the 1980´s, could be needed and also be installed.11
a b
Introduction
3
This means a change in the raw material situation can boost commercialized
processesa.
Also the availability of a cheap feedstock, as it is currently the case for natural gas in
the US, might lead to commercialization of new processes. The 2008 founded
company Siluria Technologies started up a demonstration plant for production of
ethylene by oxidative coupling of methane (OCM) earlier this year, the first plant of
its kind in history.12-15
Even if the scenario expected by the IEA and others does not come true, and a
groundbreaking discovery makes hydrogen available from renewable resources,
methane and methanol will remain important for industrial organic chemistry, as they
represent versatile C1 building blocks.
The roles of methane and methanol in a changing raw material situation are
summarized and supplemented in the following three statements11
:
- Production and importance of natural gas, respectively methane, will
continuously increase over the next decades.
- Straightforward manufacture of high value chemicals (HVCs) from methane is
a topic of great interest and importance.
- Methanol is an important platform molecule which can be used as a fuel, a
source for producing oxygenates, e.g., formaldehyde, or a source for olefins
and aromatics, via MTP/MTO and methanol-to-aromatics (MTA).
Consequently, aside from improvements in the established Syngas based multi-step
processes, alternative and ideally direct ways to convert methane into HVCs should
be established.16
A direct route with less process steps comprises one possibility to
increase efficiency and consume less energy and raw material, a major goal in
chemical industry.
Any selective oxidation of methane into oxygenates is such an alternative, and the
investigation of such a reaction, i.e., the selective oxidation of methane to methyl
bisulfate (MBS), a methanol derivative, in concentrated/fuming sulfuric acid is the
scope of this thesis.
aThe first world scale MTP complex started operation in 2010 in China. It is, however, based on coal
gasification and the drivers are different in this case, i.e., not cheap shale gas and a shift in the
ethylene/propylene production ratio but cheap coal and vital economic growth. Thus, China is building
capacity for MTP as well as methanol-to-olefin (MTO) to produce ethylene.
Selective oxidation of methane in sulfuric acid
4
1.2 Scope of this thesis
As is evident from the previous section, the selective oxidation of methane,
respectively any saturated hydrocarbon,17,18
is an obvious dream reaction for
chemists.19
Compared to Syngas such a type of reaction has the inherent problem that
the desired HVC is an intermediate in a complex reaction network, and consecutive
reactions take place, leading to undesired total oxidation to carbon dioxide. This issue
can be overcome by several tricks, each of them having usually also a drawback. One
of these tricks is protection of the intermediate, the desired HVC, with a protecting
group. This concept has been used in the selective oxidation of methane to methanol
in solution, based on C-H activation with metal complexes.
Methane oxidation with metal complexes has originally been achieved with the so-
called Shilov system which operates in aqueous hydrochloric acid and leads directly
to methanol.20-22
It was shown that methane and methanol have approximately the
same intrinsic reactivity towards the catalyst23-25
which is for direct methane
conversion already quite good.
Subsequent improvements in selectivity were achieved by the group of Roy A.
Periana which transferred this chemistry into concentrated/fuming sulfuric acid
thereby giving not directly methanol, but methyl bisulfate which was estimated to
have - compared to methane - approximately 100 times lower reactivity to the
catalyst. This system was actually developed at a company, Catalytica Inc., and is
known as the Catalytica or Periana system.26-30
However, one drawback is the molecular catalyst used, posing problems on separation
and recycling of the same. This issue had been approached in previous studies by the
development of a solid catalyst, mimicking the molecular one, in our group at the
Max Planck Institut für Kohlenforschung in Mülheim an der Ruhr, Germany.31
This
first generation solid catalyst was established to be indeed a solid counterpart of the
molecular one.32
Second33
and third generation solid catalysts were developed mainly
by my predecessor Mario Soorholtz.32
The original proposal for my thesis was to modify the third generation solid catalyst
that it could be used in a continuous way, in contrast to its development in batch
reactors. However, first experiments on the stability of a modified third generation
Introduction
5
solid catalyst resulted in questions regarding the fundamental understanding of the
system, notably relevant for both molecular and solid catalysts.
Therefore, the focus of my thesis was changed towards the fundamental
understanding of the molecular system covering the three aspects, relevant to describe
a catalyst, i.e., activity, stability and selectivity.34
Investigation of these aspects
comprises the main part of the thesis.
For the successful implementation of a catalyst all three parameters need to be
addressed and balanced, which is not always simple, due to their interplay. As an
example, OCM has been known since 198235
but rapid deactivation of the catalysts
due to the high temperatures has been a showstopper for this reaction although
activity and selectivity were acceptable. At the heart of Siluria´s OCM technology,
now tested in a demonstration plant, is a dedicated catalyst which tackles the problem
of high temperatures and rapid deactivation. In this case a more active catalyst allows
for lower operational temperatures, and consequently longer catalyst lifetime, and
most likely also higher selectivity are achieved.12-15
It should be noted that Mario Soorholtz and I worked more than one year together in
the laboratory on this project and part of the developments are a result of this team
effort. Consequently, there are aspects in our theses which overlap. Also, part of this
thesis is in preparation for publication and word-for-word matching with the
manuscripts exists.
The thesis is structured in the following way: After a brief overview about industrial
production of methanol and comparison with possible direct conversion routes for
methane, including C-H activation and the Periana system, the main difficulties with
respect to experiments are given, as well as, a short summary of Mario Soorholtz´s
thesis, 32
. Then the macroscopic modification of the third generation solid catalyst is
discussed, together with stability tests which lead to questions on the system in
general. These questions are answered in the next section, discussing the influence of
sulfur trioxide concentration and catalyst concentration on activity, stability, and
selectivity. This part is mainly in preparation for publication, but has been
complemented by additional results. The summary and outlook section comments
critically on the achievements made, evaluates the Periana system, or in a broader
sense the selective oxidation of methane to methyl bisulfate in concentrated/fuming
sulfuric acid, and guides the work for further improvement.
Selective oxidation of methane in sulfuric acid
6
2 Conversion processes for methane
Around 10% of the produced methane is used for production of chemicals. More than
90% of this fraction is utilized via intermediate production of Syngas, and other direct
uses, e.g., production of halogenated methanes, prussic acid, acetylene and carbon
disulfide, represent minor applications. Syngas itself is the most important way for
production of hydrogen, which is used for synthesis of ammonia (51%), hydrotreating
and hydrochracking in refineries (35%), methanol production (8%), and Oxo-
synthesis.11
2.1 Syngas production and methanol synthesis
2.1.1 Syngas production
Requirements on composition and purity of Syngas differ widely, and depend on its
use. Specifications can be met by choice of feed (natural gas, naphtha, coal, etc.), type
of process (SMR, autothermal reforming, and partial oxidation), and post treatment
(water gas shift reaction). Processes for Syngas production have distinct advantages
and disadvantages, which can serve as a framework for the requirements, that direct
conversion processes should meet.
One big advantage of Syngas is obviously that it does not suffer from consecutive or
side reactions (carbon formation can be excluded if the right conditions are chosen),
and thermodynamic equilibrium controls the reaction, resulting in high selectivity.
Many direct processes instead are rather kinetically controlled.
Nevertheless, the SMR reaction (1) is endothermic, catalyzed by nickel supported on
oxides, and requires temperatures between 800 and 950 °C at 20 bar. This is achieved
by external heat supply with a considerable amount of heat-exchange operations.11
𝐶𝐻4 + 𝐻2𝑂 → 𝐶𝑂 + 3𝐻2 𝛥𝐻 > 0 (1)
In contrast, autothermal reforming (ATR) and partial oxidation (POX), irrespective if
catalytic or non-catalytic, are exothermic and combust a part of the hydrocarbon
Conversion processes for methane
7
source inside the reactor, reaction (2), respectively oxidize the hydrocarbon partially,
reaction (3). Actually, ATR shows high selectivity towards reaction (3) and thus
represents in a sense such a dream reaction, mentioned in chapter 1.2.11
𝐶𝐻4 + 2𝑂2 → 𝐶𝑂2 + 2𝐻2𝑂 𝛥𝐻 < 0 (2)
𝐶𝐻4 +1
2𝑂2 → 𝐶𝑂 + 2𝐻2 𝛥𝐻 < 0 (3)
ATR and POX do not need external heat supply, but generally pure oxygen is
required instead, usually produced by energy intensive cryogenic distillation of air in
an air separation unit (ASU). If air were used instead of oxygen, too much inert gas
would be carried through the whole process, lowering the volumetric productivity and
overall efficiency. The choice of the process depends much on factors like available
feedstock and size of the plant, but in general big plants prefer ATR and small ones
SMR.11
Consequently, in order to improve efficiency of methane conversion any direct
conversion process should ideally be (mildly) exothermic and not require energy
intensive separation or recycling steps, i.e., utilize at best air as oxidant. SMR loses
efficiency as (1) is endothermic and part of the feedstock has to be burned externally
for heating purposes. ATR loses efficiency by energy intensive cryogenic distillation
of air.11
2.1.2 Industrial methanol production
Production of methanol is rapidly growing with an annual output of 34.5 106 tons in
2005, more than 40 106 tons in 2009 and around 60 10
6 tons in 2012. Annual growth
rates of 10 to 20% are expected, mainly due to increasing production in China.11
Usually, synthesis proceeds at 220 to 280 °C and 50 to 100 bar over a copper zinc
oxide catalyst which is structurally promoted by alumina. The reaction is exothermic
and limited by thermodynamic equilibrium under these conditions which makes
recycling of unconverted Syngas necessary. The mechanism is still under debate,
however, carbon dioxide in the feed is needed for high activity and the stoichiometric
number, i.e., the ratio of hydrogen minus carbon dioxide to the sum of carbon oxides,
is slightly above 2.11
Selectivity for methanol formation is very high, i.e., greater than 99%. However, in
the overall process efficiency is lost during Syngas formation and carbon yield for
Selective oxidation of methane in sulfuric acid
8
methanol production from natural gas lies in a range of 70 to 80%a.36
Another
important benchmark is the space-time-yield (STY) as a measure of productivity
which lies for commercial processes around 1 t m-3
h-1
.
A less superficial overview on Syngas and methanol production can be found in the
literature.11
2.2 Direct conversion processes
Given the maturity of Syngas based indirect routes, direct ones hardly seem to be able
to compete,37
and the selective and direct conversion of methane to HVCs is one of
the most challenging problems in chemistry (with respect to both economy and
chemistry) and often designated as “a holy grail of chemistry”.16,38-40
Part of the
chemical challenge lies in the high standard enthalpy for homolytic bond dissociation
of the methane C-H bond, approximately 439 kJ mol-1
, and the fact that the enthalpies
of functionalized molecules like methanol are lower, 402 for the C-H and 385 kJ mol-
1 for the C-O bond
b.41
Accordingly, various approaches and numerous attempts from several branches of
chemical science exist, and the overview given here is neither exhaustive nor are the
examples comprehensively described.
One rare example of an endothermic reaction used for the direct conversion of
methane is the aromatization of methane with a MoC2/H-ZSM-5 catalyst to benzene
and hydrogen.42,43
The reaction proceeds at temperatures of 700 °C at moderate
pressure and equilibrium conversion of 12% can be obtained. Recently, also the
conversion of methane to ethylene, hydrogen and aromatics was reported over iron
exchanged zeolite at 1090 °C.44
These approaches have similar to SMR the drawback
that external heat supply is required which decreases the overall carbon yield, but
aExact values are difficult to find and vary. In case of MTO process with SAPO-34 an overall carbon
yield to HVCs of 62% is reported along with 10% coke formation. Thus, the carbon yield for Syngas
production and methanol formation is in the given range. bThis is oversimplified since H2 is readily activated although the bond dissociation energy of
436 kJ mol-1
is similar to methane. Also symmetry and position of the molecular orbitals contributing
to bond energy have to be considered.
Conversion processes for methane
9
aromatics represent interesting products, usually not formed with other direct
conversion processesa.
OCM has already been mentioned and designates the reaction of methane, at or above
atmospheric pressure, forming methyl radicals on the surface of a solid catalyst,
mostly different classes of oxide materials. The radicals desorb and couple in the gas
phase to first yield ethane, which undergoes subsequent oxidative dehydrogenation to
ethylene.35,45,46
Undesired side reactions include the reaction of methyl radicals with
dioxygen leading to carbon oxides. The reaction is exothermic and the abstracted
hydrogen atoms are oxidized to water. The high reaction temperature (above 800 °C;
claimed to be lower in the Siluria process) and the associated problem with catalyst
lifetime have already been mentioned and it remains to be answered if Siluria will
succeed.12-15,47
In any case, catalysts are known which exhibit at 30% conversion
selectivity towards C2 hydrocarbons greater than 70%, corresponding to a carbon
yield that is in the range of the benchmark.
Methane was also shown to be oxidized with oxide materials to methanol and
formaldehyde by dioxygen at temperatures ranging from 400 to 800 °C. However, at
pressures higher than 10 bar, radical side reactions prevail, and the reactor material is
catalytically active. At atmospheric pressure systematic study of catalytic materials
was possible. However, in both cases selectivity towards methanol or formaldehyde is
rarely above 50%, even though conversion is usually kept below 10%.48
Thus, the low
yields make this approach seemingly unattractive. Improvements were made by using
oxidants other than O2, especially nitrous oxide,49
but the use of any expensive
oxidant, i.e., other than air or pure oxygen, even if recycled, contributes to process
costs and allows their use only for the production of intermediates, e.g.,
cyclododecanone,50
and not base chemicals. It still shows another issue for oxidative
approaches, i.e., activation of dioxygen. Dioxygen is a diradical in a triplet state and
direct insertion into C-H bonds is spin-forbidden. Instead radical pathways with
hydrogen abstraction prevail.22,51
Consequently, several approaches in the field of direct methane conversion make use
of transfer oxidants.
aBased on thermodynamic considerations aromatization could be more carbon efficient than SMR.
Based on standard heat of formation SMR could reach a theoretical carbon yield of roughly 80%
whereas aromatization could yield 90%.
Selective oxidation of methane in sulfuric acid
10
Catalytic chlorination of methane with subsequent hydrolysis of the methyl chloride
to give methanol and HCl was suggested by George A. Olah.52,53
Amongst others,
platinum nanoparticles on alumina can be used as catalyst at temperatures from 180 to
250 °C and atmospheric pressure with 99% selectivity to methyl chloride at 36%
conversion, and in contrast to the commercial production of chloromethanes this
process does not involve radicals. Hydrolysis requires higher temperatures, and
besides methanol also dimethylether is formed. These values look promising, but the
massive corrosion encountered with Cl2/HCl and little experience with HCl oxidation
on industrial scale (Deacon process) questions the viability of such a technology.54
Current research focuses on Br2 as transfer oxidant55
and Gas Reaction Technologies
Inc., founded in 1999, tried to commercialize a methane to gasoline process based on
this chemistry together with Marathon Oil Corporation.56
No brand-new
announcements appeared recently, and the technology was sold to Reaction35 LLC
where the focus of this technology lies on converting NGLs to lower olefins and
aromatics.57
It is thus questionable if this technology will be used for converting
methane, but the technology uses a closed cycle, has a claimed carbon efficiency of
70%, and has been tested in a demonstration plant.56-60
Other examples with alternative oxidants utilize H2O2 and often precious metal based
catalysts.61-65
Reaction temperatures are usually below 100 °C, pressure ranges from
30 to 50 bar, and aqueous media are utilized. H2O2 is either supplied directly or
generated in situ by direct synthesis from H2 (which might be generated by water gas
shift reaction from CO and H2O) and O2. These systems are obviously quite selective
towards C1-oxygenates but due to the expensive oxidant (and non-productive
decomposition of it) not expected to reach commercialization. Higher temperatures
would also be desired in order to make use of the heat released by the exothermic
reaction.
In this context also carboxylation under oxidative conditions and oxidative
carbonylation of methane to acetic acid should be briefly mentioneda. This chemistry
was strongly influenced by the group of Fujiwara and is often catalyzed by
Pd(OAc)2/Cu(OAc)2 or VO(acac)2 with K2S2O8 or O2 as oxidant in trifluoro acetic
acid.66-69
Depending on the catalysts the reaction might proceed via electrophilic or
Conversion processes for methane
11
radical mechanisms.68
In case of carbon dioxide as a carbon source70
this chemistry is
for good reasons debated since the reaction is thermodynamically restricted.71,72
Thus
oxidative conditions are necessary, and several other pathways for formation of acetic
acid have been proposed.73
Some of the examples in aqueous phase are related, or can actually be seen as
biomimetic approaches, mimicking enzymes, called methane monooxygenases.
Structural elucidation of these enzymes, and mimicking of their properties is a wide
field where either only the reaction cascades are mimicked, as just exemplified for the
H2O2 mediated reaction, or the structure as well, either with metal complexes, often
porphyrine complexes,74
or solid catalysts, mostly ion exchanged zeolites.75-77
However, a big problem of truly biomimetic systems is to generate closed cycles, and
if this is realized, to achieve reasonable activity. Thus, such approaches might be seen
as least developed and far from practical application.
With the exception of aromatization and catalytic halogenation, radicals, or at least
peroxo species which readily form radicals, are involved in the direct conversion
reactions discussed so far. Since radicals are reactive intermediates, they can lead to
undesired side reactions, as in case of OCM where radicals do not only react to ethane
but also carbon oxides, or to formic acid instead of methanol in H2O2 mediated
oxidations of methane. Consequently, non-radical conversion should have an intrinsic
advantage.78,79
C-H activation with metal complexes is another way of activating methane without
formation of radicals and part of the catalytic transformation investigated in this
thesis. Some basics of C-H activation and subsequent functionalization of alkanes are
given in the following. It should be stated, however, that a clear distinction can be
difficult, and activation of alkanes via metalloradical pathways is sometimes included
in this field.
aThe authors named the formation of acetic acid from methane carboxylation, irrespective if CO or CO2
is the source of the second carbon atom. For the sake of clarity I prefer the nomenclature oxidative
carbonylation and carboxylation under oxidative conditions.
Selective oxidation of methane in sulfuric acid
12
2.3 Selective oxidation based on C-H activation
C-H activation, as it is used here, describes the insertion of a metal complex into a
carbon hydrogen bond of an alkane, forming a metal carbon bond. One of the early
discoveries of C-H activation at saturated hydrocarbons is Shilov chemistry, and this
is also one of the few examples which leads to an actual functionalization of the C-H
bond. Over the course of time, more and more examples of C-H activation have been
found, and understanding of the underlying molecular mechanisms continuously
improved. In general, C-H activation is facile, and can proceed even below room
temperature, but conditions required for actual functionalization, e.g., the presence of
oxidants or water, are incompatible with many organometallic complexes, and
examples of functionalization are relatively rare.80
The two prominent examples, where functionalization occurs, are the Shilov20
and the
Periana system.29
Both are believed to operate by electrophilic C-H activation, a term
which has mechanistic implications, i.e., release of a proton during C-H activation
which is accepted by a base, and usually observed with late transition metals in protic
media. Scheme 1 shows a generally accepted mechanism, which is a sequence of four
steps, i.e., coordination of methane leading to a σ-complex, breaking of the C-H bond
(sometimes σ-complex formation and breaking of the C-H bond are taken together as
C-H activation), oxidation and functionalization.23
Further mechanistic details are
comprehensively reviewed and not further discussed here.80-82
Scheme 1: Mechanism of Pt(II) catalyzed methane functionalization by electrophilic C-H activation.
Conversion processes for methane
13
2.3.1 Shilov system
The Shilov system was reported in 1972 and uses a mixture of Pt(II) and Pt(IV)
chloro complexes in aqueous hydrochloric acid at around 120 °C and up to 100 bar
methane, to oxidize amongst other substrates methane to methanol and methyl
chloride. Pt(II) serves thereby as catalyst whereas Pt(IV) is a stoichiometric oxidant
and not capable of C-H activation. The reaction is autocatalytic since Pt(II) is
generated by reduction of Pt(IV). Recycling of the, if stoichiometrically used,
unsuitable oxidant Pt(IV) in a Wacker oxidation like fashion showed limited success
and subsequent work aimed at fundamental understanding of the reaction.20-23
It was found that the platinum chloro species undergo speciation in solution with
changing population of different chloro complexes, depending on chloride
concentration. Most active in terms of both, H/D exchange and oxidation, is
PtCl2(H2O)2. Furthermore, the substrate scope was extended and chemoselectivity
found to favor primary over secondary and secondary over tertiary C-H bonds.23
Since features of H/D exchange by Pt(II) and functionalization in the Shilov system
are similar with respect to rate and selectivity, C-H activation (actually methane
coordination) is accepted to be the rate determining step. Only under special
conditions, i.e., low oxidant concentration and temperatures above 100 °C in acetic
acid, oxidation was found limiting.23
With respect to practical application, the use of Pt(IV) as oxidant and only few
turnovers of the catalyst due to deactivation by formation of metallic platinum,
presumably as a result of Pt(II)-CH3 oxidation by inorganic Pt(II) are clear
showstoppers. Intrinsic selectivity of the catalyst towards unfunctionalized and
functionalized C-H bonds seems to be in an acceptable range of around 3:2. Clearly,
this would not allow full conversion in one pass, but operation at 30% conversion
would still result in high selectivitya.21
Productivity in the range of 0.2 mol L-1
h-1
with a starting concentration of 15 mM
potassium tetrachloroplatinate and 50% conversion of Pt(IV) is too low, but this rate
was measured at 140 °C and rates at higher temperatures have not been reported to
my knowledge. Extrapolation to a useful rate of 20 mol L-1
h-1
would require a
aBased on figure 2 in reference 21, and assuming that overoxidation to CO2 would also require Pt(IV),
it can be concluded that at 50% oxidant conversion selectivity to methanol and methyl chloride is close
to 100% and still around 80% at full oxidant conversion.
Selective oxidation of methane in sulfuric acid
14
temperature of approximately 220 °C (with the given EA,app of 108.9 kJ mol-1
and
0.02 mol L-1
h-1
at 120 °C).21
A comprehensive overview and mechanistic details are given elsewhere.23
2.3.2 Periana/Catalytica system
In the 80´s and 90´s modifications of Shilov chemistry with different solvents and
catalysts appeared. One important modification is the work of Sen´s group which used
Pd(CH3COO)2 for stoichiometric oxidation of methane in trifluoroacetic acid to the
corresponding methyl ester.83
Subsequent work realized a catalytic transformation
with Pd(CF3COO)2 as catalyst and H2O2 as oxidant, and mentions that methanol is
more easily oxidized than the ester.84
Around the same time Moiseev´s group reported
the catalytic oxidation of methane with Co(CF3COO)3 and air in trifluoroacetic acid
and showed methyl trifluoroacetate to be stable under typical reaction conditions.85
Although it is not clear if these examples proceed by electrophilic C-H activation
(involvement of peroxo compounds!) and inefficient catalysis appears to be a bigger
issue than selectivity in the Shilov system, the concept of esterification was picked up
and extended to methane oxidation in concentrated sulfuric acid/oleum and coined
product protection by Roy A. Periana and Henry Taube at Catalytica Inc. A truly
noteworthy retro perspective of the considerations and activities leading to the
development of methane oxidation in sulfuric acid, respectively the Periana system, is
available.86
2.3.2.1 General features
The concept of the Periana system consists of three reactions, (4) to (6).27,29
Reaction
(4) is the oxidation of methane to methyl bisulfate with simultaneous reduction of
sulfur trioxide to sulfur dioxide and formation of water. Reaction (5) is the hydrolysis
of the ester, leading to methanol and regeneration of sulfuric acid. Reaction (6) is
reoxidation of sulfur dioxide which is on paper the same reaction used for large scale
production of sulfuric acid. Combination of equations (4) to (6) basically result in
equation (7): Oxidation of methane to methanol with oxygen, the desired dream
reaction. Since reactions (5) and (6) are necessary to obtain methanol it can be
questioned if this is a direct conversion at all, but increase in the overall carbon yield
could still be possible.
Conversion processes for methane
15
In any case, only reaction (4) is investigated in this thesis!
𝐶𝐻4 + 𝑆𝑂3 + 𝐻2𝑆𝑂4 → 𝐶𝐻3𝑂𝑆𝑂3𝐻 + 𝑆𝑂2 + 𝐻2𝑂 (4)
𝐶𝐻3𝑂𝑆𝑂3𝐻 + 𝐻2𝑂 → 𝐶𝐻3𝑂𝐻 + 𝐻2𝑆𝑂4 (5)
𝑆𝑂2 +1
2𝑂2 → 𝑆𝑂3 (6)
𝐶𝐻4 +1
2𝑂2 → 𝐶𝐻3𝑂𝐻 (7)
Sulfuric acid has the following four functions:
- Solvent, thus the catalyst should show good solubility in the medium
- Protective group, as mentioned in the historical review
- Reactant, i.e., oxidant, thus should be considered as such
- Cocatalyst, as discussed below for Pt(bpym)Cl2
2.3.2.2 Early catalysts – Hg(II)
The first academic report and patent about this system mention several cations which
can be used to oxidize methane to methyl bisulfate, sometimes catalytically. Usually
these cations are soft electrophiles and comprise, e.g., Tl(III), Pd(II), Pt(II), Au(III)
and Hg(II). Initial focus was on Hg(II). At 180 °C 50% methane (at 34.5 bar) is
converted within 3 h and 85% selectivity (43% yield) in 100% sulfuric acid to methyl
bisulfate with 0.1 M Hg(OSO3H)2 as catalyst (gas/liquid ratio 2.33:1). This
corresponds to volumetric productivities of 0.4 mol L-1
h-1
and a turnover frequency
(TOF) of 3.6 h-1
.26,27
There had already been indications that mercury salts in combination with sulfur
trioxide as oxidant might catalyze the selective transformation of methane, but the
previous example was a rather undefined process with pure, gaseous sulfur trioxide,
and the subsequent work of Periana´s group was clearly a milestone.87,88
Remarkably, the intermediate CH3HgOSO3H was identifieda, and based on
independent synthesis and treatment of this compound a mechanism was proposed,
consisting of electrophilic, meaning heterolytic cleavage of the C-H bond, methane
activation by mercuric bisulfate and formation of methyl mercuric bisulfate,
aTo my knowledge this is the only intermediate in the field of methane conversion in sulfuric acid
which has been undisputed spectroscopically identified!
Selective oxidation of methane in sulfuric acid
16
functionalization with sulfuric acid to give methyl bisulfate and mercurous bisulfate,
which is finally reoxidized to mercuric bisulfate.
Different mercury compounds were subsequently studied as catalyst by other groups,
but without significant improvement towards a viable process.25,89,90
2.3.2.3 Pt(bpym)Cl2 - the Periana catalyst
Although platinum salts were found to be active they were described as “poor (…)
because of the irreversible bulk-metal formation”a, and in the search for more efficient
catalysts focus was put on development of ligands that stabilize cationic platinum
species in sulfuric acid.29
This search resulted in the complex η2-(2,2´-
bipyrimidyl)dichloroplatinum(II) ((bpym)PtCl2, 1, scheme 2) which is often named
Periana/Catalytica catalyst, respectively the
combination with concentrated sulfuric acid or oleum
Periana/Catalytica system.
At 220 °C 90% of methane (at 34 bar) is converted
within 2.5 h and 81% selectivity (73% yield) in 102%
sulfuric acid to methyl bisulfate with 50 mM 1 as
catalyst (gas/liquid ratio 2.75:1). This corresponds to
volumetric productivities of 0.5 mol L-1
h-1
and a TOF
of 10 h-1
. In separate experiments volumetric
productivities of 3.6 mol L-1
h-1
and a TOF of 36 h-1
were found, and the final methyl bisulfate
concentration was 1 mol L-1
. Based on these values 1
was claimed to be more effective than mercuric salts.29
It is important to note that 1 is often designated as a highly active catalyst. However,
too low activity is one of the biggest issues for this catalyst, respectively, the system.
For industrial application, activity should be two to three orders of magnitude
higher.23,91,92
Indeed, Periana clearly points out that cis-diamminedichloroplatinum(II)
shows higher initial activity, but is transformed and precipitates with a half-life of
aIt has been found in this thesis that in case of K2PtCl4 precipitate is not bulk metal, but PtCl2.
Conversion processes for methane
17
15 min at 180 °Ca. The 2,2´-bipyrimidyl ligand leads to reduced activity, but keeps the
cationic platinum species solubilized.
Based on the fact that H/D exchange is observed at 150 °C in D2SO4 - not oleum -,
but no oxidation, it is accepted that the oxidation step is limiting above an acid
concentration of 85%.29
Subsequent reports concerning 1 and related complexes with nitrogen containing
ligands were often computational studies and aimed at mechanistic understanding.
Presumably due to the special solvent, these studies often lead to conflicting
conclusions.93-100
Without being exhaustive, the cocatalytic effect of the acid is briefly surveyed as an
example. This effect was conceptually developed for C-H activation, based on the
finding that Pt(II) catalyzed H/D exchange below acid concentrations of 80% sulfuric
acid essentially does not proceed. However, it increases continuously, if the
concentration of the acid is increased. In 96% sulfuric acid H/D exchange is around
1000 times faster compared to 80% sulfuric acid.
In analogy to alkane dehydrogenation systems, ground state stabilization was given as
explanation.91,92
Ground state stabilization describes the coordination of ligands,
stronger than the substrate, to the metal center, which inhibits the reaction. If acidity
drops in sulfuric acid, nucleophiles, e.g., methanol or water, are less and less
protonated and might coordinate to the metal center. This was substantiated by
computational methods and sounds generally reasonable, but is without clear proof,
and it has to be considered that the Shilov system operates in aqueous solution and
does obviously not face serious problems with ground state stabilization (methane
coordination is limiting under common conditions but proceeds at measurable rates).
A later computational96
study finds methane to readily replace water from the
platinum center and suggests instead acid catalyzed ligand exchange as reason, i.e.,
protonation of the leaving bisulfate ligand accelerates coordination of methane.
Unambiguous proof for this hypothesis is still missing.
Recently a new mechanistic study appeared,101
refining the original mechanistic
proposal, and addressing another important issue in Pt(II) catalyzed methane
functionalization. Pt(IV) catalyzed C-H activation is rarely observed, and, if so, under
aInitial TOF of 3.6 h
-1 at 180 °C with cis-(NH3)2PtCl2 is reported without further information about the
used conditions.
Selective oxidation of methane in sulfuric acid
18
special conditions, e.g., photochemical activation.80,100,102-104
Consequently, strongly
oxidizing conditions could lead to depletion of active Pt(II) and deactivation of the
whole system. The recent study concludes that this so-called over-oxidation
dilemma23,105
could be unproblematic, if a self-repair mechanism was in operation,
where Pt(IV) is the actual oxidant and regenerates Pt(II). Again, clear proof for that is
missing.
One relatively young direction in this field is the development of solid catalysts which
mimic 1.31
The development of solid catalysts for large volume chemicals is crucial,
since catalyst separation is more difficult with molecular ones. A proof of concept
was given with Pt-CTF, a Pt(II) modified covalent triazine framework (CTF), and
subsequent work revealed close similarities between 1 and Pt-CTF on the molecular
level.32
More solid analogues have been developed,33,106
and reference 33 has led to
important consequences for this thesis.
2.3.2.4 Further catalysts and developments
1 is perhaps the most investigated catalyst for reaction (4), and although platinum
salts were described as not promising, studies exist which make use of platinum salts,
sometimes even Pt(IV) compounds which should be even less promising in terms of
C-H activation.107,108
This is surprising, because PtCl4, which was used as catalyst, is
insoluble in sulfuric acid or oleum, but TON of approximately 800 was obtained. The
difference between studies undertaken by Periana and the just mentioned ones lies
mainly in the use of sulfuric acid or oleum.
The importance of this difference becomes evident in a study about a modified
system, i.e., the platinum catalyzed oxidative carbonylation of methane with carbon
monoxide to acetic acid.109
One series of experiments in this study was conducted
without carbon monoxide, i.e., under standard conditions to give methyl bisulfate, and
clearly shows that the amount of methyl bisulfate formed in oleum is around two
orders of magnitude higher. Additionally, this is one of the few studies where a direct
comparison between 1 and other platinum compounds was undertaken, clearly
showing 1 to be inferior to other platinum compounds.
Actually, oxidative condensation or carbonylation of methane in sulfuric acid to give
acetic acid is usually done with palladium catalysts and was first reported by Periana
with PdSO4.110,111
Additional carbon monoxide is not needed with palladium catalysts,
Conversion processes for methane
19
most likely due to formation of CO from methyl bisulfate. Again a clear influence of
sulfuric acid and oleum with respect to the formed product (methyl bisulfate or acetic
acid) as well as the amount of formed product is observed.112,113
The catalyst scope was further extended to Au(I) and Au(III) with Se(VI) as oxidant
in sulfuric acid,114
and iodine as well as iodine compounds in oleum with low sulfur
trioxide115
concentration, or high sulfur trioxide116,117
concentration. It was reported
that I2 in sulfuric acid is not active and oleum is required for the reaction.115
Furthermore, it is commented that differences in these publications with respect to
yield, reaction order, etc. are likely a result of the differences in the sulfur trioxide
concentration.116
Especially the activity in the study with 65% oleum is promising,
and the TOFs exceed for the first time in methane oxidation in sulfuric acid/oleum
clearly a value of 100 h-1
.
Summarizing selective oxidation of methane in sulfuric acid/oleum it is clear that
Periana not only invented this transformation but also contributed a lot to subsequent
developments. However, Periana usually used concentrated sulfuric acid as medium
whereas other groups more often used oleum.
Further developments targeted at ligand design to increase the rate of C-H activation,
sometimes in non-oxidizing strong acids, with other substrates, but limited
success.118,119
Two noteworthy directions are the use of ionic liquids to increase the solubility of
platinum compounds in sulfuric acid120
and methane oxidation based on C-H
activation with air as oxidant. Incorporation of air to obtain a closed cycle is
paramount for further progress in this field. Examples are the use of redox
couples,121,122
Wacker type systems,111
silica supported hybrid compounds consisting
of an analogue of 1 and polyoxometalate for activation of oxygen123
and a high-
throughput approach testing combinations of catalyst/ligand/cocatalyst
compositions.124
In some cases, methane oxidation might proceed by other pathways,
involving radicals or peroxo species.
2.3.2.5 Critical remarks on methane oxidation in sulfuric acid
Actually, a huge number of compounds has been used for oxyfunctionalization of
methane in sulfuric acid, more commonly in oleum, including V2O5,125
TiO2, Cr2O72-
,126
but also metal free compounds like K2S2O8127,128
and other peroxides129
or
Selective oxidation of methane in sulfuric acid
20
combinations like urea/H2O2/RhCl3.130
This is only a small selection and especially
patent literature has many more versions of oxyfunctionalization in sulfuric
acid/oleum.26,28,30,131-133
The feature by which these systems can be categorized is if functionalization occurs
via pathways involving radicals or not. Systems based on a free radical mechanism
usually operate below 100 °C and form methanesulfonic acid. Although formation of
carbon dioxide is not considered in reports on these systems, selectivity seems to be
comparable to the systems described beforea. Nevertheless, the actual problem in
categorizing is that methane sulfonic acid is converted in oleum to methyl bisulfate
above 150 °C, even without catalyst.134
Since the systems, claimed to operate by C-H
activation, usually work above this temperature, it is not a priori clear, if activation
proceeds by a non-radical mechanism. It could be argued that this is not important, as
also the radical systems are selective, but if we want to understand the chemistry and
rationally design a system, this is an important question.
Pt(II), Pd(II), and iodine based systems are indeed very likely to operate by a non-
radical mechanism since methane sulfonic acid has never been observed as main
product with these catalysts.130
Other halogens and mercury, however, were found to
form methanesulfonic acid, depending on the conditions, as main product. It might
even be possible that Hg(II) initiates radical reactions as well as catalyzes methane
conversion by C-H activation.25,90,127
Aside from mechanistic ambiguity another problem is evident in the field of methane
oxidation via C-H activation in sulfuric acid/oleum. Comparison of the performance
of catalysts is sometimes based on results from different publications, obtained under
different conditions. 1, for example, has been claimed to be more effective than
Hg(II).29
This might indeed hold for selectivity if methyl bisulfate is the desired
product and not methane sulfonic acid, but not necessarily for activity. Comparing
TOFs reveals that Hg(II) exhibits a TOF of 3.6 h-1
, whereas 1 shows a value of 10 to
36 h-1
. The latter value is higher, but it was obtained at 220 °C, whereas the former at
180 °C. The same applies to conversion: 90% clearly looks better than 50%, but the
gas/liquid ratios, and the sulfur trioxide concentration are different. In case of Hg(II)
100% sulfuric acid was reported, while 102% sulfuric acid was used with 1.
aProductivity lies in a comparable range to Periana´s work but does not seem to be easily increased,
since decomposition of initiator is difficult to control at higher temperature.
Conversion processes for methane
21
This might look like almost identical but in fact the nomenclature for sulfuric acid and
oleum can be misleading by referring either to mol% or wt%. To avoid any
misunderstanding only wt% is used in this thesis and corresponds in case of oleum
(where the molar ratio of sulfur trioxide to water exceeds 1, i.e., sulfur trioxide is
present in other forms than sulfuric acid, H2SO4, e.g., as H2S2O7) to the wt% of excess
sulfur trioxide present in the sulfur trioxide/sulfuric acid mixture and in case of
sulfuric acid to the wt% of H2SO4 present in the sulfuric acid/water mixture. With
respect to 102% sulfuric acid in Periana´s original work,29
the used oleum
specification was clarified in a subsequent theoretical study coauthored by Periana,96
where it is stated that 102% sulfuric acid corresponds to 9% oleuma. Literature tends
to be not very precise on this and sometimes no clear distinction is made between
sulfuric acid and oleum. Furthermore, cited conditions can differ from the conditions
in the original report.31,39,95,106
Table 1 Comparison of conditions used for testing Hg(II)27
and 129
in the oxidation of methane
parameter Hg(II) 1
solvent 300 mL 100% sulfuric acid 80 mL 9% oleum
temperature /°C 180 220
methane /mmol 640 115a
“free” SO3 /mmol 0 90
“accessible” H2SO4 /mmolb 990 260
maximum possible CCH4 /%c 77 100
experimental CCH4 /% 50 90 aIf the volume of the gas phase given in the experimental description is used for the calculation the amount of methane would be
185 mmol. bbased on the difference between 100 and 85% sulfuric acid cassuming a 2:1 stoichiometry of accessible oxidant and
methane; The exact stoichiometry with respect to the oxidant is not clear, yet.
Finally, 100% sulfuric acid corresponds to 18.7 mol L-1
H2SO4 and 9% oleum to a
solution formally containing 2.1 mol L-1
sulfur trioxide and 17.6 mol L-1
H2SO4.
Considering that methane oxidation slows down upon dilution and does not proceed
below 85% sulfuric acid (15.4 mol L-1
H2SO4) it is clear that these differences are
significant and cannot be neglected. Table 1 gives a rough estimate of the amount of
a102% sulfuric acid is obviously used to describe a molar ratio, most likely the molar ratio of SO3:H2O
which would be then 1.02:1. The conversion to wt% is likely achieved by correcting the 2 mol% excess
SO3 by the molar weights of SO3 and H2O (factor of 4.44). This procedure can be found in the internet,
but is questionable, since the reference should be the total mass of the SO3/H2SO4 mixture and not only
of the excess SO3 and H2O.
Selective oxidation of methane in sulfuric acid
22
methane and oxidant (sulfur trioxide and sulfuric acid above 85% sulfuric acid/ sulfur
trioxide alone) as well as the maximum theoretical conversion of methane for Hg(II)
and 1.
These numbers show that the conditions used in previous publications differ with
respect to the composition of the solution as well as sulfuric acid/sulfur trioxide has to
be considered as reactant and can be limiting! Consequently, use of inappropriate
amounts of oxidant and/or the particular experimental approach can obscure intrinsic
properties of the investigated catalyst.
2.4 Perspective for methane conversion
The Periana System improved several drawbacks of the Shilov system, e.g., use of a
more practical oxidant, higher selectivity, and catalyst stability. However, aside from
the inherent corrosion issues, there are two major hurdles, preventing
commercialization: Separation of methyl bisulfate from the reaction solution, which is
complicated by the large sulfuric acid to methane molar ratio,135
and, more
importantly, the catalyst activity, which seems to be two to three orders of magnitude
too low for industrial application.23,39,91
It is difficult to foresee if any of the mentioned approaches will be implemented on
commercial scale. The ones, currently tested in demonstration plants, are favored, and
the Periana system seems to lag behind. Syngas, however, can be expected to
maintain a strong position for economic reasons, e.g., depreciated plants, and the
efficiency of this mature process.37
If alternatives are implemented, it will be first for
small scale applications. Nevertheless, technologies for Syngas production and
conversion on small scale are also developed, based on once through concepts,
meaning air can be used in Syngas production and subsequent conversion does not
need recycling of unconverted Syngas.136,137
In case of methanol several concepts
exist. Liquid phase processes could be beneficial for good heat control.11,138,139
Without the need to recycle gas, it might be affordable to carry undesired ballast gas
through the reactor. These developments show that separation and processing are
important factors for production on large scale, and any possible solution has to be
analyzed comprehensively. Focus on a catalyst alone is not sufficient. Separation,
e.g., of product from the reaction mixture, is of exceptional importance.
Results and discussion
23
3 Results and discussion
3.1 Reading advice
The previous section already hints that there are still many open questions with
respect to methane oxidation in sulfuric acid. Also, comparison of the experimental
conditions, used for testing Hg(II) and 1, shows differences, and indicates that results,
reported in the literature, might be masked.
Indeed, several findings, presented in this thesis, seemingly conflict with some of the
conclusions presented in the literature survey. In order to improve readability, and to
avoid confusion, the most important findings are given here as brief notes. This
should be seen as supplement to the introduction.
- Acid and sulfur trioxide concentration have a significant impact on reactivity.
- Differences in sulfur trioxide concentration can lead to orders of magnitude
difference in catalyst activity.
- Many conclusions in literature are based on experiments in concentrated
sulfuric acid and not oleum.
- Platinum salts are not unstable catalysts, and achieve under certain conditions
high activity and stability.
- The TOF is a function of catalyst concentration, and many conclusions in
literature are based on high catalyst concentrations.
- Different catalyst concentration leads to different speciation of catalyst
precursors.
- Oxidation of methane to methyl bisulfate is part of a consecutive reaction
network which has to be kept in mind with respect to experimental approach
and selectivity.
Currently, literature generalizes from a narrow set of conditions. However, this thesis
shows that specific statements only apply to a specific set of conditions, and revisits
the whole picture, based on experiments from a broad range of conditions.
Selective oxidation of methane in sulfuric acid
24
3.2 Experimental issues and approach
From the critical discussion in chapter 2.3.2.5 it is clear that mechanistic as well as
basic experimental issues are still present and need to be solved, before textbook
understanding of the Periana system is achieved.
Since the focus of this thesis is mainly on Pt(II) catalyzed methane oxidation, radical
mechanisms are of minor concern. All experiments yielded methyl bisulfate as the
main product with carbon dioxide being a side product. Other possible products, like
methanesulfonic acid, were not detected. Small signals in 1H NMR spectra from trace
impurities were present in some cases but hard to assign. It remains elusive, although
unlikely, if methyl bisulfate is formed directly or via methanesulfonic acid as
intermediatea.
However, one important aspect is comparability of measurements and conditions as
well as description of activity: A specific conversion value after a certain time is
clearly not sufficient to describe the system. Consequently, attention was paid at the
experimental approach and obtaining values which can be compared between
different experiments.
One challenge lies in the use of a batch setup where as a rule of thumb conversion of
reactants with a reaction order of 1 should stay below 10% in order to obtain rates
which are not obscured by consumption of reactant.140
As methane and also sulfur
trioxide/sulfuric acid - likely with a reaction order greater than one - are reactants, the
conversion of both has to be limited. In case of sulfur trioxide and sulfuric acid this is
complicated by the fact that not just the total amount has to be considered but also
which part is accessible and can participate in the reaction. The experimental setup
which was developed together with Mario Soorholtz accounts for this and limits under
standard conditions (20% oleum, reaction stopped after defined pressure drop) the
conversion of methane to a maximum of 30%, but usually 20%. The conversion of
sulfur trioxide is under assumption of a 2:1 stoichiometry for sulfur trioxide/methane
comparable. Importantly, there is no transition from oleum to sulfuric acid in most
cases, since sulfur trioxide is in slight excess. A brief description of a standard
experiment is given in the following and further details can be found in the
Results and discussion
25
experimental part. Additional experimental issues are discussed where they are
relevant.
Scheme 3: Sketch of the experimental setup; taken and modified from reference 32.
In a typical experiment with the two autoclave setup, depicted in scheme 3, where one
autoclave serves as reactor and the other as reservoir for methane, 15 mL oleum (this
corresponds in case of 20% oleum to 70 mmol free sulfur trioxide) with catalyst
(typically around 600 µM) is heated under stirring (1000 rpm) to reaction
temperature, usually 215 °C, pressurized with methane to a total pressure of around
70 bar (30 mmol methane, pCH4 60 bar), which had been preheated to the same
temperature, and finally all reactions, except where otherwise stated, are stopped after
different reaction times, but at the same drop of the pressure-volume product,
indicating similar conversion, by quenching in a water bath. Figure 2 shows typical
p,T-t profiles: The pressure in the reactor increases during heat-up, due to evaporation
of sulfur trioxide, and after pressurizing with methane decreases (from a higher level)
due to conversion of methane.
aThe unexpected signals in
1H NMR spectra may originate from impurities in the capillary, added as
lock reference. The same signals are observed by measuring the capillary immersed in pure sulfuric
acid.
Selective oxidation of methane in sulfuric acid
26
Figure 2: p,T-t profiles of a typical experiment.
Lower conversions would be desirable, but reaction time and time for pressurizing
and quenching have to be accounted for, too. If possible, comparisons are only made
at identical conversion. Nevertheless, it is still a simple batch setup and improvements
can be made, albeit with modifications of hardware. All in all this procedure reflects
the catalytic activity for a given set of conditions rather than the maximum possible
yield. For the understanding of the system, both types of information are important.
Also the lowest level of acidity, at which a catalyst is still working, would be
important for the description and comparison of catalysts. It might be that a less active
catalyst tolerates more dilute conditions.
Previous work of the group did partly follow integral approaches, and results should
be seen in that context: The initial work on Pt-CTF reported TONs based on the
amount of Pt(II) used in the first reaction in five recycling experiments and showed
this TON to stay constant.31
Final methyl bisulfate concentration indicates that this
does not necessarily mean that the initial amount of Pt is stably coordinated to CTF,
and also not that the activity is constant after recycling but due to the long reaction
times the reaction essentially consumes all accessible oxidant, respectively sulfur
trioxide, and the TON is predetermined, within error margin, by the sulfur trioxide
concentration. It is expected that the fresh catalyst ends up faster in the sulfuric acid
regime, and also leached platinum species contribute to activity (compare with the
thesis of Mario Soorholtz where it was shown that a typical CTF with 15 wt%
Results and discussion
27
nitrogen content can accommodate 5 wt% platinum after recycling several times
while the platinum content of Pt-CTF reported in the initial report amounted to
approximately 40 wt%). For the recycled catalyst the reaction time is still sufficient to
reach the sulfuric acid regime, although a part of the platinum has been leached.
This problem with an integral approach was further substantiated and addressed in a
subsequent study33
where the TON of a solid catalyst in recycle experiments was
shown to increase with an integral approach if the reference for each recycle is the
actual amount of Pt used in each reaction. Essentially all reactions lead to the same
amount of methyl bisulfate along with less and less platinum used in each experiment.
Nevertheless, although the Pt content is constant at ~6 wt% for the Pt(II) modified
nitrogen containing carbon investigated, the TOF measured with the differential
approach decreases in contrast to the TON from the integral approach which shows
that deactivation other than leaching of platinum exists. The same behavior has been
observed for polyacrylonitrile derived N-containing carbon materials modified with
Pt(II) (Pt-PAN-C).32,141
This underlines the importance of the experimental approach.
However, also as a result of the dependency of the TOF on catalyst concentration,
vide infra, the new “differential” approach in batch reactors is not necessarily
sufficient to understand the complex deactivation behavior and continuous
experiments seem to be the only way for unravelling these intricate mechanisms.
Attempts to better understand the deactivation behavior of the solid catalysts are still
ongoing.
In general, previous publications should be critically revised, considering the
particular experimental approach chosen.
The stated TOFs and rates are based on the product methyl bisulfate in the reaction
mixture, determined by 1H NMR, although the TOFs calculated either by the pressure
drop or the amount of methanol determined by HPLC after hydrolysis show the same
trends and both parameters describe the catalytic activity reasonably well. The
pressure drop was used for the determination of rates in previous work33,89,116,117
but
the exact composition of the gas phase during the experiment is not determined and
remains undefineda.
aAlthough it is reasonable to use the pressure drop as measure of activity, the actual composition of the
gas phase is not known and changes in the amount of CH4, CO2, SO2 and SO3 contribute to the pressure
drop. Under the mentioned standard conditions with 30% conversion based on the amount of methyl
Selective oxidation of methane in sulfuric acid
28
In any case, this experimental approach differs from most previous reports. Until now
most reports follow the integral approach of Periana´s 1998 Science paper,29
reporting
a yield of oxidation product after a certain reaction time, usually a few hours, and not
at the same conversion.107,112
In some studies, the integral approach seems to be
appropriate because the used conditions (low sulfur trioxide concentration or low
temperature) usually lead to rates which can still be satisfactorily resolved, even if the
reaction is conducted for several hours. However, most of the experiments presented
in this thesis show rates that would not be resolved by an integral approach.
Furthermore, it is important to keep in mind that the reported TOFs, respectively
rates, in this work as well as in all previous work (including H/D exchange) are
apparent. They are significantly influenced by the changing properties of sulfuric
acid/oleum with changing sulfur trioxide concentration and different catalyst
concentrations.109
Consequently, care has to be taken in comparing results from
different publications.
3.3 Thesis of Mario Soorholtz
Since parts of Mario Soorholtz´s thesis32
were the starting point for this thesis it seems
convenient to wrap up the main conclusions from that work, which had two main
directions. The first direction was the characterization of the first generation solid
catalyst Pt-CTF, and the elucidation of its structure at the atomic level. It was
essentially found that it is adequate to call Pt-CTF a solid analogue of 1 although
subtle differences between both exist.
The second direction was the development of solids that can act as ligand for Pt(II)
and after modification with Pt(II) catalyze the oxidation of methane in oleum/sulfuric
acid. Initially the CTF material was modified by applying milder synthesis conditions,
i.e., lower synthesis temperature. This resulted in materials which were chemically
different, albeit unstable in oleum/sulfuric acid at 215 °C. Later experiments with
other nitrogen containing carbons as potential solid ligands for methane oxidation
showed that the parent material has to be treated at last at 600 °C, preferentially at
higher temperature, to avoid complete decomposition after 2.5 h in oleum at 215 °C.
bisulfate, determined by 1H NMR, the pressure drop would indicate 15% conversion if a change in the
partial pressure of methane was the only contribution.
Results and discussion
29
The higher stability might be induced by mild carbonization at or above 600 °C and
formation of graphitic domains. Nevertheless, by no means should it be concluded
that materials surviving 2.5 h at 215 °C in oleum/sulfuric acid are infinitely stable
towards the harsh conditions.
Different precursors were used to vary the atomic level (different coordination
environment) as well as the structural properties (porosity) of CTFs. Micro- and
mesoporous CTFs with different nitrogen content as well as different ratios of surface
nitrogen species were obtained and catalytic tests were conducted in batch
experiments similar to Periana´s integral approach, which was shown to be possibly
inappropriate in the previous sections. A comparison between most of these tests
seems to be reasonable because the majority shows an effective rate of reaction too
low to reach the maximum possible conversion (both, intrinsic activity and amount of
catalyst were rather low). In recycling experiments only the original Pt-CTF showed a
stable platinum loading between 4 and 5 wt%. Higher TONs, comparable to values
obtained with 1, but unstable platinum coordination were obtained if the CTF
exhibited a higher C/N ratio. Actually, comparable values were obtained in these two
cases, as conversion of oxidant was close to maximum conversion.
However, improvement of activity with unchanged stability was desired and a variety
of nitrogen containing carbons was screened for their ability as solid ligand in
methane oxidation. Interestingly, materials were identified which exhibited greater
TONs than 1 and showed increasing TONs in recycling experiments. This at first
sight peculiar behavior has just been mentioned and is a result of the experimental
approach, together with the relatively high activity of the catalysts. In subsequent
experiments less platinum was used, as part of it had been leached. In addition,
recycling resulted in additional loss of material, but the activity of the catalyst and the
rate of reaction were still sufficiently high to get close to the maximum possible
conversion. In case of 20% oleum care has to be taken if the methyl bisulfate
concentration gets higher than 1 mol L-1
(this concentration had previously been
reached with 1, and also the most active, but unstable, Pt-CTF at a platinum
concentration of 1 mM.).
This made modification of the setup mandatory. At the same time another material,
derived from lobster shells, did not only show increased activity, but also stable
platinum loading in recycling experiments. This was the first solid catalyst
Selective oxidation of methane in sulfuric acid
30
investigated with the setup described in the previous section (at this stage reactions
lasted 30 min which still does not allow comparison at identical conversion but the
TOFs were determined from a defined pressure drop at the beginning.). Remarkably,
the material was around twice as active as 1 and showed a stable platinum
coordination over five recycling experiments.33
Figure 3: Platinum loading after four recycling steps as function of the specific surface area for
different Pt-PAN-C carbonized at 1100 °C. Taken and modified from reference 32.
Unfortunately, this material, as it is derived from lobster shells, is not well suited for a
systematic investigation. Luckily, at the same time Pt-PAN-Cs were also found to be
promising candidates. These materials can chemically be modified by different
thermal treatments,142
resulting in different nitrogen content as well as different ratios
of the surface nitrogen species. Furthermore, also the pore structure can be varied, and
two synthesis routes are available: The first one is based on hard templating, where
the pore structure can be influenced by use of different silica templates. The second
one is based on post synthesis treatment of a non-porous bulk carbon by means of
water steam activation. These two routes allow for a wide variety of hierarchical
structures, including micro and mesoporous carbons, but also core shell materials.
Investigation of the Pt-PAN-C revealed that, depending on the nitrogen content and
specific surface area, a certain amount of platinum is probably stably coordinated.
Results and discussion
31
This is shown in figure 3 with a modified graph from the thesis of Mario Soorholtz,
where “stable” platinum loading corresponds to the value of apparently constant wt%
of platinum during the course of four recycling experiments in 20% oleum for 2.5 h at
215 °C.32
Furthermore, these studies corroborated what had already been evident:
- Higher nitrogen content, as a result of lower carbonization temperature, gives
materials which exhibit more sites for stable platinum coordination.
- Higher nitrogen content leads to less active materials.
- Deactivation not attributable to platinum leaching occurs, but is very difficult
to investigate due to interaction of various factors.
3.4 Solid catalysts
The PAN based catalysts looked quite promising and were the starting point of this
thesis with the original proposal to use them in a continuous manner. In order to
evaluate if they are suitable for application in a continuous setup, deactivation needs
to be better understood. Aside from deactivation mechanisms different from simple
platinum leaching, it is also important to clarify if platinum is released under reaction
conditions and formation of methyl bisulfate is catalyzed by dissolved species which
are coordinated again upon cooling the reaction mixture. To answer this question hot
filtration experiments are necessary which means that the operating catalyst is
removed from the hot reaction mixture. As oleum at 215 °C is challenging to filter,
the catalyst needs to be modified macroscopically that filtration can be done by means
of a frit or capillary. Consequently, the size of the carbon particles has to be increased
from around 10 - 100 µm to several mm.
3.4.1 Macroscopic modification
Due to the physical properties of the carbon, pelletizing by simply pressing is not
possible, and also the use of binders did not result in useful pellets. Thus, the
approach was changed from post carbonization shaping to pre-carbonization shaping.
Selective oxidation of methane in sulfuric acid
32
Two ways were pursued: Post synthesis generation of porosity by water steam
activation (WSA)a, and hard templating with a pre-shaped template.
3.4.1.1 Water steam activation
In contrast to the resulting carbon, powdered PAN can be readily pelletized or already
polymerized in a mold which allows basically shapes of any kind, obviously ideally
suited for introduction of porosity by WSA. In comparison to the synthesis protocol of
the powdered Pt-PAN-C obtained by WSA there is no big difference: Radical
polymerization slightly above room temperature, stabilization at 200 °C in air,
subsequent carbonization at 1100 °C, and finally introduction of micropores at 850 °C
by WSA. Some examples of shaped bodies (pellets, crushed material, monoliths) and
different stages during synthesis of a monolith are shown in figure 4.
Figure 4: Photograph of porous PAN-C pellets, crushed material and monoliths (lower part; from left to
right PAN monolith, stabilized PAN monolith, PAN monolith after carbonization and WSA) obtained
by WSA route.
As shown by nitrogen physisorption isotherms in figure 5 of crushed PAN-C before
and after WSA, porosity is introduced by WSA, and the resulting material is
microporous, exhibiting a type I isotherm.
aWater steam activation is oxidative treatment of carbons with steam to create micropores.
Results and discussion
33
Figure 5: Nitrogen physisorption isotherms of PAN-C before and after WSA.
BET equivalent surface area increased from 1.6 to 556 m² g-1
(ZIR-ZA-035-04;
corresponds to upper right material in figure 4). This material was modified with
potassium tetrachloroplatinate to give a formal platinum loading of 3.2 wt%. At the
time of these experiments the correlation seen in figure 4 was not known yet which
means that part of the platinum is expected to leach during methane oxidation in
concentrated sulfuric acid in the retro perspective.
Block face analysis with scanning electron microscopy/energy dispersive X-ray
analysis (SEM/EDX) mapping indicated that the bodies have a core shell structure
with a (highly) porous shell and a solid core. Figure 6 shows the electron image (left)
of a block face cut from embedded crushed material and the corresponding EDX
mapping of platinum (right). Platinum seems to be enriched at the edges of the
material.
Selective oxidation of methane in sulfuric acid
34
Figure 6: SE micrograph of crushed Pt-PAN-C-WSA (left) and corresponding Pt EDX mapping (right)
Also simply cutting some of the monoliths showed that there is a solid core. Core
shell structure has already been shown for the powdered PAN-C, obtained by WSA.32
Analysis of some carbon particles, which had fallen off during cutting the embedded
material, with SEM and scanning transmission electron microscopy (STEM) revealed
the presence of platinum nanoparticles smaller than 3 nm (figure 7).
Figure 7: Micrographs of carbon particles from cutting embedded Pt-PAN-C-WSA; from left to right:
SEM, STEM dark field, STEM bright field
Pt-PAN-C-WSA was tested for methane oxidation in a trickle bed reactor at
atmospheric pressure as well as in hot filtration under standard conditions and the
results are described and discussed in the following sections.
Nevertheless, the synthesis of the shaped materials needs to be better understood and
improved as treatment of bulky material is likely to result in different properties
compared to the powdered materials. For example, monolithic PAN-C after
stabilization and carbonization showed inhomogeneous optical appearance and the
solid core was sometimes not carbon at all but hard, glassy polymer. These
differences may result from limited access to the surface. Furthermore, thermal
treatment of the shaped PAN led to mechanical stress and often resulted in cracks.
Also the formation of platinum nanoparticles is not desired at all and also here
synthesis protocols need to be adapted.
Results and discussion
35
3.4.1.2 Hard templating
Since without changing the synthesis protocol the materials scope of PAN-C by WSA
is limited to materials which only stabilize a low amount of platinum, and mechanical
stability proved to be an issue in catalytic testing, another route was followed: Hard
templating with preshaped silica templates which offers more variability with respect
to pore structures as well as carbonization temperatures. Based on literature reports
silica monoliths were synthesized by soft templating.143
After calcination of the
resulting material, a PAN-C replica of the silica monolith was obtained, following the
standard synthesis protocol for powdered PAN-C: Impregnation of the silica with a
stabilization at 200 °C under air, carbonization at the desired temperature, and finally
leaching of the silica template. Since the silica can be molded in various ways, also
here a wide variety of bodies can be synthesized. Figure 8 shows the photograph of a
silica monolith and its carbon replica after silica leaching.
Figure 8: Photograph of a porous silica monolith and a PAN-C replica after silica leaching
The silica showed ordered mesoporosity as evidenced by low angle X-ray diffraction
(XRD) shown in figure 9 and exhibited a bimodal pore system as measured by
nitrogen physisoprtion, shown in figure 10. This was expected, since two different
structural directing agents were used for synthesis. Adsorption properties are given in
table 2.
Selective oxidation of methane in sulfuric acid
36
Figure 9: X-ray diffractogram of a crushed silica monolith
Figure 10: Nitrogen physisorption isotherms of SiO2 monoliths, the composite material after
polymerization and at various stages towards Pt-PAN-C monoliths, as well as the PAN-C monoliths.
Results and discussion
37
Table 2: Results of analysis of N2 physisorption for SiO2 monolith and PAN-C replica
parameter SiO2 PAN-C
BET /m² g-1
864 612
total pore volume at p/p0 0.97 /cm3 g
-1 1.01 0.58
mesopore range /nm (3-4), 20 (3) - 6
In qualitative agreement with the powdered PAN-C elemental analysis gave a
nitrogen content of 3.9%, if 1100 °C had been used during carbonization, and 16.4%
in case of carbonization at 650 °C.
Elemental analysis also revealed that modifying PAN-C with Pt(II) might be more
difficult than in case of powdered material. In case of Pt-PAN-C with a formal
loading of 3.8% this value varied with sampling and parts from the shell showed
higher platinum content.
Since in the meanwhile the focus of the thesis was changed towards understanding of
the molecular system, these materials were neither further analyzed nor tested for
methane oxidation.
Still, future studies should use hard templating rather than WSA, since templating has
a broader scope with respect to carbonization temperature, and PAN-Cs are less bulky
and likely closer to the powdered materials. Similar to the monoliths obtained by
WSA, magnetic stirring led to abrasion of small particles. Elaborate synthesis of the
silica template does not seem to be necessary and pelletizing of silica particles used as
template in the synthesis of powdered PAN-C seems sufficient.
3.4.2 Continuous experiments and hot filtration
Continuous experiments and hot filtration were only performed with catalysts
obtained by the WSA route.
Continuous methane oxidation was done in a trickle bed reactor at atmospheric
pressure of methane. The reactor was made of glass and the Pt-PAN-C was supported
on a glass frit. Method development and knowledge about the system was small at the
time of the experiments, and thus results were only of limited value. Anyway,
conceptually it is possible to carry out continuous methane oxidation in this way.
Mostly sulfuric acid (not oleum) was used as reactant, and traces of methyl bisulfate
Selective oxidation of methane in sulfuric acid
38
/methanol were found by NMRa. The signal was slightly bigger, if 20% oleum was
used, but not present if PAN-C was in the reactor instead of Pt-PAN-C. Qualitatively,
after around 13 h of methane oxidation at 215 °C and 200 mL of sulfuric acid had
passed over the catalyst still platinum could be found by elemental analysis in the
used catalyst (100 mg of catalyst were initially used and in the middle of the
experiment the amount was reduced to 33 mg). Consequently, these experiments
prove that methane oxidation with Pt-PAN-C is possible in a continuous fixed bed
reactor, although the use of atmospheric pressure and sulfuric acid, instead of oleum,
vide infra, led to low rates with limited practical use.
Hot filtration was carried out with a modified one autoclave batch setup: The
autoclave was equipped with two valves. One valve was used for supplying methane;
the other valve was connected to a capillary in the inside of the autoclave. The
capillary reached to the bottom which means that upon opening the valve a
sufficiently pressurized autoclave will release more than 90% of the liquid.
Preconditioning, by first immersing the catalysts in sulfuric acid, and only afterwards
in oleum, was necessary, since immediate contact of the catalyst with 20% oleum
often led to disintegration of the catalyst.
In one series, the catalyst was first used in 96% sulfuric acid for 2.5 h at 215 °C and a
total pressure of approximately 95 bar. The catalyst was recycled after cooling to
room temperature and stored in 20% oleum without washing and drying. This way of
recycling the catalyst, i.e., recovering from the reaction mixture at room temperature,
was repeated five times in 20% oleum (1st 15 mL oleum, subsequent runs with
20 mL). Subsequently, the catalyst was recycled five times by hot filtration at 215 °C,
in 20% oleum after 2.5 h of reaction. Figure 11 shows the obtained amount of methyl
bisulfate for this series in comparison with the amount of methyl bisulfate formed
under identical conditions with a similar amount of PAN-C material not modified
with Pt(II) (“blind”). Since the amount of methyl bisulfate formed in these
experiments is rather low, limitations by oxidant or methane are not a concern at all
and conducting the reaction for 2.5 h is justified. It is obvious that the amount of
methyl bisulfate formed in the 2nd
run after the first cold filtration (first experiment
aFor comparison, if 1 (2.5 – 5 mM) was used continuously in a 2:1 mixture of 96% sulfuric acid/ 20%
oleum MBS concentration was in the range of 20 mM which is around 1 order of magnitude more than
with the solid catalysts. This experiment was conducted over 5 weeks and the concentration of MBS
went through a maximum. If this was due to deactivation of the catalyst or drop in acidity is not clear.
Results and discussion
39
with oleum) is around three times higher compared to all other experiments. This is
likely due to unstably coordinated platinum (the initial platinum loading was 2.6 wt%
at BET equivalent surface area of 556 m² g-1
; figure 3 suggests that the use of 1.5%
would be more appropriate). However, subsequent cold and hot filtrations showed a
stable methyl bisulfate production, which is twice as big as the amount formed with
PAN-C in blind experiments under identical conditions. Only the last hot filtration
shows a decrease in the amount of formed methyl bisulfate, but this could already be a
result of loss of catalyst particles split off from the shaped body. This series is
promising and indicates that there seems to be no difference between hot and cold
filtration and more important that not all of the platinum is released at 215 °C and
reccordinated at room temperature. Another indication for this hypothesis is that
platinum was still detected by elemental analysis in the catalyst (0.5 wt% after three
more cold filtrations; activity in these experiments was not different from blind
experiments, but the catalyst had been washed and dried in between which could lead
to other types of deactivation). However, the activity in recycling tests could be
blurred by mass transfer limitations as the shaped catalyst has only a low external
surface area.
Figure 11: Amount of methyl bisulfate formed in 2.5 h at 215 °C in recycling experiments.
Selective oxidation of methane in sulfuric acid
40
In order to get more indication about the stability of the platinum coordination, the
activity of the filtrate, obtained by hot and cold filtration was tested with Pt-PAN-C
pellets from the same batch. In both cases 150 mg of shaped catalyst (pellets, 1 wt%
platinum, 370 m² g-1
; pellets seem to be less resistant to attrition compared to
monoliths or crushed material.) was heated in 20 mL 20% oleum with methane to
215 °C. Then, either hot filtration was performed, or the autoclave was rapidly cooled
in a water bath to room temperature, where the solution was filtered off “cold” by
opening the filtration valve.
In case of hot filtration, the filtrate activity was 0.0200 mol h-1
, compared to
0.0175 mol h-1
for cold filtration (methane was used for filtration and the filtrate
activity was corrected for the already formed amount of methyl bisulfate during heat
up in the filtration). Activity in hot filtration is higher but not significantly and it is
clear that differences during filtration, e.g., as a result of viscosity or the pressure
difference, might contribute to the difference. Actually, the sulfur trioxide partial
pressure, which is a good measure of sulfur trioxide concentration, indicated that in
case of testing the filtrate of the hot filtration experiment, the sulfur trioxide
concentration was higher. BET equivalent surface area decreased in both experiments
(to 184 m² g-1
in case of cold filtration, 198 m² g-1
in case of hot filtration), indicating
that substantial loss of catalytically active particles had happened. However,
meaningful comparison would also require testing the activity of the parent Pt-PAN-C
pellets. Such experiments were not undertaken, since these materials did generally not
show the required support stability for a meaningful complete dataset. For
comparison, if half of the platinum was leached out, one could expect a rate of
0.0713 mol h-1
(based on a measurement with 30 µM potassium tetrachloroplatinate).
100 mg of powdered Pt-PAN-C (carbonized at 1100 °C, 1wt% platinum) showed a
rate of 0.0025 mol h-1
. These values indicate that a substantial part of the platinum
might be leached under reaction conditions, since the rate of both filtrates lies in the
range of potassium tetrachloroplatinate. However, the comparison between the
powdered Pt-PAN-C and the Pt-PAN-C pellets, used for hot filtration should be taken
with caution, since they are different materials, with respect to porosity (micro vs.
meso porous and hierarchy) and surface chemistry (pellets are synthesized with an
additional WSA step).
Results and discussion
41
Hot filtration with 430 mg of monolith (330 m² g-1
, 0.8wt% platinum) and N2 showed
filtrate activity (0.062 mol h-1
), but the solid catalyst was still active afterwards
(0.025 mol h-1
), and platinum was found in the solid by elemental analysis. This
suggests, in agreement with the previous experiments, an intermediate situation,
where only part of the platinum is leached, albeit the leached species seem to be more
active.
In general, continuous methane oxidation and hot filtration experiments do not
satisfactorily answer the question, if catalytic action originates from truly
heterogeneous catalysis or if platinum is, under reaction conditions, released to the
solution and recoordinated upon cooling. The filtrate showed substantial activity,
which suggests, either leaching of platinum under reaction conditions, or support
instability and loss of catalyst particles. Based on comparison with potassium
tetrachloroplatinate and powdered Pt-PAN-C it seems reasonable that a substantial
part of the filtrate activity originates from leached platinum. Nevertheless, continuous
experiments and recycling of Pt-PAN-C after hot filtration also shows that a fraction -
although a minor one - of the activity and platinum is retained on the material. Future
experiments should improve the experimental setup and the stability of the pellets,
because mass transfer limitations and attrition/support corrosion might mask results.
3.4.3 Carbon corrosion
Deactivation of a solid catalyst can occur by means of several processes, e.g.,
poisoning of active sites, sintering, or formation of deposits. In liquid phase processes
leaching is a big concern. However, also corrosion and abrasion of support, especially
with carbon supports under oxidizing conditions, should not be neglected.144
Abrasion and/or carbon corrosion was already obvious in hot filtration experiments.
To further test these deactivation pathways, 200 mg of powdered mesoporous PAN-
C-1100 was heated to 215 °C in a 5/1 mixture of concentrated sulfuric acid and 20%
oleum. After two weeks 135 mg of carbon was recovered and the solution had
significantly darkened. However, this result is ambiguous, because the flask was
obviously leaking, and air could have got in the flask. Nevertheless, a strong smell of
sulfur dioxide was observed, indicating sulfuric acid as oxidant and not aira. In order
aThermogravimetry revealed PAN-C to be kinetically stable in air until around 400 °C.
Selective oxidation of methane in sulfuric acid
42
to remove sulfuric acid from the micropores, part of the material was subsequently
carbonized for a second time at 1100 °C and around half of the material was lost
during this step. Nitrogen physisorption and elemental analysis was conducted at all
three stages. Isotherms are shown in figure 12 and table 3 summarizes adsorption
properties before and after sulfuric acid treatment.
Figure 12: Nitrogen physisorption isotherms of PAN-C before and after sulfuric acid treatment.
Table 3: Results of analysis of N2 physisorption for PAN-C before and after sulfuric acid treatment
parameter before
treatment
after
treatment
after 2nd
carbonization
BET /m² g-1
453 323 334
total pore volume at p/p0 0.97 /cm3 g
-1 0.954 0.238 0.253
mesopore range /nm 6-8 (3 – 4) (3 – 4)
Besides oxidation of carbon as evidenced by weight loss, especially the mesoporous
structure suffers from the sulfuric acid treatment, and mesoporosity is essentially lost.
This is unlikely due to blockage of pores by sulfuric acid, because the isotherm is
hardly influenced by a second carbonization at 1100 °C, which should lead to
Results and discussion
43
evaporation of any remaining sulfuric acid. However, additional material is lost
during this second carbonization step, maybe due to loss of surface functionalities like
SO3H groups.
The structural changes are corroborated by transmission electron microscopy (TEM)
which also shows loss of mesoporous structures. Figure 13 shows typical TEM
micrographs of mesoporous PAN-C (left column) and of PAN-C after sulfuric acid
treatment (right column). Clearly, the mesoporous structures are lost upon prolonged
heating in sulfuric acid.
Figure 13: TEM micrographs of mesoporous PAN-C before (left column) and after sulfuric acid
treatment (right column) at 215 °C for two weeks.
Elemental analysis supports the hypothesis of surface functionalization and loss of
these functionalities after a second carbonization. Table 4 lists the C, N and S content
Selective oxidation of methane in sulfuric acid
44
of the materials. Upon sulfuric acid treatment a considerable amount of oxygen, the
balance of the weight, is found in the material, and the sulfur content is lower after the
second carbonization.
Table 4: C, N, S content and weight of PAN-C at the various stages of the stability test
sample C /% N /% S /% weight /mg
before treatment 79.4 3.5 n.a. 200
after treatment 63 3.1 3.7 135
after 2nd
carbonization 67 2.2 0.7 68
The stability of different materials is apparently different. PAN-C carbonized at
1100 °C is faster oxidized than CTF carbonized at 600 °C in 20% oleum, as shown by
the pressure time curve in figure 14, where 200 mg of material was heated to 215 °C.
Carbon dioxide was the majority of infrared active component in the gas phase after
completion of these experiments.
Figure 14: Pressure time curves of stability tests in 20% oleum.
Rough estimation with the ideal gas law results in around 2 mmol of carbon dioxide
in case of PAN-C. Elemental analysis shows substantial oxygen and sulfur
modification of the materials after oleum treatment.
Results and discussion
45
Another sign of support corrosion is found in liquid phase 1H NMR spectra of all
reaction solutions obtained with nitrogen containing carbons. Exemplarily the 1H
NMR spectrum of the solution from the stability test with PAN-C in 20% oleum is
shown in figure 15. The signal at 10 ppm corresponds to the acid protons, the signal at
2.9 ppm corresponds to methanesulfonic acid added as standard, and the signal at
2.5 ppm originates from the lock standard DMSO-d6. The other signals were not
assigned, except the triplet at around 5 ppm with a 1:1:1 intensity which originates
from 14
NH4+, a clear sign of nitrogen release from the support.
Figure 15: 1H NMR spectrum of the solution from the stability test with PAN-C in 20% oleum.
Consequently, there is strong indication that the carbon support is slowly (but very
likely still too fast for application in a continuous process) oxidized under reaction
conditions, and it can be questioned if a carbon support can be successful under these
conditionsa.
In conclusion nitrogen containing carbons can be used as solid ligands for methane
oxidation. However, there are several drawbacks:
aIt is also known that PAN-C can be used for the oxidative removal of SO2 from flue gases. Oxygen is
present in these cases but it is clear that the surface is not inert towards the components present under
reaction conditions.145
Selective oxidation of methane in sulfuric acid
46
- High nitrogen contents are needed to prevent platinum from leaching which is
detrimental for activity
- Activity of presumably stably coordinated platinum is low
- Stability of the support against oxidation and attrition needs to be improved
3.5 Molecular catalysts
Already during the thesis of Mario Soorholtz as well as in the hot filtration
experiments with catalysts of high platinum loading and low nitrogen content it
became evident that leached species might actually be very active and contribute to
formation of methyl bisulfate. In order to quantify how much dissolved species
contribute to catalysis in the filtrate activity tests described in the previous section, a
small amount of potassium tetrachloroplatinate was used as catalyst in 20% oleum.
This combination was not only very active, but also stable with several hundred
turnovers. This result was unexpected and different from literature where platinum
salts are described as unstable. The unexpected stability, the high activity, and the
consequences for solid catalysts (small amounts of leached species might be
responsible for the majority of the activity) led to the decision to focus on the
understanding of the molecular system before further development of the more
complex system with solid catalysts should continue.
Since the catalytic activity observed was much higher compared to literature, and one
obvious difference to the work of Periana is the sulfur trioxide content of oleum, the
starting point was a systematic investigation of this parameter.
3.5.1 Influence of sulfur trioxide concentration: 100-fold
activity increase in oleum
The main content of this section will be published in “Revisiting the Periana System”.
3.5.1.1 Periana catalyst
The dependency of the TOF of 1 on the sulfur trioxide concentration is shown in
figure 15 (TOF is given as molMBS molPt-1
h-1
; molMBS determined by means of
1H NMR; most of the values are still underestimated due to the not strictly differential
Results and discussion
47
approach and oxidation of methyl bisulfate to carbon dioxide). Compared to 96%
sulfuric acid (18 mol L-1
H2SO4), the use of ~40% oleum (9.5 mol L-1
SO3,
11.6 mol L-1
H2SO4) leads to an increase of the TOF by a factor of 100. Further
increase of the sulfur trioxide concentration decreases the TOF by a factor of 2. These
measurements were done with a catalyst concentration of 600 µM which is around
two orders of magnitude lower compared to the 50 mM of Periana´s original paper.29
The highest observed TOF50bar is ~1250 h-1
, 35 times higher than in Periana´s original
report.29
The methyl bisulfate concentration in these experiments usually reached
0.5 mol L-1
, only a factor of two lower. Consequently, with two orders of magnitude
lower catalyst loading, but the right sulfur trioxide concentration, a similar volumetric
productivity compared to Periana´s original report is reached.29
The TOFs in figures
16 and 17, vide infra, are extrapolated to a partial pressure of 50 bar methane, the
reaction order of which is assumed to be 1 (the raw data are given in figures 18 and 19
and a detailed description of the extrapolation is described in the experimental part).
Figure 16: Dependency of TOF for methane oxidation with 1 on sulfur trioxide concentration at
215 °C. TOF based on methyl bisulfate formed, 15 mL solution, [1] 600 µM, CCH4 < 30%, TOFs
extrapolated to methane partial pressure of 50 bar (see experimental part for details, actual partial
pressures at 215 °C 35-75 bar).
Selective oxidation of methane in sulfuric acid
48
The partial pressure of methane is calculated based on the amount of methane
introduced to the reactor. Extrapolation does not change the general picture of the raw
data but seems to be necessary as the observed partial pressure of sulfur trioxide at
215 °C differs drastically for different sulfur trioxide concentrations (~1 bar for 100%
sulfuric acid, up to 60 bar for 65% oleum), and it is thus difficult to precisely control
the partial pressure of methane introduced to start the reaction.
3.5.1.2 Potassium tetrachloroplatinate
Due to the strong influence of the sulfur trioxide concentration, also potassium
tetrachloroplatinate was investigated as catalyst. In general, simple Pt compounds are
known to be even more active than 1a, but believed to deactivate after few turnovers,
respectively after a short time, upon formation of polymeric Pt compounds, e.g.,
platinum(II) chloride, or “Pt black” (based on standard electrode potentials Pt(0) does
not seem to be a pathway for deactivation in the Periana system and formation of
Pt(0) seems only possible under less oxidizing conditions, e.g., Shilov conditions.).29
However, closer inspection of the patent,30
which served as the basis for Periana´s
original report,29
reveals that e.g., (NH3)2PtCl2 shows a better performance than 1 in
some instances (c.f. table 2 ibidem). Indeed, at low catalyst concentrations
deactivation of potassium tetrachloroplatinate after few turnovers is not observed,
and, as illustrated in figure 17, a similar behavior with respect to sulfur trioxide
concentration compared to 1 is found, albeit more pronounced, overall with higher
activity over the whole sulfur trioxide concentration investigated. At the optimum
sulfur trioxide concentration the TOF50bar is more than 20 times higher compared to 1,
i.e., ~27000 h-1
. Such a value is to my knowledge far bigger than anything else
reported for the direct conversion of methane75
and clearly in the range of large scale
industrial homogenously catalyzed processes.
aSimilar to the comparison between Hg(II) and 1, a TOF of 3.6 h
-1 at 180 °C is given for (NH3)2PtCl2
without further specification of the used conditions.29
Results and discussion
49
Figure 17: Dependency of TOF for methane oxidation with K2PtCl4 (1 shown for comparison) on
sulfur trioxide concentration at 215 °C. TOF based on methyl bisulfate, 15 mL solution, [K2PtCl4]
600 µM, CCH4 < 30%, TOFs extrapolated to methane partial pressure of 50 bar (see experimental part
for details, actual partial pressures at 215 °C 35-75 bar).
3.5.1.3 Pressure dependency
Figure 18 shows the raw data of both sulfur trioxide concentration series. As
mentioned, the general trend does not depend on extrapolation.
Figure 18: Raw data of the sulfur trioxide dependency series for a): 1 and b): K2PtCl4 (1 shown for
comparison). The pressure of methane in the preheating autoclave was kept constant at approximately
53 bar at room temperature and the partial pressure in the reactor thus changes.
a b
Selective oxidation of methane in sulfuric acid
50
To exclude that methane has a substantially different reaction order than one, pressure
dependency series were measured in 96% sulfuric acid, 20% oleum and 65% oleum
for both catalysts. The results are shown in figure 19. The partial pressure of methane
at reaction temperature (x-axes) is again calculated based on the amount of methane
introduced into the reactora.
Most of the series show within error margin a linear correlation between the partial
pressure and the TOF. In 96% sulfuric acid with 1 as catalyst (figure 19a) the TOF
seems to become saturated with increasing pressure. However, the TOF is based on
methyl bisulfate found in the crude reaction mixture, and the carbon dioxide formed
in the experiment with the highest pressure is higher compared to the other
experiments of this series. This illustrates the problem of underestimation of TOF if
rMBS is low in general. It might be in this case that the rate of methyl bisulfate
decomposition already equals rMBS.
In case of the pressure dependent experiments in 96% sulfuric acid the conversion of
methane was below 5%. In 20% oleum conversion ranged from 15 to 40% and in 65%
oleum it was ~10% in case of 1 and ~30% for potassium tetrachloroplatinate. Due to
the differences in conversion these results should be seen qualitatively and a true
order of reaction should not be extracted. Due to the interdependencies of the
parameters it is difficult to perform a clean and distinct pressure series. To solve this
problem, the experimental setup has to be further modified, so that operation under
constant pressure is possible, i.e., methane should be continuously fed into the reactor
and the conversion of sulfur trioxide limited. It seems still acceptable to use a reaction
order of one for extrapolation in the sulfur trioxide concentration series.
aA table with the values of figure 19 is in the appendix and shows that the calculated partial pressure
might have a considerable error. The values in oleum cannot be compared because the partial pressure
of SO3 contributes significantly. In case of sulfuric acid, however, the calculated partial pressure is
higher compared to the measured total pressure. This is unexpected since the volume of the gas phase
should be, due to expansion of the liquid phase, smaller than assumed for the calculations.
Results and discussion
51
Figure 19: Pressure dependency of the TOF for a): 1 in 96% sulfuric acid, b): K2PtCl4 in 96% sulfuric
acid, c): 1 in 20% and 65% oleum, and d): K2PtCl4 in 20% and 65% oleum at 215 °C with a catalyst
concentration of 600 µM in 15 mL sulfuric acid/oleum. The ranges of methane partial pressure are
different due to the increasingly higher sulfur trioxide partial pressures limiting the experimentally
accessible window. The total pressure, the calculated methane partial pressure, TOF and temperature of
the respective experiments can be found in the appendix.
3.5.1.4 Catalyst stability
With simple platinum salts, deactivation has always been described as a problem in
literature. However, at a catalyst concentration of 600 µM the reaction solution
remains clear, and, more importantly, the TONs usually exhibit values greater than
500 (both for 1 and potassium tetrachloroplatinate). Also in 96% sulfuric acid a TON
greater than 150 was observed for potassium tetrachloroplatinate (The TON of 1 is
only 18 at identical reaction time due to the lower activity. Although TON is usually
related to stability it should be noted that a low TON can also result from a stable
catalyst with low activity combined with a “short” reaction time.) As for cis-
(NH3)2PtCl2 a half-life of 15 min at 180 °C was reported,29
and some of the reaction
times presented for potassium tetrachloroplatinate are below this value, additional
experiments were performed with potassium tetrachloroplatinate where TON > 16000
was observed (40 µM, 20% oleum, 2 h). Titration of the solution after reaction yields
a b
c d
Selective oxidation of methane in sulfuric acid
52
a concentration of sulfur trioxide equivalents of 19.1 mol L-1
. Although this value
corresponds to oleum, the contributions from methyl bisulfate and sulfur dioxide to
the titrated acidity are not known, and an unambiguous conclusion is not possible.
Also, the pressure vs. time curve indicated activity over the whole reaction time. The
pressure drop significantly decreases over the course of the reaction which, however,
can be expected due to the consumption of sulfur trioxide and the shown dependency
of activity on sulfur trioxide concentration. Consequently, the quoted TON can be
considered as lower limit. It can be expected that with a constant sulfur trioxide
concentration the TON would be significantly higher.
Furthermore, keeping the reactor for 5 h at reaction temperature (this is two orders of
magnitude longer compared to the shortest reaction time, and 20 times longer
compared to the reported half-life of (NH3)2PtCl2 at 180 °C) before pressurizing with
methane shows the same activity within experimental error (320 µM, 20% oleum,
TOFwithout preheating 25540 h-1
, TOF5 h preheating 27670 h-1
).
In order to prove that deactivation is in general not a problem for simple Pt
compounds and not only particularly for potassium tetrachloroplatinate, several
platinum compounds were briefly investigated, including cis-(NH3)2PtCl2 and
platinum(II) chloride. The compounds tested in 20% oleum are listed in table 5. It is
remarkable that all of the platinum containing precursors exhibit a decent stability
with a TON greater than 500, and deactivation does not seem to be a considerable
issue. Furthermore, the activity of all tested compounds is at least one order of
magnitude higher compared to 1 and additional bpym (5 extra equivalents, entry 2)
reduces the activity of 1 by a factor of 2. This decrease of activity can perhaps also
explain the previously observed lower activity of the solid analogue Pt-CTF compared
to 1: Whereas 1 has a 4:1 molar ratio of N:Pt, addition of 5 extra equivalents of bpym
leads to a ratio of 24:1 which is much closer to a 5 wt% Pt-CTF which has
approximately a 42:1 ratioa.
The differences in selectivity are not necessarily intrinsic, and selectivity will be
discussed in detail later. It is interesting to note that the system shows high reactivity,
irrespective of whether the oxidation state of platinum in the precursor is 0, II or IV. It
is important to realize that the higher activity in oleum compensates the lower catalyst
aA sound comparison should be made based on the amount of surface nitrogen. However, this was not
attempted.
Results and discussion
53
concentration and the volumetric productivity is even higher compared to the original
report (4 vs. 11 mol L-1
h-1
).29
Table 5: Performance of platinum compounds for methane oxidation in 20% oleum at 215 °C
entry catalyst
precursor
concentration
/µM
TOF
/h-1
TONa
SMBS
/%
rMBS
/mol h-1
1 1 600 1281 651 96 0.0115
2 1/bpym 600/3000 558 764 91 0.0050
3 cis-(NH3)2PtCl2 600 19201 651 98 0.1728
4 PtCl2 600 15214 769 97 0.1373
5 (bipy)PtCl2b
600 14399 840 97 0.1296
6 Pt black 680 20094 837 98 0.2060
7 Pt(acac)2 680 22532 883 97 0.2292
8 K2PtCl4
600 22998 945 98 0.2050
9 K2PtCl4
670 23246 768 99 0.2352
10 K2PtCl6
670 24011 774 98 0.2421
Reaction conditions: 15 mL 20% oleum, CCH4 < 30%, pCH4,215 °C ~65 bar, ptotal,215 °C ~72 bar
adifferences in TON are a result of slightly different conversion (20-30%) and catalyst concentrations bbipy stands for 2,2´-
bipyridyl and should not be confused with bpym.
3.5.1.5 Methane oxidation with pure sulfur trioxide
Before discussing the results presented so far in more detail, a brief discussion of
methane functionalization with pure sulfur trioxide seems appropriate at this point. A
patent of Dow, using amongst other materials Pd modified Nafion as catalyst,146
and
patents from the 1950´s, utilizing mercury sulfate as catalyst,87,88
have reported the
selective oxidation of methane with pure sulfur trioxide. In order to test potassium
tetrachloroplatinate as possible catalyst, it was loaded together with liquid sulfur
trioxide in an autoclave, heated to 215 °C and pressurized with preheated methane.
Some activity was observed, but it appeared to be rather low compared to methane
oxidation in oleum. The TOF based on the pressure drop is approximately 160 h-1
(figure 20). This value is obtained with the ideal gas law under the assumption of the
stoichiometry in equation (8):
𝐶𝐻4 + 2𝑆𝑂3 → 𝑀𝐵𝑆 + 𝑆𝑂2 (8)
This means three gas molecules would be transformed into a liquid and a gas
molecule. Consequently, per bar methane converted the pressure would drop by two
Selective oxidation of methane in sulfuric acid
54
bar. The IR spectrum of the gas phase after cool down revealed the presence of
methane, carbon dioxide and sulfur dioxide. Sulfur trioxide is difficult to assign as
absorption by sulfur trioxide overlaps with absorption bands of sulfur dioxide.
Figure 20: Pressure vs. time curve for methane oxidation with pure sulfur trioxide as oxidant and
K2PtCl4 as catalyst at 215 °C.
After opening the reactor, a few drops of a strongly green liquid were seen which
could be collected by rinsing with concentrated sulfuric acid. A 1H NMR of the
resulting solution was measured (figure 21). The intense signal at 10.4 ppm originates
from the acid protons. At 3.5 ppm a broad, presumably due to leached metal cations
from the reactor, signal in the range expected for the methyl resonance of methyl
bisulfate is found. In contrast the quintet at 2.5 ppm (residual protons of DMSO-d6)
and the singlet at 3.3 ppm (H2O) are sharp, as DMSO-d6 is added in a capillary to the
NMR tube and thus has no contact to the cations.
Results and discussion
55
Figure 21: a): NMR spectrum of the sulfuric acid rinse of methane oxidation with K2PtCl4 at 215 °C
and pure sulfur trioxide as oxidant b): extension from 2 to 4 ppm.
It is yet unclear, if the lower activity can be explained by the fact that the solid
potassium tetrachloroplatinate exhibits only a low surface area and a solid catalyst
with atomically dispersed platinum31,33
could lead to a higher TOF. Despite the
a
b
Selective oxidation of methane in sulfuric acid
56
analytical issues it remains to be answered if a process utilizing pure sulfur trioxide
could add any advantage. Possibly separation might be easier. A pure gas phase
process seems appealing, but will not be easy to realize, as moisture will be present in
any case, e.g., as a result from over oxidation of methane, which would react to
sulfuric acid with the sulfur trioxide.
3.5.1.6 Discussion of sulfur trioxide concentration series
Two key points of these results need a thorough discussion: First why potassium
tetrachloroplatinate and other platinum compounds are significantly more active than
1 over the whole sulfur trioxide concentration range investigated, and second why
both catalysts show a strong and at first sight unusual dependency on sulfur trioxide
concentration.
Regarding the higher activity of potassium tetrachloroplatinate it has already been
mentioned that simple Pt compounds are known to be more active than 1,29
but no
direct comparison has been presented up to now showing the extreme difference in
activity. The higher activity is usually traced back to the ease of oxidation of the
different compounds and the compound which is more easily oxidized shows higher
activity. Obviously, bpym exhibits π-acceptor properties which makes oxidation
difficult, respectively stabilizes the oxidation state (II). However, the actual surprise is
the stability of these materials which disproves the current standpoint of the scientific
community. Anyway, there has been already clear evidence for the stability of simple
platinum compounds in literature.107
A reason for the misunderstanding with respect
to catalyst stability over the last years might be the lack of details regarding the
influence of process parameters like sulfur trioxide or catalyst concentration and
especially their interplay. Later the reason for this misunderstanding will become
more obvious when the influence of the catalyst concentration on the rate of methyl
bisulfate formation is discussed. A crucial point is the much lower catalyst
concentration of the present study in combination with the individual solubility of the
compounds used herein, and compensation effects caused by the tremendous
differences in activity in oleum and sulfuric acid.
3.5.1.6.1 Differences between sulfuric acid and oleum – oxidation potential
The dependency of the activity on the sulfur trioxide concentration of 1 and potassium
tetrachloroplatinate is similar. Therefore it seems reasonable to assume the same
Results and discussion
57
factors being causal. The merit of the quasi differential approach is to resolve these
features (fuming of the solutions after reaction reveal that experiments with 20%
oleum applying the standard protocol remain in the oleum region; titrations indicate
the same but are ambiguous as mentioned).
An integral approach would likely show a monotonic increase in the yield with
increasing sulfur trioxide concentration, provided a constant gas to liquid ratio,
methane in excess, and sufficient reaction time. I.e., 65% oleum should lead to a
higher final methyl bisulfate concentration and higher yield compared to 40% oleum,
where the maximum rate is observed. The aforementioned studies on related systems
indicate such a behavior.89,107,109,112,113,116,117,125
A comprehensive overview of these
systems is given elsewhere, but the conditions used in the individual studies should be
evaluated carefully.147
Similar dependencies of activity on the composition of the reaction medium have
been observed for the oxidative carbonylation of methane in CF3COOH/H2O. They
were explained by speciation of the rhodium catalyst due to protonation of OH-
ligands depending on proton activity.148
Furthermore, bases can participate in
oxidation reactions and influence the rate.22
Aromatic nitration in sulfuric acid/oleum
is such an example where it has been found that the rate of nitration is a function of
the sulfur trioxide concentration with a maximum at 92 - 94% sulfuric acid.149
In this
context the concentration of a proton acceptor and hence the dissociation properties of
sulfuric acid were found to be important. However, another obvious explanation could
be a change of the oxidation potential with change in sulfur trioxide concentration. It
is reasonable that the rate is correlated with the oxidation potential. This is known for
the Wacker oxidation of ethylene to acetaldehyde, where the rate of oxidation in the
two step process (oxidation of ethylene and regeneration of catalyst in separate
reactors) is not stationary and changes with the potential of the solution.150
The latter seems to be an obvious explanation for the increase in activity up to around
40% oleum. Oleum is known to be a stronger oxidant and shows a higher oxidation
potential compared to sulfuric acid.151-153
As oxidation is believed to be rate limiting
an increase of the oxidation potential could lead to a higher rate. It can be expected
that the reaction pathway does not change with sulfur trioxide concentration. Thus, it
seems appropriate to assume a linear activation energy – reaction energy relationship
Selective oxidation of methane in sulfuric acid
58
like the Brønsted-Evans-Polanyi relation. Gibbs free energy, ΔG, and potential, E, are
linked by equation (9) which suggests proportionality between rate and potential.
∆𝐺 = −𝑛𝐹𝐸 (9)
Indeed, the resemblance of TOF and oxidation potential at the transition from sulfuric
acid to oleum is intriguing. At this point the TOF as well as the oxidation potential
increase drastically, in case of the potential by 0.6 Va.152
A similar change can be
observed for the Hammett parameter in this region.153,154
This is not surprising, since
the Nernst equation (10) couples proton activity and potential if protons participate in
the redox reaction, which is often the case. It can be expected that reduction of sulfur
trioxide and concomitant formation of water during methane oxidation involves the
transfer of protons. However, in the context of oxidations the redox potential seems to
be more appropriate for a discussion instead of proton activity.
𝐸 = 𝐸0 +𝑅𝑇
𝑧𝐹𝑙𝑛
𝑎(𝑂𝑥)
𝑎(𝑅𝑒𝑑) (10)
3.5.1.6.2 Decrease in activity above 40% oleum - solubility of methane
One might be tempted, to interpret the decrease above 40% oleum as a consequence
of the over-oxidation dilemma (see explanation of the over-oxidation dilemma in the
introduction). However, this has not been proven experimentally for the Periana
system, yet. Moreover, the same holds for the oxidation state of the active species. As
both catalyst precursors, 1 and potassium tetrachloroplatinate, show the same
dependency, the properties of the solvent should also be considered here for possible
explanations of the decrease of the TOF for high sulfur trioxide concentration.
As mentioned, the availability of a base could be crucial. In the proposed mechanisms
HSO4- appears usually and the self-dissociation of sulfuric acid/oleum changes over
the whole concentration range.155
Consequently, the concentration of the bisulfate
anion continuously changes. In order to increase the concentration of the bisulfate
anion, respectively the hydrogen disulfate anion, potassium bisulfate or potassium
disulfate was added to the reaction mixture. Adding potassium salt to 65% oleum at a
concentration of 1.6 mol L-1
resulted in a completely unexpected outcome: The TOF
Results and discussion
59
Table 6: Influence of salt addition by salting out on methane oxidation
entry salt, [K+, NH4
+] /M [SO3] /M
a TOF /h
-1 TOFwith/TOFwithout /%
65% oleum
blind testb
1 no addition, 0 23.55 144 2.1
1
2 no addition, 0 23.55 330 100
3 K2S2O7, 1.6 21.43 184 55.8
K2PtCl4
4 no addition, 0 23.55 6742 100
5 KHSO4, 0.8 21.96 1891 28.0
6 K2S2O7, 1.6 21.02 755 11.2
7 (NH4)2SO4, 1.6 21.70 813 12.1
20% oleum
blind testb
8 no addition, 0 20.20 53 0.23
K2PtCl4
9 no addition, 0 20.20 22998 100
10 KHSO4, 0.8 19.86 16886 73.4
11 KHSO4, 1.6 19.18 9300 40.4
12 K2S2O7, 0.8 19.89 18911 82.2
13 K2S2O7, 1.6 19.45 11958 52.0
14 (NH4)2SO4, 0.8 19.70 19528 84.9
15 (NH4)2SO4, 1.6 19.10 10011 43.5 aThis concentration corresponds to the titrated amount of sulfur trioxide equivalents. The sulfur trioxide partial pressure of
experiments with salt addition was slightly lower compared to the experiments without salt addition which indicates slight
dilution in agreement with the titrations. However, both traits agree with the range of sulfur trioxide concentration discussed.
bTOFs of blind tests were calculated as if 600 µM of catalyst was present. Blind test in 65% oleum was conducted for 5 min (as
for the measurements in 65% oleum with K2PtCl4) and 2.5 h in 20% oleum. The experiment with 1 in 65% oleum was done for
0.5 h and thus the blind test is not directly comparable. As the difference between blind activity and catalytic activity is more
significant for K2PtCl4 this catalyst was used for most of the investigations. In 65% oleum much higher blind activity is observed
which is not fully clear yet. Measurements in 65% oleum are more challenging due to temperature increase and material stability
issues especially for Teflon parts which might release more active contaminations in 65% oleum (not clear if due to the higher
sulfur trioxide concentration or temperature effects).
aTo assess the exact redox potential in oleum is particularly challenging. By measuring the potential of
a concentration cell I could qualitatively confirm that the oxidation potential increases with increasing
SO3 content. However, I was unable to determine and reproduce exact values.
Selective oxidation of methane in sulfuric acid
60
of 1 in 65% oleum decreased by approximately a factor of 2. This phenomenon was
further investigated with potassium tetrachloroplatinate (600 µM) in 65% and 20%
oleum with different concentrations of K+ and NH4
+. The results are summarized in
table 6, and it is obvious that addition of substantial amounts of salt decreases the
TOF. Although the decrease of TOF is unexpected at first sight, it is known that the
solubility of nonpolar substances is decreased upon addition of salt to the solution (in
case of methane-water at similar temperature and pressure, the addition of 1 M NaCl
reduces the solubility of methane by around 20%.156,157
). This effect of salting out
decreases the anyhow low concentration of methane further, and consequently also
decreases the rate of methyl bisulfate formation.
These observations and a publication that shows that the solubility of methane in
aqueous sulfuric acid is strongly dependent on the concentration of the acid158
evidence that the decrease of TOF at high sulfur trioxide concentrations is most
probably due to a lower solubility of methane in oleum with a high sulfur trioxide
content, respectively the TOF is in general convoluted by the concentration of
methane in solution. This implies that not only added ions can lead to salting out, but
also self-dissociation or changes in speciation of the medium. Interestingly, the
maximum of the TOF occurs close to 44.9% oleum where the medium has the formal
composition of H2S2O7 and a turning point for several properties.155
It seems unlikely
that the decrease can be explained by a reaction order of methane different from one
and differences in partial pressure (see figure 20): Comparison of the data from the
pressure dependency measurements in 20 and 65% oleum supports a lower solubility
of methane in 65% oleum, where the TOF values are lower, respectively level off
towards higher pressure.
A closer look shows that addition of 1.6 mol L-1
of K+ or NH4
+ to 65% oleum (entries
6 and 7) decreases the TOF by a factor of ten compared to the TOF without salt
addition (entry 4). In 20% oleum the same amount of salt decreases the TOF only by a
factor of 2 to 3 and the TOF is still higher than in 65% oleum without salt addition
(entries 11, 13 and 15). In addition to the bigger effect of salt addition in 65% oleum
it has to be considered that addition of salt dilutes the medium and it can be expected
that the concentration of sulfur trioxide, thus also acidity and oxidation potential, are
reduced. Therefore, if salting out was not the reason for the decrease, and, e.g., the
oxidation potential instead, the dilution by salt addition to 65% oleum should increase
Results and discussion
61
the TOF because the starting point of the reaction is moved from the very right side of
figures 16 and 17 towards the maximum of the TOF which is obviously not the case
(the titrated sulfur trioxide concentration of entries 6 and 7 is close to the optimum).
For 20% oleum a decrease is expected anyway because the reaction proceeds on the
left side of the maximum in figures 15 and 16, and dilution by salt addition moves the
starting point of the reaction further to lower sulfur trioxide concentrations. In
conclusion these experiments show that solubility of methane seems to be a bigger
issue in 65% oleum than in 20%. In any case, for the understanding of this system it is
necessary to consider all physicochemical properties of the solvent, and not only
acidity.
Summarizing, the dependency of the activity on the composition of the medium can
be explained by redox and solubility properties. Both factors have already been
mentioned in a publication by the group of Bell for the carbonylation of methane to
acetic acid with Pd2+
in sulfuric acid.112
3.5.1.6.3 Comparison of different catalyst precursors
The lower activity of 1 compared to the other platinum compounds is likely a
consequence of differences in the ease of oxidation, as has been investigated
computationally for cis-(NH3)2PtCl2 and 1.93,99
As the oxidation step is assumed to be
rate limiting in the Periana system, a compound, respectively an intermediate of the
catalytic cycle, which is more easily oxidized should show higher activitya.
The main difference between 1 and the other Pt(II) compounds is the bpym ligand.
Whereas ammine and chloro ligands function as σ-donors, bpym also functions as π-
acceptor. Donors should rather stabilize Pt(IV), easing oxidation of Pt(II) to Pt(IV),
whereas acceptors stabilize Pt(II) and oxidation to Pt(IV) is more difficult.159
The
reverse step, reduction of Pt(IV) to Pt(II) has been studied in detail with the same
conclusions.160
The stabilizing effect for Pt(II) should additionally be increased by
protonation of bpym. It might be that such a picture is oversimplified and it has been
suggested in a computational study that considering bpym as a π-acceptor might not
be entirely complete and also its ability as σ-donor should be considered.
Nevertheless, the weaker σ-donor capability of bpym would still lead to the same
aIf the oxidation step was not limiting, inferior methane coordination properties would result in lower
activity. Compare to the catalyst concentration series, the model based on quasi steady state, and that
Selective oxidation of methane in sulfuric acid
62
trend in the ease of oxidation. Also protonation of bpym reduces donation capability,
but influences on the other steps of the catalytic cycle have to be considered too.93,94
Furthermore, a study on the trends in the ease of oxidation shows the unexpected
result that (tmeda)PtI2Me2 (a Pt(IV) compound with σ-donors only) is capable of
oxidizing Pt(II) ligated with α-diimine ligands.161
However, in a subsequent study it
was clearly shown that diamine complexes of Pt(II) are more easily oxidized than the
corresponding α-diimine complexes.105
Both studies mention the importance of steric
factors. Consequently, even though the various influences are complex and do not
seem to be fully understood, bpym seems to stabilize Pt(II) and makes oxidation to
Pt(IV) more difficult.
The detrimental effect of excess bpym (table 6, entry 2) in combination with the
(20 mM) < 20% oleum with 1 (20 mM). In none of the experiments methane was
observed, indicating that functionalization is irreversiblea.
aTwo publications made similar experiments: The original publication where essentially the same is
observed.29
Another publication with palladium catalysts reports the backward reaction to methane.112
Results and discussion
111
Table 14: Decomposition of NaMBS and carbon dioxide formation in blind experiments
entry conditionsa
rox, total
/mol L-1
h-1
rox, cat only
/mol L-1
h-1
rox, cat only/rox,
total
1 96% H2SO4 0.049 1
2 20% oleum 0.065 1
3 96% H2SO4, 20 mM 1 0.119 0.070 1.43
4 20% oleum, 20 mM 1 0.137 0.072 1.11
rCH4→CO2b
/mol L-1
h-1
rCH4→CO2
/rox, total,20 mM
5 blind test 96% H2SO4 0.0058 0.049
6 blind test 20% oleum 0.0072 0.053 a215 °C, 2.5 h, 15 mL, 1 M NaMBS bbased on 15 mL reaction volume excluding gas phase oxidation
These experiments show that uncatalyzed decomposition of methyl bisulfate (or
catalyzed by non Pt compounds, vide infra) significantly contributes to methyl
bisulfate oxidation, i.e., uncatalyzed decomposition proceeds with almost the same
rate as decomposition catalyzed by 20 mM of 1 in oleum (rox,cat only is obtained by
subtracting the rate of NaMBS decomposition without catalyst from the rate obtained
with catalyst addition). Furthermore, uncatalyzed decomposition is slightly faster in
oleum which might be due to the higher oxidation potential. Catalytic decomposition,
however, seems to be rather insensitive to the sulfur trioxide concentration. Finally,
entries 5 and 6 show that direct formation of carbon dioxide from methane is around
one order of magnitude slower than carbon dioxide formation via methyl bisulfate. It
cannot be excluded that carbon dioxide formed in blind experiments originates from
methyl bisulfate with an extremely low steady state concentration. Indeed, traces of
methyl bisulfate were found in oleum without addition of catalysta. Due to the various
contributions the most general description of selectivity in this system is given by
equation (22) with integral rates.
𝑆 =𝑛𝑀𝐵𝑆
𝑛𝑀𝐵𝑆+𝑛𝐶𝑂2=
𝑟𝑀𝐵𝑆−𝑟𝑜𝑥,𝑐𝑎𝑡−𝑟𝑜𝑥,𝑛𝑜𝑛𝑐𝑎𝑡
𝑟𝑀𝐵𝑆+𝑟𝐶𝐻4→𝐶𝑂2 (22)
aCatalytically active contaminations might be present. Teflon parts, e.g. the coating of stir bars, can
soak up reaction solution (with catalyst) and release it in the next experiment. Because tiny amounts of
catalytically active material show substantial activity in oleum this has to be considered and taken care
of (experiments with 30 mM I2 as catalyst changed the appearance of Teflon parts to purple. These
parts showed substantial activity without additional catalyst). Consequently, care has to be taken and
Teflon parts have to be frequently replaced. It might indeed be the case that Teflon itself shows a small
activity. If a lot of Teflon parts were inside the reactor blind activity was slightly increased.
Selective oxidation of methane in sulfuric acid
112
Thus, fitting selectivity as a function of rMBS with constant kox,apparent is partially
justified: Due to the relatively big contribution of noncatalytic decomposition, the
ratio of k1/kox increases with increasing catalyst concentration. If catalyzed methyl
bisulfate decomposition was similarly influenced as methane oxidation this would not
be valid and k1/kox would be constant. For low catalyst concentration decomposition,
can almost exclusively be attributed to noncatalytic decomposition. Anyway,
comparing the value of 0.082 mol L-1
h-1
obtained for kox,apparent from the fit in figure
44 and correcting it with a factor of four for the lower concentration of methyl
bisulfate (final concentration of methyl bisulfate around 0.5 mol L-1
) shows that these
values are in the same order of magnitude but by far not similar, a result of the rough
model.
Actually, the values of catalytic decomposition show that the concept of protection
works much better than assumed so far: Transforming the value, under assumption of
methyl bisulfate being first order in catalytic decomposition, of 0.07 mol L-1
h-1
into
TOF gives a value of 3.5 h-1
compared to the TOF for methane oxidation with 50 mM
of 1 in 20% oleum of 264 h-1
(table 8 entry 4). This is already two orders of
magnitude faster which gives - if a two orders of magnitude lower methane
concentration is assumed (10 mM) - around 10000-fold higher intrinsic reactivity
towards methane as a lower limit.
It is remarkable that decomposition in 96% sulfuric acid proceeds slower, although
methanol can be observed in NMR and a part of the ester was obviously saponified.
Methanol itself might be protonated and still sufficiently protected. Anyhow, as
decomposition of NaMBS is slower in concentrated sulfuric acid, it does not seem
likely that Kp already dropped to an alarming value in 96% sulfuric acid. Thus, if
oxidation potential and solubility of methane could be increased in concentrated
sulfuric acid it might be possible to achieve reasonable activity still with high
selectivity. The determination of Kp as well as kox as a function of sulfur trioxide
concentration, especially in more dilute acid (50 – 100%), could be an important goal
in order to define a region where methane oxidation could be viable in terms of
product protection.
3.5.4.3.1 “Uncatalyzed” methyl bisulfate decomposition and other issues
In order to test the precision and generality of the values obtained in the
decomposition experiments, the amount of carbon dioxide expected to be evolved is
Results and discussion
113
calculated based on these values and compared with the actual amount of carbon
dioxide determined experimentally. This did not prove to be very precise.
As an example: For methane oxidation in 20% oleum with 142 µM of 1 0.00188 mol
carbon dioxide were determined experimentally. The predicted amount of carbon
dioxide amounted 0.00099 mol (uncatalyzed decomposition assuming first order of
methyl bisulfate and half the final concentration as concentration is included, as well
as the amount of carbon dioxide formed in blind experiments representing direct
oxidation of methane to carbon dioxide is added; catalytic decomposition was
neglected due to the low catalyst concentration). In general this discrepancy might
partially be caused by the relatively high error of the determination of carbon dioxide
as already discussed which should rather lead to overestimation. The general picture
is retained however.
Nevertheless in order to exclude other effects, investigation of decomposition was
started with higher throughput in glass vials. Actually, the first series of experiments
(change of NaMBS concentration without catalyst in 96% sulfuric acid) raised several
new questions. Decomposition in the glass vial is roughly by a factor of 1.7 slower
compared to the Hastelloy autoclave (comparison of experiments with 1 mol L-1
NaMBS). This could possibly be explained by decomposition catalyzed by metal
cations which were leached from the autoclave material, most likely Ni the major
component of Hastelloy G35. This notion was supported by the color of solutions
after reaction (note that experiments in Hastelloy autoclaves were considered as
“uncatalyzed decomposition” up to this point). If leached cations cause
decomposition, also corrosion properties of the medium have to be considered, which
shows again that not only the dominant property of sulfuric acid/oleum, i.e., acidity,
but also other properties have to be considered. Furthermore, it was observed that the
relationship between rate and concentration of NaMBS is not linear, implying that
decomposition is not first order in NaMBS, but increases progressively with
increasing concentration (figure 49). This might be caused by decomposition via
several pathways one of which might be bimolecular via intermediate formation of
dimethyl ether. These observations reveal decomposition, respectively selectivity, to
be much more complex and less understood than expected, especially under real
conditions. Further experiments in that direction are ongoing at the moment.
Selective oxidation of methane in sulfuric acid
114
Figure 49: Concentration dependency of NaMBS decomposition in glass vials: 1 mL 96% sulfuric acid,
glass vials, no catalyst, 2.5 h, 215 °C, CNaMBS <10%. Double logarithmic plot does not give a clear
linear correlation.
The main facts, learned about the selectivity of methane oxidation in the Periana
system, in this thesis are:
- selectivity pattern can be explained by a consecutive reaction network
- methyl bisulfate decomposition is less influenced by sulfur trioxide
concentration compared to methane oxidation
- methyl bisulfate oxidation is partly catalyzed by the catalyst for methane
oxidation, but also uncatalyzed, and/or catalyzed by contaminations
- deprotection does not seem to be an issue in 96% sulfuric acid
- quantification of carbon dioxide needs to be improved
Conclusion and Outlook
115
4 Conclusion and Outlook
In conclusion, it has been shown in this thesis that macroscopically shaped platinum
modified nitrogen containing carbon materials can be used as catalysts for continuous
methane oxidation with sulfuric acid/oleum. Activity of materials with a stable
platinum loading is rather low, and carbon corrosion occurs, posing question marks on
the long term applicability of these materials under the harsh conditions. Hot filtration
experiments were not conclusive with respect to stability of the platinum
coordination, but indicated that dissolved platinum species can be tremendously
active.
Observation of the high activity depends much on the experimental approach and is
massively influenced by the sulfur trioxide concentration: In short, oleum with its
high oxidation potential is needed for high rates. It is important to consider all
properties of sulfuric acid/oleum, and not only acidity.
Simple platinum compounds, which were regarded as pretty active but unstable
catalysts so far, were found to be unprecedentedly active with TOFs greater than
20000 h-1
, likely the highest TOF ever observed for methane oxidation, and
unexpectedly stable with TONs reaching values of 16000. Comparison with the
Periana catalyst showed superior performance of simple platinum compounds under
most conditions. Deactivation by dilution of the medium, i.e., decrease of the sulfur
trioxide concentration, is seemingly the bigger issue than deactivation of the catalyst
itself. Nevertheless, solubility of most platinum compounds is low. This is part of the
reason why these compounds are often regarded as unstable. Catalyst solubility is an
issue and needs to be approached rigorously since catalyst precursors can be
transformed under reaction conditions.
Not only solubility of the catalyst is a problem but also the low solubility of methane.
This leads to a dependency of the TOF on catalyst concentration which can be
explained with a model assuming quasi steady state of a platinum methane sigma
complex. Consequently, TOFs are always apparent and need to be compared with a
reference measured under identical conditions.
Selective oxidation of methane in sulfuric acid
116
Selectivity was found to be determined by the laws of a consecutive reaction network,
and uncatalyzed decomposition of methyl bisulfate is a major pathway for undesired
side reactions. Deprotection in 96% sulfuric acid has not been found to be a serious
problem.
To underline the main improvements that have been achieved in this thesis a
comparison of the current performance with the one of Periana´s original report,29
recommended guidelines,91,92
as well as a large scale industrial process, i.e., the
Cativa process for methanol carbonylation to acetic acid189
and conventional methanol
synthesis11
is made in table 15.
Table 15: Comparison of several parameters of the original publication, guidelines, this thesis, the
Cativa process and conventional methanol synthesis.
entry parameter Periana
199829
guide-
line 91,92
this
thesis
Cativa
process189
MeOH
synthesis11
1 rvol /mol L-1
h-1 a
3.6 3.6 >15 40 20 - 30
2 [catalyst] /mM 50 1 0.6 18
3 TOF /h-1
10 - 100 3600 >25000 >2000
4 TON 1000 105-10
6 >16000
b
5 S /% 81 >90 >95 99 >99c
6 CCH4 /%
Coxidant /%
90
<10d
20
20
20 – 30
<10d
15 - 50
7 separation difficult easy difficult distillation distillation abased on liquid phase for liquid phase process bbased on separate experiment cselectivity of MeOH synthesis only; generation of
SynGas decreases overall selectivity dnot only based on free sulfur trioxide but sulfur trioxide and sulfuric acid to illustrate the
problem of recycling the oxidant, respectively the high sulfur trioxide/sulfuric acid to CH4 ratio
Finally, even though activities relevant for large scale production of bulk chemicals –
seen as the biggest issue so far - have been demonstrated, it is not expected that a
process relying on this reaction can enter the phase of commercialization soon,
because the conditions needed for high activity lead to sulfuric acid as a byproduct,
and ways to separate methyl bisulfate from concentrated sulfuric acid or oleum need
to be developed. Dilution of the reaction solution does not look attractive since
reconcentration of these streams would decrease the overall carbon yield.190
Knowledge about alkylation of isoparaffin/olefin mixtures in sulfuric acid, especially
about economics and plants in operation, might be helpful for the assessment of how
much methanol per sulfuric acid has to be produced before regeneration of spent acid
can be afforded.191
Conclusion and Outlook
117
First attempts with respect to separation were undertaken with limited success during
this thesisa, and some approaches can also be found in literature, e.g., low pressure
membrane distillation192
or transesterification.132
With respect to further improvements on the catalytic reaction, mechanistic
understanding should be further developed and clarified if the knowledge from this
thesis can be transferred to other conditions and solvents, which would allow easier
separation. It would be very desirable to increase the solubility of methane which
would be beneficial for activity, selectivity and methyl bisulfate concentration.
A pressing mechanistic question is the role of Pt(IV) in this chemistry. Clearly, this
thesis shows that the over oxidation dilemma does not seem to be a problem in this
system. If this can be attributed to a self-repair mechanism, as recently suggested by
Periana´s group, remains to be answered.101
(In-situ) spectroscopic studies are needed
to corroborate such mechanistic proposals. Due to the reasonable solubility in sulfuric
acid/oleum, 1 is a good compound for such studies. Identifying the actual oxidant,
respectively clarifying, if a high oxidation potential is sufficient, goes hand in hand
with closing the cycle, i.e., incorporation of air or oxygen. It is indeed obvious from
literature that redox couples can be used, little restrictions exist with respect to the
actual oxidant, and oxygen can be incorporated.63-65,111,121-124,180,193,194
Nevertheless, it
remains to be shown how these approaches can be combined with the high activity
conditions presented in this thesis.
The difference between sulfuric acid and oleum might be universal and can at least
partially be transferred to the manifold examples of non-radical methane oxidation in
sulfuric acid/oleum with electrophilic cations, e.g., Hg2+
,27,89
Pd2+
and precursors
thereof,26,110-113
cationic iodine115-117
and gold,114
vanadiumoxo compounds,125
and
many others.28,126
Besides platinum catalysts, iodine is a very interesting candidate for
this reaction, because it is not only very active, but also seems to operate under more
dilute conditions, at least in the presence of sulfur dioxide.
Nevertheless, understanding of the Periana system is still in its infancy and further
investigations are necessary to achieve the same level of understanding, developed for
other reactions. Thus, more sophisticated experiments are necessary to circumvent the
many interdependencies present in the system and overcome ambiguity.
aTransesterification was not successful but should be further pursued. Stripping of reaction solution
with steam indeed leads to hydrolysis of MBS as evidenced by NMR but sufficient amounts of liquid
could not be condensed in the outlet.
Selective oxidation of methane in sulfuric acid
118
5 Experimental
5.1 Catalytic testing
5.1.1 Continuous methane oxidation
For continuous methane oxidation at ambient pressure glass reactors were used. The
catalyst was supported on a glass frit and the reactor had an inner diameter of
approximately 5 mm. The oven was controlled by a thermocouple inside the catalyst
bed. sulfuric acid or oleum was supplied without preheating by a syringe pump at a
rate of 15 mL h-1
, and methane with an uncalibrated mass flow controller at 0.05
standard liters per minute. The liquid was collected at the outlet and analyzed.
For continuous experiments with 1, methane at ambient pressure was bubbled through
a mixture of 96% sulfuric acid/20% oleum (2:1) at 225 °C. Samples were taken
periodically from the solution.
In both cases the effluent gas was vented and not further analyzed.
5.1.2 Autoclaves, materials, and equipment
All autoclaves were designed in house (see appendix for technical drawing), and a
schematic drawing is shown in scheme 8.
Scheme 8: Autoclave
additional capillary for hot filtration
Experimental
119
Autoclaves for methane oxidation consist of Hastelloy G-35 (balance Ni, 33% Cr, 8%
Mo, ≤ 2% Fe, ≤ 0.6% Si, ≤ 0.5% Mn, ≤ 0.4% Al, ≤ 0.05% C) and were, depending on
corrosion, remachined from time to time. The autoclave for preheating methane, vide
infra, consists of stainless steel (1.4571).
All autoclaves are equipped with pressure meters (JUMO dTrans p30; the corrosive
conditions lead to drift, and pressure meter have to be controlled and recalibrated
from time to time) and thermocouples (type K; the metal housing was additionally
protected by a Teflon hose). Autoclaves are heated separately via external heat
supply, controlled by a previously optimized PID controller, and stirring is achieved
by magnetic stirring with a Teflon coated stir bar. Pressure and temperature are
monitored and recorded by a LabView program designed by Marjan Thomas, a
screenshot of which is shown in figure 50.
Figure 50: Screenshot of the user interface of the Lab View program taken and modified from 32.
In order to reduce contact of the reaction mixture with the reactor wall, a glass inlet
was used in all reactions.
All volumes were analyzed with a physisorption analyzer by measurement of the
pressure difference between the calibrated known volume manifold and the sum of
manifold and autoclave after opening the in house made connection to the autoclave.
Helium was used as gas, and the measurement was done at room temperature and a
pressure ranging from 0.5 to 1 bar. The ideal gas law was used for calculation.
Sealing materials consist of gold or Teflon. Teflon is used at less demanding
connections and has to be replaced more frequently. In case of 65% oleum, Teflon
Selective oxidation of methane in sulfuric acid
120
sealing proved to be inappropriate. Polychlortrifluorethylene proved to be less
suitable compared to Teflon.
Valves and connections are made of stainless steel, and were replaced when needed.
The use of Hastelloy valves resulted in prolonged valve lifetime but does not
sufficiently prevent corrosion.
5.1.3 Hot filtration
Hot filtrations were done using a modified autoclave with an additional capillary as
indicated by the arrow in scheme 5. For safety reasons the hot filtration valve
(stainless steel) could be equipped with a modified drilling machine which allowed
opening from a safe place outside the pressure proof cubicle. This is not optimal with
respect to valve lifetime since not only the hot sulfuric acid/oleum leads to increased
chemical stress, but also opening in the described way increases mechanical stress.
Opening with the drilling machine is not gentle and usually turns the valve many
times towards its endpoint. Ball valves which are easier to open have considerable
shorter life time. The released liquid was collected in a glass vessel with an opening to
the exhaust and the liquid was further analyzed and used.
5.1.4 Standard experiment
Catalytic methane oxidation was carried out in a two autoclave setup described in
section 3.1, in reference 33 and is shown in scheme 3. The two autoclaves are
connected via two valves and a short capillary (dead volume of the capillary and
valves 0.9 respectively 1.3 mL). One autoclave (54.6 mL) was exclusively filled with
methane and served as preheating reservoir. The other autoclave, named reactor, was
used for the actual reaction. Two different autoclaves were used for this purpose (the
volumes, reduced by their glass inlet and a magnetic stir bar amounted 36.2, 37.3 mL
for reactor 1 and 30.4, 32.9, 34.8 mL for reactor 2; several values are given for each
reactor corresponding to volumes after remachining).
Usually, the preheating autoclave was filled with methane to a pressure of 53 bar at
room temperature. The reactor was filled with, unless otherwise stated, 15 mL of
sulfuric acid or oleum and catalyst, respectively other substances if indicated, and
purged with Argon before closure. Both autoclaves were connected via the capillary,
Experimental
121
but still separated by valves, and placed in their heating blocks. Heating was started,
in case of the reactor together with stirring. It usually took 30 min until reaction
temperature was reached. Before pressurizing, the reactor temperature was allowed to
stabilize for ~5 min. The reaction was started by opening the valves to the preheating
autoclave for ~10 s which led to a total pressure of ~70 bar in the reactora. The
reaction was run until a certain pressure drop was reached, or, in case of slow rates,
after a certain time, mostly 2 h. The pressure drop was 8.3 bar in case of reactor 1 and
10 bar in case of reactor 2, and most experiments were performed with a drop of the
pressure volume product of approximately 180 bar mL at reaction temperature. The
reaction was stopped by removing the reactor from the heating block and quenching it
in a water bath under stirring.
The temperature and pressure time profiles of the reactor and preheating autoclave for
a typical experiment are shown in figure 3 with the heat up period (until ~1600 s)
leading to autogenous pressure increase of ~10 bar in the reactor due to evaporation of
sulfur trioxide from the solution. Upon pressurizing, the pressure of reactor and
preheating autoclave are identical, i.e., ~70 bar. The reaction, however, leads to a
pressure drop in the reactor. At the beginning of the last period (~4000 s), pressure
and temperature of both autoclaves drop rapidly due to quenching in a water bath.
After room temperature was reached, the gas phase was vented into a gas sampling
bag and part of it transferred with a gas tight syringe into the gas cell of the IR
spectrometer and the relative composition analyzed.
An aliquot (~1 mL) of the liquid phase was mixed with a weighted amount of
methane sulfonic acid (MSA) as standard and analyzed by 1H NMR (It was separately
verified that methanesulfonic acid is not formed under the used conditions.). The
amount of methyl bisulfate formed during the reaction was determined based on the
integrals of the methyl resonances of methyl bisulfate (3.6 ppm) and methanesulfonic
acid (2.9 ppm). The exact positions of the resonances depend on the acidity (lower
acidity leads to up field shifts).
Another aliquot (10 mL) of the reaction solution was added slowly to 20 mL of
chilled distilled water under stirring and heated to 90 °C for 3 h. This mixture was
afterwards analyzed by HPLC for its methanol content. In most cases hydrolysis was
aAt the beginning of this thesis the exact times were not adhered to strictly, but strong influence is not
expected.
Selective oxidation of methane in sulfuric acid
122
not complete, and only 80% of methyl bisulfate was converted to methanol. NMR of
the hydrolysate showed that the remainder stays unaffected.
In some cases an aliquot (1 mL) of the sulfuric acid/oleum was titrated after and/or
prior to reaction. To reduce the concentration of protons, 1 mL of the solution was
mixed with distilled water to give a volume of 100 mL. 3 mL of this solution were
mixed with distilled water to give a solution of 50 mL and half of this solution was
finally titrated. Due to the amount of dilution steps, unknown contributions from
methyl bisulfate and sulfur dioxide and the vigorous reaction in the first dilution step
(especially with oleum of high sulfur trioxide content), titrated sulfur trioxide
concentrations should be taken with care (in case of salt or NaMBS addition, further
ambiguity arises): E.g., the stock solution of 96% sulfuric acid (17.97 mol L-1
H2SO4)
was titrated to contain 17.83 mol L-1
H2SO4 (~95%). After addition of NaMBS to
obtain a 1 M solution titration resulted in a value of 16.54 mol L-1
(this would
correspond to ~90% sulfuric acid). Correcting for the Na+ and the fact that the methyl
bisulfate anion does not have acidic protons this would still be 17.54 mol L-1
“SO42-
“
(corresponding to ~94% sulfuric acid). This only holds, if methyl bisulfate is not
hydrolyzed during the dilution steps, and a clear statement is difficult. Anyway, the
results of the titrations correlate with the calculated sulfur trioxide concentration as
well as with the sulfur trioxide partial pressure.
The sulfur trioxide concentration of mixtures was calculated based on the volumes of
the stock solutions, disregarding density changes upon mixing with densities of the
stock solutions of 1.8355 (96% sulfuric acid), 1.9 (20% oleum) and 1.925 g mL-1
(65% oleum). Titration and fuming of the reaction solutions indicate that experiments
with 20% oleum in quasi differential mode stay in the oleum region. Rough
estimation based on an average titration value of 20.2 mol L-1
for the combined
concentration of sulfur trioxide and H2SO4 (calculated value of 20.23 mol L-1
), a
methyl bisulfate concentration of 0.5 mol L-1
and a 2:1 stoichiometry for the reaction,
indicates the same (100% sulfuric acid has a concentration of 18.67 mol L-1
).
Finally, the reactor was rinsed with acetone and water, dried overnight in a drying
oven and checked for its tightness before the next experiment.
Experimental
123
5.1.5 Methane oxidation with sulfur trioxide
After melting sulfur trioxide, 1 mL of liquid sulfur trioxide (0.025 mol) was charged
in reactor 2 together with 3.3 mg potassium tetrachloroplatinate as catalyst and heated
to 215 °C. After pressurizing with methane to a total pressure of 66 bar a pressure
drop of 11 bar within 3.5 h was noticed (figure 23). Already at the beginning of the
experiment the pressure was above 20 bar which was due to the use of a partially
damaged pressure meter. This pressure meter could not measure pressure below
20 bar any more, but the values between 20 and 100 bar agreed well with a reference
pressure meter.
A pressure drop of 11 bar corresponds to approximately 0.0045 mol methane
respectively methyl bisulfate and a TOF of 160 h-1
.
5.2 Error assessment
Due to the amount of sources contributing to experimental error, fully quantitative
assessment is difficult and error bars are not shown in the graphs. However, trends are
reproducible, although errors under different conditions are different. As a general
measure of precision, the values of six experiments under “standard conditions”
(600 µM 1, 20% oleum) are shown in table 16.
Table 16: Reproducibility of experiments with 600 µM of 1 in 20% oleum
operator TOF1H-NMR /h-1
TOFHPLC /h-1
Tobias Zimmermann 1228 987
Tobias Zimmermann 1260 996
Tobias Zimmermann 1319 974
Tobias Zimmermann 1319 1004
Mario Soorholtz n.a. 1060
Mario Soorholtz n.a. 1044
average value 1281 1011
standard deviation 39 31
These experiments were conducted over the course of one year, during which pressure
meters had to be recalibrated, autoclaves remachined, and different batches of
chemicals were used. Additionally, two operators performed the experiments, both
Selective oxidation of methane in sulfuric acid
124
reactors, and different batches of 1 were used. As mentioned before, hydrolysis is not
complete and TOFs based on HPLC are usually only 80% of the value based on
NMR. This shows a generally high precision and reproducibility. The majority of the
other experiments are done under similar conditions (20% oleum, low catalyst
concentration).
Higher errors were observed for measurements in 65% oleum and experiments with
potassium tetrachloroplatinate, especially at high catalyst concentration. Four
measurements with 50 mM of potassium tetrachloroplatinate as precursor, albeit with
different pretreatments (grinding and waiting time at 215 °C before the reactor was
pressurized) resulted in an average rMBS of 0.0654 mol h-1
(0.0414, 0.0687, 0.0905,
and 0.0609 mol h-1
, respectively) with a standard deviation of 0.0176 which
corresponds to a relative standard deviation of 27%.
5.3 Calculations
5.3.1 Determination of methane and carbon dioxide
The amount of methane introduced in the reactor was calculated based on the
measured volume of the preheating autoclave and temperatures and pressures before
heat up and after quenching. The Benedict-Webb-Rubin (BWR, equation (23), see
table 17 for substance specific constants) equation of state was used for the
calculations.195,196
The solver add in of Excel was used for numerical solution of the
BWR equation. Using the ideal gas law leads to differences of up to 10%. The amount
of methane and carbon dioxide after the reaction was calculated with the BWR
equation, too, under assumption of Dalton´s law, i.e. the sum of the partial pressures
of methane and carbon dioxide equals the total pressure. The partial pressures are
obtained from integration of IR spectra (2800-3200 cm-1
for methane and 2150-
2550 cm-1
for carbon dioxide). Apparent integral extinction coefficients were
separately determined and resulted in a 10.55 times higher integral absorption of
carbon dioxide in the indicated spectral range which was used accordingly in the
calculation of the partial pressures.
Experimental
125
𝑝 = 𝑅𝑇𝑛
𝑉+ (𝐵0𝑅𝑇 − 𝐴0 −
𝐶0
𝑇2)
𝑛2
𝑉2+ (𝑏𝑅𝑇 − 𝑎)
𝑛3
𝑉3+ 𝑎𝛼
𝑛6
𝑉6
+𝑐
𝑛2
𝑉2
𝑇2(1 + 𝛾
𝑛2
𝑉2exp (−𝛾
𝑛2
𝑉2)) (23)
Table 17: BWR constants for methane and carbon dioxide (Units in atm, L, mol and K; R 0.08207