Material DRX

Catalytic tar reforming in
biomass gasification
Tungsten bronzes and the effect of gas alkali
during tar steam reforming
OLOF FORSBERG
Supervisor: Angélica V. González, PHD
Examiner: Klas Engvall, professor
Master of Science Thesis
Royal institute of Technology, KTH
Department of Chemical Technology
Stockholm, Sweden 2014

Abstract
Gasification of biomass is today facing several problems with the high amount of tar
produced and compounds such as alkali that can harm the catalyst in catalytic tar reformation.
This is why the focus in this master thesis study was to create a catalyst for secondary tar
steam reforming. The aim was to create a catalyst that was suitable for tar steam reforming
and also evaluate the effect that alkali had on the catalyst.
A catalyst composed of 20%Bronzes –ZrO2 impregnated with nickel was prepared in this
study and characterised with XRD and BET. The bronzes consisted of K0.25WO3 and was
prepared with two different methods and analysed with XRD to see if there was some
difference in the structure and purity. Three different weight loads of nickel: 5-,10- and 15
wt% , was prepared for each catalyst that was named Method 1 and Method 2. In total six
catalysts was tested in an experimental test rig that was situated at the Royal Institute of
Technology in Stockholm. In addition a blank test was performed for comparison of the
catalytic activity.
For the experiments 1-methylnaphthalene was decided to be used as a simulated tar. The
experiments were divided into two parts where in Part 1 a S/C ratio of 4 was used and Part 2 a
S/C ratio of 6 was used. The experiments were conducted at reactor temperatures of 700 °C
and 800 °C with or without alkali aerosols. Other parameters changed in the experiments were
the catalyst load, 1-methylnaphthalene flow and the gas hourly space velocity. Results were
analysed with 4 micro gas chromatographs and solid phase adsorption.
Results from the catalyst characterisation indicated that the wanted catalyst had been prepared
however in Method 2 a higher purity of the bronzes was reached compared to Method 1. The
results from the BET analysis gave a surface area of between 40-46 m2/g for the different
catalysts.
In the experiments from Part 1 a very high gas hourly space velocity was used and the results
indicated that there was almost tar reduction compared to the blank test. In Part 2 the gas
hourly space velocity was lowered and a higher tar reduction and was obtained. One test was
also conducted at 900 °C where the highest tar reduction was obtained, almost 40 %.
From the results it could be seen tar reduction and 1-MN/naphthalene ratio was increasing
with higher temperatures and nickel loadings. Catalysts prepared from Method 2 also showed
a higher tar reduction and 1-MN/naphthalene ratio compared to Method 1which could
indicate that it was more stable and had a higher purity of the bronzes. The results from
atomic absorption spectrophotometer showed that the mass of potassium in the catalyst before
the experiment decreased between 3-29 % compared to after the experiment. From the rather
low decrease in potassium the hexagonal structure of the bronzes clearly protects the
potassium from evaporating within the bulk. Also introducing alkali aerosols had a positive
effect on the tar reduction.
3
Table of Contents
Abstract ...................................................................................................................................... 3
1 Introduction ............................................................................................................................. 6
1.1 Aim and delimitations ...................................................................................................... 7
1.2 Nomenclature.................................................................................................................... 8
2 Background ............................................................................................................................. 9
2.1 Principles of biomass gasification .................................................................................... 9
2.1.1 Tar and tar reforming .............................................................................................. 10
2.1.2 Alkali compounds and its effect on biomass gasification ....................................... 14
2.2 Catalytic tar reforming.................................................................................................... 15
2.2.1 Choice of catalyst .................................................................................................... 15
2.2.2 Catalyst preparation................................................................................................. 17
2.2.3 Instruments used for characterisation of the catalyst .............................................. 19
2.3 Analysing instruments for product gas composition ...................................................... 21
2.3.1 GC for measurement of gases, hydrocarbons and low molecular weight tar .......... 21
2.3.2 Surface ionization technique for measurement of alkali levels ............................... 21
3 Experimental section ............................................................................................................. 24
3.1 Catalyst preparation ........................................................................................................ 24
3.1.1 Preparation of support ............................................................................................. 24
3.1.2 Nickel IWI method .................................................................................................. 26
3.2 Catalyst testing ............................................................................................................... 27
3.2.1 Experimental set-up................................................................................................. 27
3.2.2 Calibration of analysing equipment ........................................................................ 28
3.2.3 Activity evaluation .................................................................................................. 28
4 Results and discussion ........................................................................................................... 32
4.1 Catalyst characterisation ................................................................................................. 32
4.1.1 BET results .............................................................................................................. 32
4.1.2 XRD Results ............................................................................................................ 32
4.2 Experimental results and discussion ............................................................................... 35
4.2.1 Part 1 ....................................................................................................................... 35
4.2.2 Blank test ................................................................................................................. 41
4.2.3 Part 2 ....................................................................................................................... 43
4.2.4 Alkali loss/gain before and after experiments in the catalyst and its effect on the
results. .............................................................................................................................. 54
4
5 Conclusion ............................................................................................................................. 57
6 Further improvements ........................................................................................................... 59
7 References ............................................................................................................................. 60
8 Appendix ............................................................................................................................... 62
Catalyst characterisation, XRD ............................................................................................ 62
5
1 Introduction
Environmental issues are of great importance today and research on renewable fuels is
steadily increasing. The continuous use of fossil fuels contributes to emissions of greenhouse
gases and the temperature increase of the planet. The recent increase in the price of fossil
fuels contributes to the possibility of using renewable fuels as an alternative energy source for
heat, electricity and transport. Other factors that benefit the use of renewable fuels are
environmental and political regulations and the continuous strive for a sustainable society.
Biomass is considered as a renewable resource and can be converted to energy from thermal,
biological and physical processes. A thermochemical process which has gained much interest
is gasification of biomass. This process offers higher energy efficiencies compared to other
thermochemical conversions such as combustion or fast pyrolysis [1].
Gasification of fossil fuels is well known and commercially available today. The gasification
of biomass faces several problems of today due to higher volatile matter content and varying
composition. One of the drawbacks is the tar production and to increase the efficiency and
make it economically feasible more research has to be spent on the downstream and
upgrading of the product gas [2].
Tar production is a large drawback in biomass gasification and this study will focus on
catalytic steam reforming of tar. In last year master thesis study a new promising catalyst was
prepared but it has not been properly tested or evaluated yet. Hence this study will continue
the research and properly test and evaluate the catalyst. The support material consists of
potassium tungstate with zirconium dioxide and the active phase is nickel. The support is
synthesised based on previous methods with the structure KxWO3 – ZrO2. Nickel is used as an
active phase because of its high efficiency in steam tar reforming compared to other known
catalysts but it can suffer rapid deactivation from carbon deposition and sintering and also
from alkali poisoning. Gasification of biomass also generates alkali aerosols and other
inorganics such as chlorine and sulphur that can harm the catalyst. The purpose of this study
is therefor to create a support that can prevent nickel from fast deactivation.
The potassium, that will be located inside the bulk of the tungsten, will act as a promoter by
increasing the number of active sites. The hexagonal tungsten structure will protect the
potassium from evaporating and thereby leaving the support which could have been a problem
if it had been on the surface of the support. The mix of potassium tungsten and tetragonal
zirconia shows a strong solid acidity. Strong solid acids can perform catalytic reactions at
lower temperature and thereby lower the temperature of the steam reforming process making
it more energy efficient.
The catalytic activity is tested at different steam to carbon ratios and temperatures with 1-
methylnaphthalene, 1-MN, as a simulated tar. Experiments will be conducted with and
without alkali to investigate the impact alkali has on the catalyst. In the experimental set-up
6
the catalyst is placed in a heated bed reactor. An inlet stream consisting of nitrogen, 1-MN,
steam, hydrogen and alkali is used. For the measuring of the composition after the reactor
three different instruments are used: micro-GC, surface ionization, SI, and solid phase
adsorption, SPA.
1.1 Aim and delimitations

The aim and objective of this project was to create a catalyst that could be suitable for tar
steam reforming in biomass gasification. Gasification of biomass generates a lot of
components that could cause operational problems and harm the catalyst. In last year master
thesis project Matteo Diomedi investigated a promising catalyst candidate for tar steam
reforming, alkali doped tungsten bronzes mixed with zirconia impregnated with nickel. This
project is a continuation of Matteo Diomedi’s master thesis project where the catalyst will be
further evaluated. Focus will be on trying to improve the preparation of the catalyst and also
further testing in a test rig at KTH. Measurements will be conducted on how alkali in the inlet
stream will affect the catalyst. The deactivation of the catalyst will be analysed by measuring
the decrease in tar reduction over time. A blank test will be used to compare the tar reforming
properties of the catalyst. The S/C ratio will also be varied to see how this affects the
reforming process. At last conclusions whether the catalyst is suitable or not will be drawn
from the results obtained.
There were a few delimitations for this study which is presented in punctuation form:
A simulated tar was used, 1-Methylnaphthalene, to replace other tars that can form in
biomass gasification.
Experiments were conducted in a test rig at KTH and the results may differ if tested in
a large scale reactor.
7
1.2 Nomenclature
1-MN – 1-Methylnaphthalene
AAEM – Alkaline and alkali earth metal
AAS – Atomic absorption spectrophotometer
GC – Gas chromatograph
GHSV – Gas hourly space velocity
HTB – Hexagonal tungsten bronzes
ITB – Intergrowth tungsten bronzes
IWI – Incipient wetness impregnation
MW – Molecular weight
PAH – Polycyclic aromatic hydrocarbons
S/C – Steam to carbon
SI – Surface Ionization
SPA – Solid Phase Adsorption
TPO- Temperature programmed oxidation
TPR – Temperature programmed reduction
TTB – Tetragonal tungsten bronzes
8
2 Background
2.1 Principles of biomass gasification
Gasification of biomass is a process of non-complete combustion, adding value to low- or
negative- valued feedstock for production of heat, transport fuels, electricity and for
production of chemicals such as ammonia. Biomass is chemically converted to gaseous
products in the presence of a gasifying agent. The gasifying agent can be steam, air, oxygen,
carbon dioxide or a mixture of these [3]. The major components of the gaseous products, also
called product gas, generated from gasification can be seen in Figure 1.
𝐵𝑖𝑜𝑚𝑎𝑠𝑠 + 𝑔𝑎𝑠𝑖𝑓𝑦𝑖𝑛𝑔 𝑎𝑔𝑒𝑛𝑡 + 𝑕𝑒𝑎𝑡 𝐶𝑂, 𝐶𝑂2 , 𝐻2 𝑂, 𝐻2 , 𝐶𝐻4 + 𝑜𝑡𝑕𝑒𝑟 𝑕𝑦𝑑𝑟𝑜𝑐𝑎𝑟𝑏𝑜𝑛𝑠

𝑇𝑎𝑟, 𝑐𝑕𝑎𝑟 𝑎𝑛𝑑 𝑎𝑠𝑕
𝐻𝐶𝑁, 𝑁𝐻3 , 𝐻𝐶𝑙, 𝐻2 𝑆 + 𝑜𝑡𝑕𝑒𝑟 𝑠𝑢𝑙𝑝𝑕𝑢𝑟 𝑔𝑎𝑠𝑒𝑠
Figure 1 – Major components in product gas from biomass gasification.
The product gas composition varies depending on the raw material. When evaluating a raw
material for biomass gasification the material properties are critical. Properties of importance
are moisture content, calorific value, proportions of fixed carbon and volatiles, ash/residue
content, alkali metal content and cellulose/lignin ratio. [4]
The gasification process is performed at high temperatures between 800- 1800 °C depending
on characteristics of biomass and reactor type [5]. The process generally includes several
steps[1]:
Drying process - Evaporation of moisture

Pyrolysis process – In this process volatile matter is produced from cellulose,
hemicellulose and lignin in the form of steam and gaseous- and condensable
hydrocarbons in absence of an oxidizing agent. The temperature range is between
300-500 °C.
Gasification process - Partial oxidation of the solid char, pyrolysis- tars and gases.
Combustion process- Either internally or separate combustion of char and volatile
products generating heat needed for the other processes.
In these processes various reactions takes place involving carbon, carbon monoxide, carbon
dioxide, water, steam hydrogen and methane. The most important reactions are described
below: [5]
𝐶 𝑂 𝐶𝑂 (Eq. 2.1)
𝐶𝑂 𝑂 𝐶𝑂 (Eq. 2.2)
9
𝐻 𝑂 𝐻𝑂 (Eq. 2.3)
𝐶 𝐶𝑂 𝐶𝑂 (Eq. 2.4)
𝐶 𝐻𝑂 𝐶𝑂 𝐻 (Eq. 2.5)
𝐶 𝐻 𝐶𝐻 (Eq. 2.6)
𝐶𝑂 𝐻𝑂 𝐶𝑂 𝐻 (Eq. 2.7)
𝐶𝐻 𝐻𝑂 𝐶𝑂 𝐻 (Eq. 2.8)
Equations 2.1 to 2.3 describe the combustion process which occurs in the oxidation zone
where C, CO and H2 reacts with O2. These reactions are complete and are not considered in
determining an equilibrium product gas composition. Heat is generated from these reactions
to carry on the other endothermic reactions in the other processes. Equations 2.4 to 2.6 are the
Boudouard reaction, the water gas reaction and the methanation reaction respectively. These
are heterogeneous reactions where some of the tars and char are converted into gas. Equations
2.7 to 2.8 are the CO-shift reaction and the steam methane reforming reaction. These are
homogeneous reactions which may alter the composition of the product gas.
There are many parameters in the gasification process that can affect the composition of the
product gas. Some of the parameters are feed composition, water content, gasification
temperature, the extent of oxidation of the pyrolysis products and the effect of catalyst [1]. For
syngas production the product gas must contain high amounts of hydrogen and carbon
monoxide. The quality of syngas can vary depending on the end use. Syngas quality is defined
based on carbon conversion and ratios of H2/CO, CH4/H2, gas yield and gasification
efficiency [6]. For Fischer-Tropsch synthesis the optimal ratio of H2/CO is 2:1 [7].
Depending on the usage of the product gas and the feed composition there are different
reactor types. The three most common gasifier categories are moving-bed-, fluid-bed- and
entrained flow gasifier.
2.1.1 Tar and tar reforming

The product gas formed from gasification contains many impurities, e.g. particulates, tars and
inorganic compounds. Tar formation is of great concern since tar can cause several problems
if not removed such as operational problems by blocking gas coolers, filter elements and
engine suction channels [8]. From Milne et al. tar is defined as: “The organics, produced under
thermal or partial-oxidation regimes (gasification) of any organic material, are called “tars”
and are generally assumed to be largely aromatic” [9]. A more simple definition is aromatics
with a molecular weight larger than benzene can be classified as tar.
10
Formation of tar is based on series of complex reactions during gasification process and is
dependent on reaction conditions. A tar formation scheme was proposed by D.C Elliot [10]. It
is shown in Figure 2 below and illustrates the transition of tars as a function of process
temperature from primary products to phenolic compounds to aromatic hydrocarbons.
𝑀𝑖𝑥𝑒𝑑 𝑃𝑕𝑒𝑛𝑜𝑙𝑖𝑐 𝐴𝑙𝑘𝑦𝑙 𝐻𝑒𝑡𝑒𝑟𝑜𝑐𝑦𝑐𝑙𝑖𝑐 𝐿𝑎𝑟𝑔𝑒𝑟

𝑂𝑥𝑦𝑔𝑒𝑛𝑎𝑡𝑒𝑠 𝐸𝑡𝑕𝑒𝑟𝑠 𝑃𝑕𝑒𝑛𝑜𝑙𝑖𝑐𝑠 𝐸𝑡𝑕𝑒𝑟𝑠 𝑃𝐴𝐻 𝑃𝐴𝐻
400 °C 500 °C 600 °C 700 °C 800 °C 900 °C
Figure 2 – Tar formation scheme, where PAH is polycyclic aromatic hydrocarbons
Variables that affect the tar formation are the composition of feedstock and the temperature of
the reactor [8]. Studies from Baker et al. show that tar yield is reduced at higher temperatures
gasification temperatures [11]. Reforming of tar by thermal cracking alone is however not
economically favourable.
There are four major product classes of tar, primary-, secondary-, alkyl tertiary and condensed
tertiary products from thermal conversion of biomass. The different types of tars together with
the product classes can be found in table 1. The product classes are temperature dependent, all
primary products have been found thermally cracked before the tertiary products appear [9].
Table 1- Classification of tars.
Product class Tars

Primary products Cellulose- derived products such as
levoglucosan, hydroxyacetaldehyd and
furfurals. Hemicellulose derived products and
lignin derived methoxyphenols
Secondary products Phenolics and olefins
Alkyl tertiary products Methyl derivatives of aromatics, methyl
acenaphtylene, methyl naphthalene, toluene
and indene
Condensed tertiary products PAH series without substituents, benzene,
naphthalene, acenaphtylene,
anthracene/phenanthrene and pyrene
Depending on the usage of the product gas there are strict limitations of the tar content. For
syngas production, gas turbine and fuel cells the levels are very low, ranging from 0.05-5
g/Nm2. In other applications such as internal combustion engines and pipeline transport the
limits of tar contents are significantly higher, 50-500 g/Nm2 [12].
When the temperature decreases, the tars in the product gas become over-saturated and
therefore start to condense [8]. This depends on the tar vapour pressure and the saturation
11
pressure of the tar. Tars have individual saturation pressure and it depends on the tar dew
point which is the temperature when the total partial pressure of tar equals the saturation
pressure of tar. Tar reforming is therefore an important process and thereby lowering the total
partial pressure of tar or complete decomposition of the tar. Hence tar condensation is avoided
by lowering the total partial pressure of tar.
Tar can be decomposed in several ways: [8]
Thermal cracking: 𝑝𝐶 𝐻 𝑕𝑒𝑎𝑡 𝑞𝐶 𝐻 𝑟𝐻 (Eq. 2.9)
Steam reforming: 𝐶 𝐻 𝑛𝐻 𝑂 𝑛 𝐻 𝑛𝐶𝑂 (Eq. 2.10)
Dry reforming: 𝐶 𝐻 𝑛𝐶𝑂 ( )𝐻 𝑛𝐶𝑂 (Eq. 2.11)
Carbon formation: 𝐶 𝐻 𝑛𝐶 ( ) 𝐻 (Eq. 2.12)
Here CnHx represents tar and CmHy represents hydrocarbons with less carbon. Tar can also be
removed physically with wet/dry technologies, but since this study is limited to steam
reforming of tar with alkali doped bronze catalyst the physical removal of tar will not be
considered [9]. There are primary and secondary methods for tar reduction. In primary
methods the tar is removed in the gasifier while in secondary methods tar is removed after the
gasifier, described in Figure 3 below [13].
12
Primary method
Gasifier
Biomass Application
Gas clean-up
Tar removal
Gasifying agent
Secondary method
Tar removal
Gasifier Application
Gas clean-up
Gasifying agent
Figure 3 – Primary and secondary method of tar reforming
In this study the focus will be on secondary tar removal with a catalyst after the gasification
process. Hence a gasifier will not be used and the product gas and tar will be simulated.
To simulate a product gas stream to the inlet of the reactor 1-MN was used as a tar
component. 1-MN is an aromatic compound with the formula C11H10. It was bought at Sigma-
Aldrich with the CAS number 90-12-0. It is a stable compound with a high boiling point of
around 240°C.
The reason 1-MN was used is that naphthalene are one of the most common tar groups
formed in biomass gasification [14]. 1-MN will replace all tar components in the product gas.
From Nair et al., the decomposition scheme of naphthalene can be seen in Figure 4 below [15].
13
Figure 4 – Naphthalene decomposition path
Naphthalene and other tars have been used in a study for steam tar reforming by Roberto Coll
et al. [16]. The study covered the order of reactivity of tars and their tendency of coke
formation. The results showed that the reactivity rate and tendency to form coke increased
with larger aromatic rings. However naphthalene was an exception with the lowest reactivity
rate of the tars tested, benzene > toluene >> anthracene >> pyrene > naphthalene. That is also
a reason why naphthalene is suitable as a simulated tar.
2.1.2 Alkali compounds and its effect on biomass gasification

Alkali compounds can be found in biomass and is released during thermal conversion in
gasification process. Compared to fossil fuel, the amount of alkaline and alkali earth metal
species, AAEM, is significantly higher in biomass, mainly K, Na, Ca and Mg [17]. Part of the
AAEM is volatilized during the gasification process and this can lead to severe problems such
as slagging, agglomeration, deposition and corrosion in advanced combustion systems. Long
Jiang et al. studied the release characteristics of AAEM and found out that during gasification
of three biomass samples the release rate, amount vaporised, of alkali and alkaline earth
metals were 12-34 % and 12-16 % respectively [17].
However, AAEM can also have a catalytic effect both in the thermal conversion process and
also by steam reforming of volatiles [17]. Alkali metals are active catalysts in the reaction with
oxygen containing species by increasing the number of active sites without altering the kinetic
mechanism [18].
14
2.2 Catalytic tar reforming
A catalyst is a substance that increases the rate of reaction without itself being consumed.
Generally it consists of a support material and an active phase but it can also contain a
promoter. A promoter enhances the catalytic activity of the catalyst. The support can
sometimes also work as a promoter.
The main purpose of a catalyst in biomass gasification is the removal of tar and increase the
gas yield, but it can also change the composition of the product gas. Tar elimination reactions
are kinetically limited and by using a catalyst the reaction rates can be increased. There are
generally three main groups of catalysts materials for tar reforming in biomass gasification,
dolomite-, alkali metal- and other metal- and nickel catalysts [19]. Z. Abu El-rub et al.,
investigated 9 groups of catalysts in a review where he divided them into two separate classes
[20]
. The two classes were synthetic catalysts and minerals. The review considers chemical
composition, factors of catalytic activity, deactivation and advantages/disadvantages of each
catalyst. According to Z. Abu El-rub et al. nickel based catalysts are 8-10 times more active
than dolomite.
2.2.1 Choice of catalyst

For the choice of catalyst studied in this project there were a few criteria. The catalyst must be
a heterogeneous catalyst effective in tar removal, resistant in deactivation from alkali, tar
fouling and sintering and it should not be expensive. The catalyst will be in powder form and
placed in a bed reactor with a constant temperature. A promising catalyst was designed in last
year master thesis project, KxWO3-ZrO2 impregnated with nickel. The catalyst will be
recreated, improved and further evaluated in this master thesis project.
2.2.1.1 The active phase

Nickel catalysts are primarily used in secondary tar reforming and have high tar removal
efficiencies [19]. However nickel catalysts also have the advantage of methane reforming and
altering the gas composition of H2/CO ratios towards syngas quality. It is relatively cheap but
it can suffer from rapid deactivation at hot gas temperatures from biomass gasification.
Deactivation is mainly from carbon deposition and sintering, but can also be caused by other
factors such as sulphur, chlorine and alkali metal poisoning that can be present in the biomass
[21]
. Since alkali content is relatively high in biomass, which has been described earlier in
section 2.1.2, this can cause a problem. Alkali acts as catalyst poison and can block the
catalyst pore system.
Carbon deposition can be minimized if an excess of steam is used in tar steam reforming. In
this project various S/C ratios will be tested to find the optimum value. There are negative
effects of using an excess of steam however due to the increased energy costs of heating water
[22]
.
15
2.2.1.2 The support material
When investigating the support material important parameters are the pore structure, metal-
support interactions and the acidity- basicity. It can influence the metal dispersion, metal
crystallite size and carbon deposition on the catalyst surface which affects the catalytic
performance and activity of the catalyst [22].
For biomass gasification it is important that the support material can protect the nickel from
deactivation and is stable at high temperatures. By using KxWO3-ZrO2 as a support the idea is
that the potassium will work as a promoter for the catalyst by increasing the number of active
sites. The potassium will be dispersed in the bulk of the support and not on the surface. This is
done in order to prevent the potassium from vaporising at high temperatures and thereby
leaving the support. The potassium also works as an electron donor to the bronzes. In Figure 5
it can be seen how alkali is dispersed in the hexagonal structure of the tungsten trioxide.
Figure 5 – Structure of hexagonal tungsten bronzes
Depending on the value of “x” the bronzes can have a number of structures. Hexagonal
tungsten bronzes are formed at values of x between 0.13-0.31 and tetragonal tungsten bronzes
are formed at values of 0.40-0.59 for potassium [23]. There are also other possible structures
that can be formed. A. Hussain studied the phase analysis of formation temperature and
potassium content and the results can be shown in the Figure 6 below [24].
16
Figure 6 – Phase analysis of formation temperature and potassium content.
The alkali concentration also affects the electrical properties. The tungsten bronzes acts as
semiconductors at x<0.25 and metallic conductors at x>0.25 [25]. This is thought to be
correlated with the variation of crystal structure at different alkali concentrations.
When x=0.25 studies have shown anomalous transport properties of the potassium tungsten
bronzes. S. Raj et al. investigated this and concluded that this phenomenon was due to hidden
one-dimensional bands [26]. This concentration of alkali will be used in the design of the
catalyst.
Potassium tungsten trioxide mixed with tetragonal ZrO2 shows a strong solid acidity. This
strong acidity is because of the formation of Brønsted acid sites on the zirconia support from
reduction W6+ in reactant environments [27]. The creation of surface acidity is still unclear but
the strong acidic sites are related to the presence of tetragonal structured zirconia. To obtain a
tetragonal structure of zirconia is therefore very important. Other structures of zirconia which
can be formed are monoclinic- and cubic structures.
Strong solid acids can perform catalytic reaction at a lower temperature compared to non-
solid acid catalysts. Solid acids generally have two problems however, the strength of acidity
is not uniform on the catalyst and the strong acid sites deactivate quickly [28].
2.2.2 Catalyst preparation

There are different ways preparing the bronzes K0.25WO3. A. Hussain used stoichiometric
reaction steps for the synthesis, Eq. 2.13 [24]. For this master thesis project the catalyst will be
prepared the same way.
𝑀 𝑂 𝑥 𝑂 𝑂 𝑀 𝑂 (Eq. 2.13)
17
M is an alkali metal, in this case potassium. For this calcination step, two more synthesis steps
must be done to obtain the reactants needed, K2WO4 and WO2.
𝐶𝑂 𝑂 𝑂 𝐶𝑂 (Eq. 2.14)
𝑂 𝐻 𝑂 𝐻 𝑂 (Eq. 2.15)
In Eq. 2.14 potassium tungstate is prepared from potassium carbonate and tungsten-trioxide
[29] [30]
, . The reaction rate is very slow in this synthesis step and requires a high temperature
and a long calcination time. Also by using this method impurities can form since the
conversion is not 100 %. Therefore another synthesis step for obtaining potassium tungstate,
Eq. 2.16 was tested and compared to the synthesis step in Eq. 2.14. In the Eq. 2.16 potassium
nitrate was added to overcome the reaction hindrance of the reaction K2CO3 – WO3. The
potassium nitrate forms an eutectic containing 3.7 weight % K2CO3 and melts at 320 °C. The
melting point of the mixture of the reactants is lowered to 650 °C and if the reaction is carried
out in liquid phase the diffusion hindrance is limited. This synthesis step has been tested and
described by G. K. Shurdumov et al. [31].
𝐶𝑂 𝑁𝑂 𝑂 𝑂 𝑁𝑂 𝐶𝑂 𝑂 (Eq. 2.16)
In synthesis step Eq. 2.15, tungsten-dioxide is formed from tungsten-trioxide and hydrogen.
There are many other oxides that can be formed such as WO2.96, WO2.90, WO2.72, WO2 and
W3O that can affect the purity of the wanted product. Therefore the variables and settings had
to be optimized to yield mainly WO2.The variables affecting the reduction of potassium-
trioxide is temperature, time, H2/N2 flow rate and ratio and pressure ratio of H2/H2O [32]. At
lower temperatures, T <1075 K, the reduction WO3 to WO2 occurs, through the intermediates
(WO3 WO2.96 WO2.90 WO2.72 WO2) and at higher temperature, T > 1075 K, the
reduction goes towards WO [32].
Zirconia can be obtained by thermal decomposition of amorphous zirconium hydroxide.

Temperatures of 400°C yields the formation of metastable tetragonal zirconium dioxide [33].
𝑟 𝑂𝐻 𝐻𝑒𝑎𝑡 𝑟𝑂 𝐻 𝑂 (Eq. 2.17)
There has been a lot of research on tungsten trioxide/zirconia as a support material but little
attention has been given for alkali doped tungsten trioxide/zirconia. A lot of studies have been
conducted on how to prepare WO3-ZrO2 with impregnation of ammonium hydroxide solution,
but for the preparation of K0.25WO3-ZrO2 none could be found. Therefore in this project the
method proposed is to mix correct stoichiometric amounts of zirconia and bronzes before
grinding and calcination at 500°C. Analysis with XRD will be performed for evaluating the
structure and composition of the support.
Incipient wetness impregnation, IWI, is a method to add nickel to the support material
K0.25WO3-ZrO2. First the pore volume of the support must be known in order to know how
much volume it can absorb before being saturated. The measured nickel complex is then
dissolved in water and added drop wise to the support. The volume of aqueous nickel added is
18
equal to the pore volume to fill the entire pore system. After the impregnation the catalyst is
dried to evaporate all the water so that only nickel remains in the pores.
2.2.3 Instruments used for characterisation of the catalyst

2.2.3.1 XRD
For analysing and determining the structure and composition of the support prepared, XRD
was used. The instrument used was a Siemens XRD Diffractometer D5000 and is used for
powder examinations. X-rays are generated by applying current and voltage to an X-ray
diffraction tube with a copper anode. The X-rays will hit the crystal structure of the sample
with an incidence angle θ when the sample holder rotates. The layers of atoms in the crystal
structure will produce a diffraction maximum that reflects the incident x-rays if the incidence
angle is correct according to Figure 7 below [34]. The detector then records the diffracted X-
rays.
1
λ
2
θ d
p r
Figure 7 – Diffraction of X-rays from two parallel planes of atoms
Bragg’s equation, Eq. 2.20, gives the relationship of when a constructing interference occurs
for a specific d-spacing and angle θ with an integer number of wavelengths. The primary
beam is reflected from the sample to the detector every time Bragg’s condition is satisfied.
𝑛 (Eq. 2.20)
2.2.3.2 BET
BET can be used to determine the surface area and pore volume of the catalyst and was
founded by Brunauer, Emmett and Teller. The instrument used for characterisation was a
Micromeritics ASAP 2000. Nitrogen was used as analysing gas and the basics of BET are that
nitrogen gas is adsorbed on the solid and the amount of gas adsorbed depends on the exposed
surface area, temperature, gas pressure and the gas-solid interactions. The adsorption of BET
is an extension of the Langmuir theory which is based on monolayer adsorption of gas
molecules.
19
The BET equation can be written as [35]:
(Eq. 2.21)
Where V is the volume of gas adsorbed at pressure P, V is the volume of gas adsorbed in the
monolayer, P0 is the saturation pressure of the adsorbed gas and C is a constant related to
the heat of adsorption in the first layer and the heat of liquefaction of adsorbed gas on all
layers.
Before analysing, the sample has to be degassed. This is done to desorb water and other
compounds from the sample to ensure that volatile impurities are removed. When running the
measurements known amounts of nitrogen is stepwise introduced into the sample holder until
saturation pressure is reached and the pressure changes during the formation of adsorption
layers are registered. The results are presented in isothermal plots of volume adsorbed versus
pressure. The two most common isotherm approaches during BET measurements are type II-
and type IV form [36].
20
2.3 Analysing instruments for product gas composition
In this section the three different analysing instruments used in this project will be described.
The instruments are surface ionization technique, four micro-GCs and solid phase adsorption.
Solid phase adsorption will not be discussed however since the samples will be sent to
Verdant Chemical Technology for analysis.
2.3.1 GC for measurement of gases, hydrocarbons and low molecular

weight tar
Gas chromatography (GC) is a high selective separation method that has the ability of
separating components of closely similar physical and chemical properties. The components
are separated by retention in the column either by vapour pressure or polarity. Since a GC
operates at rather low temperatures it is best suited for separation of components with a
molecular weight, MW, below 300 g/mol [37]. The components to be analysed is periodically
injected through an injection valve to a column that is heated. An inert mobile phase, called
carrier gas, is flowing continuously through the column. Normally the carrier gas consists of
nitrogen, hydrogen or helium. The components are separated by retention in the column and
reach a detector at a varying time interval. Different detectors can be used, such as flame
ionisation detector, thermal conductivity detector, photo-ionization detector and many more.
2.3.1.1 Instrumental set-up micro-GC

For this project four coupled micro-GCs were used for detection of gases, hydrocarbons and
low-MW tar. The instrument was a C2V-200 MicroGas Chromatograph from Thermo
Scientific. Each micro-GC measures different components in the product which can be seen in
table 2. The detector used is a thermal conductivity detector. It has a capillary column of 0.25
mm in inner diameter.
Table 2- Compounds detected by the different micro-GCs.
Micro- GC Detected compounds

1 CO2, C2H6, C2H4, C2H2
2 1-Butene, C6H6, C7H8
3 O2 + Ar, N2, CH4, CO
4 H2
The result of the analysis is presented as graphs with a varying height depending on
concentration, at different time intervals. Calculation is based on area under the graph in
correlation to the concentration and amount of components.
2.3.2 Surface ionization technique for measurement of alkali levels

Surface ionization, SI, is a technique that can be used for detection of alkali compounds. It is
an old technique that was discovered as early as 1889 by Elster and Geitel [38]. The principles
are that an atom or molecule is adsorbed on a hot surface, a metal wire, and after desorption
they are emitted in ionic or neutral form by thermal ionization. When an alkali compound is
21
directed to the hot platinum surface, positive ions are emitted after attaining the thermal
equilibriums:
𝑀 𝑀 and 𝑀 𝑀 𝑒
For this to occur, the alkali compound first melt and decomposes on the hot surface through
breaking of chemical bonds. At the hot surface the alkali atom becomes ionized and an alkali
ion is emitted while the electron is adsorbed by the hot metal surface [39].
The Saha-Langmuir equation, Eq. 2.22, describes the statistical probability of the ionic and
neutral fluxes from a molecular beam [38].
( ) (Eq. 2.22)
Where α is the ratio of ionic and neutral fluxes, wM+/ wM0 is the statistical weight ratio of ions
and neutrals, θ is the average work function, IP is the ionization potential, kB is the
Boltzmann constant and T is the temperature of the surface.
Surface work function for a solid is the minimum energy required to remove an electron and
thereby forming positive or negative ions. The work function of the heated surface therefore
needs to be higher than the ionization potential of the adsorbed atom/molecule [40]. For alkali
metals the ratio of ionic and neutral fluxes is very high with a platinum wire compared to
most other elements. This is because the ionization potential of alkali metals are very low
compared to the surface work of platinum, IPNa = 5.14 eV, IPK = 4.34 eV and θPt ≈ 5.5 eV [41].
The size of the alkali particle affects the ionization efficiency greatly. Detection of different
alkali salt particles with varying diameters at atmospheric pressure has been studied by U.
Jäglid et al. [42]. Experiments show that all types of salt particles with a size of 5 nm or below
have melted and ionized completely while larger particles become partially ionized. For larger
particles the ionization efficiency depends on the salt properties. For particles sizes of 100 nm
the ionization efficiencies have decreased to around 1 % [42].
2.3.2.1 Instrumental set-up surface ionization

The instrument used in the experimental section consists of a heated sampling line for the
extraction of product gas and a surface ionization detector. It is designed similarly to the set-
up used by Davidsson et al. [41]. The surface ionization detector is an ion collector and the
signal will be measured by a picoammeter model 6485, from Keithley instruments. A detailed
view of the surface ionization detector is found in Figure 8.
22
Figure 8 – Surface Ionization detector
During the experimental run the sampling line must be kept at temperature similar to the
reactor temperature to avoid condensation of compounds. Two thermocouples were used for
measuring of the temperature. A nitrogen flow is introduced downstream of the detector to
protect sensitive parts. Measured ion currents of alkali is generally very low for this
application and the therefore the resistance of the ion collector must be very high, >10 12 Ω.
The optimal temperature of the heated Pt-filament for potassium is 1500 K for a high
ionization efficiency [41].
23
3 Experimental section
3.1 Catalyst preparation
The catalyst preparation will be described in detail in this section. Two methods were
prepared, one from Eq. 2.14 and the other from Eq. 2.16. For every method the amount of
nickel varied with 5-, 10- and 15 wt %. In total six different 20wt%Bronze- ZrO2 catalysts
were prepared and defined as 1.1, 1.2, 1.3, 2.1, 2.2 and 2.3, where the first number indicates
the method and the second number the weight load of nickel. This can be seen in table 3
below.
Table 3- Definition of catalyst. The first number indicates the method and the second number is the Nickel weight
load.
Catalyst number Method Nickel wt %

1.1 1 5
1.2 1 10
1.3 1 15
2.1 2 5
2.2 2 10
2.3 2 15
3.1.1 Preparation of support

For the preparation of the bronzes, Eq. 2.13, an x-value of 0.25 was decided for KxWO3. First
K2WO4 and WO2 had to be prepared from Eq. 2.14 (Eq. 2.16 for method 2) and Eq. 2.15.
𝑂 𝑥 𝑂 𝑂 (Eq. 2.13)
𝐶𝑂 𝑂 𝐶𝑂 (Eq. 2.14)
𝐶𝑂 𝑁𝑂 𝑂 𝑁𝑂 𝐶𝑂 𝑂 (Eq. 2.16)
𝑂 𝐻 𝐻 𝑂 (Eq. 2.15)
The compound K2WO4 was prepared in two different ways. For the first method 20 mmol of
K2CO3 was mixed with 20 mmol of WO3 and grinded for 10 minutes in a mortar. Then it was
calcined at 750 °C for 8 hours in air. For the second method, 4.5 mmol of K2CO3 was mixed
with 9 mmol of KNO3 and 9 mmol of WO3 and grinded in a mortar for 10 minutes. This was
calcined at 650 °C for 1 hour in air.
24
Table 4- Mass and mass loss of K2WO4 prepared from Eq. 2.14 and Eq. 2.16.
Method Mass of reactants before calcination Mass after calcination Mass loss [%]
1 7.2737 g 6.3381 g 12.86
2 3.5895 g 2.8804 g 19.75
The WO2 was prepared from Eq. 2.15 with temperature programmed reduction (TPR) where
WO3 was reduced to WO2 at 850 °C for 3 hours with a 5 % H2/Ar flow. The colour changed
from light green to brown.
Table 5- Mass and mass loss of WO2 prepared by TPR from Eq. 2.15.
Method Mass WO3 Mass WO2 after TPR Mass loss [%]
1 1.7365 g 1.6053 g 7.56
2 1.7500 g 1.5895 g 9.17
For preparing K0.25WO3, stoichiometric amounts of K2WO4, WO3 and WO2 were mixed and
grinded for approximately 30 minutes according to Eq. 2.13. Then it was weighed and
transferred to a quarts tube. The quarts tube was connected to a vacuum pump (Edward model
RV8) and put in a furnace for 300 °C for 3 hours to remove volatile compounds. The vacuum
pump was then turned off with the assumption that vacuum was established in the quarts tube
and the sample was calcined at 850 °C for 24 hours. The colour of the powder changed from
dark green to dark blue after the calcination.
Table 6- Mass of K0.25WO3 prepared from Eq. 2.13.
Method Mass K2WO4 Mass WO3 Mass WO2 Mass K0.25WO3

1 1.3038 g 7.4000 g 1.1480 g 9.8518 g
2 1.7352 g 7.4044 g 1.1491 g 10.2887 g
Zirconia was prepared from zirconium hydroxide by thermal decomposition according to Eq.
2.17. Zirconium hydroxide was measured and put in a silica tube and decomposed at 420 °C
and 3 hours in air.
𝑟 𝑂𝐻 𝐻𝑒𝑎𝑡 𝑟𝑂 𝐻 𝑂 (Eq. 2.17)

Table 7- Mass and mass loss of Zirconia prepared from Eq. 2.17.
Mass Zr(OH)4 Mass ZrO2 after thermal decomposition Mass loss [%]
33.6446 g 27.9449 g 16.94
Bronzes and zirconia was then mixed and grinded for 5 minutes and then calcined at 500 °C
for 3 hours.
25
Table 8- Mass of 20% Br- ZrO2 prepared.
Method Mass K0.25WO3 Mass ZrO2 Total mass 20%Br-ZrO2

1 2.9350 g 11.7530 g 14.6880 g
2 3.0487 g 12.1429 g 15.1916 g
3.1.2 Nickel IWI method

The support was then impregnated with nickel by a method called incipient wetness
impregnation, IWI. A catalyst with three different wt % of nickel was prepared for each
method.
Due to the low pore volume of the support, which can be seen in table 10, the impregnation
had to be done in several steps with a drying process of 110 °C for 3 hours in between each
impregnation to remove the water. After the IWI the nickel impregnated catalyst was calcined
at 450 °C for 3 hours to remove the nickel precursor. In total 6 catalysts were prepared with
varying nickel load and with a mass of 5 g each which can be seen in table 9.
Table 9- Mass balance of catalyst with 5-, 10- and 15 wt% Ni.
Compound 5 wt% nickel 10 wt % nickel 15 wt % nickel

ZrO2 3.80 g 3.60 g 3.40 g
Bronzes 0.95 g 0.90 g 0.85 g
Nickel 0.25 g 0.50 g 0.75 g
Nickel complex 1.24 g 2.48 g 3.72 g
The pore volume and surface area of the support was found after BET analysis.
Table 10- BET analysis of support material for nickel impregnation.
20%Br- ZrO2 Pore Volume [cm3/g] Surface area [m2/g]

Method 1 0.054151 52.2783
Method 2 0.068197 62.5550
26
3.2 Catalyst testing
In this section the test rig and experimental testing will be described. The test rig used was in
a laboratory scale, located at KTH.
3.2.1 Experimental set-up
Figure 9 – Detailed drawing of the test rig
The main components of the test rig were the mass flow controllers, syringe pump, alkali
aerosol generator, pre-heater, reactor, SI, micro GC’s and water traps. The inlet stream from
the mass flow controllers consisted of hydrogen (g), nitrogen (g), 1-MN (l), water (l) and
alkali aerosols. H2, N2 and water were mixed in the preheater and heated up before reaching
the reactor. 1-MN and alkali aerosols were added after the preheater. The reactor was an
electrical furnace with a quarts tube inside containing the catalyst. The idea with using a
quarts tube for the catalyst bed is that it will not take part in the reactions.
After the catalytic reactions in the reactor the product gas was divided into two streams. One
stream went to the exhaust and micro GC’s as can be seen in Figure 9 while the other stream
went to the SI-instrument. Water traps were used to remove water in the product gas to protect
the micro GC’s.
27
3.2.2 Calibration of analysing equipment
The analysing instruments had to be calibrated before the experimental testing. The micro
GC’s were calibrated with three different calibration standards.
Table 11- The different calibration standards used for the micro GC’s.
Calibration standard Components

1 N2, CO2, CO, H2 and CH4
2 C2H6, C2H4, C2H2, 1-Butene, Benzene, Toluene and N2
3 Air
For constructing the calibration curve of the SI there were a few problems. There was no trend
in the results and the sensitivity seemed to be very poor. This led to the decision of not using
the SI in the experiments. An evaluation of why the problems occurred will be discussed later
in the report.
3.2.3 Activity evaluation

3.2.3.1 Catalyst reduction
Before the experiments could be performed, the catalyst had to be reduced since nickel is only
active in its metallic form and not in an oxidised state. Daniele, A et al. studied the effects of
temperature, heating rate, H2/N2 ratio and the amount of reducible species for NiO reduction
[43]
. They found out that the heating rate and the hydrogen concentration had the largest
impact on the reduction. An equation was made for the variation of heating rate and hydrogen
concentration:
(Eq. 3.1)
K is a characteristic number for appropriate operating variables, S0 is the initial amount of

reducible species, c0 is the inlet hydrogen concentration and V is the total volume of the
reducing gas.
The optimal value of K for NiO reduction is between 55s at a heating rate of 6 °C/min and
140s at a heating rate of 18 °C/min. By using this formula the reduction conditions for this
experiment was decided.
The catalyst was reduced at 600°C with a heating rate of 10 °C/min for 1.5 hours. The inlet
flows to the reactor were 71 ml/min of H2 and 400 ml/min of N2.
28
3.2.3.2 Alkali aerosol generator
The alkali aerosol generator used converted alkali in solution to alkali in aerosol with N2 gas.
The alkali concentration is therefore different in solution compared to aerosol and a
conversion factor had to be calculated. Variables affecting the conversion factor are the alkali
concentration in the solution, the amount of N2 flow and the inlet flow to the electrical
furnace.
The inlet flow to the alkali spray was 2300 ml/min N2 and the outlet flow was 2300 ml/min
N2 and 0.1 ml/min Milli-Q water containing alkali aerosols. The outlet flow from the alkali
spray is then mixed with the main stream (H2, N2, H2O and 1-MN) before the electrical
furnace.
To calculate the conversion factor a standard solution of 1000 ppm of alkali was used. Hence
1 kg of Milli-Q water contains 1g of alkali and therefore 0.1 g of Milli-Q water contains
0.0001 g of alkali.
Table 12- The total flow rate to the inlet of the reactor with the alkali spray.
Compound Flow rate [ml/min] Flow rate [g/min]

N2 2300 2.632
H2 71 or 98 0.008
H2O syringe pump 0.558 (S/C =4) or 0.837 (S/C=6) 0.558 (S/C =4) or 0.837 (S/C=6)
H2O alkali spray 0.1 0.1
1-MN 0.1
Total 3.398 (S/C =4) or 3.675 (S/C=6)
The total flow rate to the reactor is 3.398 g/min or 3.675 g/min with 0.0001 g/min of alkali.
The concentration of alkali to the inlet stream of the reactor is therefore 29.45 ppm (S/C =4)
or 27.21 ppm (S/C =6). From this the conversion factor can be calculated which is shown in
table 13. A solution of 274.8 ppm alkali was used in the experiments. This equals 8.1 ppm in
aerosols for part 1 and with the reduced flows in part 2 a concentration of 3.4 ppm.
Table 13- Alkali conversion factor for S/C ratio 4 and 6.
Alkali conversion factor Solution to aerosol Aerosol to solution

S/C=4 0.0229 33.957
S/C=6 0.0272 36.747
3.2.3.3 Experimental plan

For the experimental testing it was decided to test the different catalysts at different
temperatures and with a varying steam to carbon ratio (S/C ratio). In total 17 experiments was
conducted which can be seen in Figure 10.
29
Figure 10- An experimental scheme for testing of catalysts 20wt%Br-ZrO2 impregnated with nickel.
30
The experiments were conducted at 700°C and 800°C. For experiments 1, 2, 9 and 10 no
alkali was used in order to measure the catalytic tar reformation without the impact of alkali.
The alkali spray was used for the rest of the experiments. A S/C ratio of 4 was used for part 1
and a S/C ratio of 6 was used for part 2.
The first idea was to run all the experiments at the same condition but with a varying S/C ratio
and with or without alkali. After part 1 it was decided to lower the flow rates in order to
reduce the gas hourly space velocity, GHSV. This is why the conditions for part 2 are
different. For part 2 the settings varied from experiment to experiment in order to achieve
better measurements. The changes made for part 2 can be found in table 14 below.
Table 14- Parameters changed in experiments from part 2.
Experiment Parameter changed

10-17 Increased experimental run time from 4 to5 hours
11-17 1-MN flow was changed from 0.1 to 0.03 g/min
13-17 Catalyst load increased from 0.5 to 1.0 g
14-17 Lowered alkali in solution from 274.8 ppm to
137.5 ppm
15 Experiment conducted at 800°C only
17 Tested at a higher temperature, 900°C.
31
4 Results and discussion
4.1 Catalyst characterisation
4.1.1 BET results
The results from the BET are presented in table 15 below. The BET results of interest are the
pore volume and surface area of the catalysts.
Table 15- BET results of surface area and pore volume.
Catalyst Ni % Surface area [m2/g] Pore volume [cm3/g]

1.1 5 40.68 0.054
1.2 10 41.70 0.058
1.3 15 41.85 0.059
2.1 5 41.32 0.055
2.2 10 45.46 0.058
2.3 15 41.61 0.061
The results show that the catalysts prepared had a rather low surface area and a low pore
volume. In general a catalyst support containing zirconia and tungsten has a low surface area,
however it might be possible to construct catalysts with a higher surface area by changing the
method for preparing the support or the preparation method of zirconia. For example during
the preparation of the support, a higher purity of the material used might yield a higher
surface area. Other parameters that could affect the surface area are temperature and time for
the thermal decomposition. Hence changing these parameters could provide a catalyst with a
higher surface area than the one prepared in this paper.
4.1.2 XRD Results

The results from the XRD patterns are presented in Figure 11 and Figure 12 below.
Information of the structure and composition can be obtained from these results. The variation
of nickel inside the bulk of each catalyst can be seen at around 440 on the x-axis.
32
Catalyst 1.1
Catalyst 1.2
Catalyst 1.3
37.6
54.2
10.0
12.8
15.5
18.3
21.0
23.8
26.6
29.3
32.1
34.8
40.4
43.1
45.9
48.6
51.4
56.9
59.7
62.4
65.2
68.0
70.7
73.5
76.2
79.0
81.8
84.5
87.3
2-Theta scale
Figure 11- XRD results of catalysts from method 1
Catalyst 2.1
Catalyst 2.2
Catalyst 2.3
18.6
47.2
75.8
10.0
12.9
15.7
21.4
24.3
27.2
30.0
32.9
35.7
38.6
41.5
44.3
50.0
52.9
55.8
58.6
61.5
64.3
67.2
70.1
72.9
78.6
81.5
84.4
87.2
2-Theta scale
Figure 12- XRD results of catalysts from method 2
The XRD results for both methods look very similar. The variation in nickel weight load can
clearly be seen where catalysts 1.1 and 2.1 had the lowest nickel loading and catalyst 1.3 and
2.3 had the highest nickel loading. The difference in preparing the catalyst by method 1 and
method 2 was however the preparation of the potassium tungstate, K2WO4. From the XRD
33
results above it can therefore be hard to see any difference in the two methods. In the
appendix under section 7.1 in Figure 47 and Figure 48 XRD results of the K2WO4 attained
from both methods can be found. If one analyse these results it can be seen that K2WO4 with a
higher purity is formed from method 2 where the peaks overlap better. The purity is in fact
how much of the reactants that have reacted to K2WO4 or how complete the reactions from
method 1 and 2 are.
XRD results of the bronzes, the support and the finished catalyst is also found in the appendix
under section 7.1. It can clearly be seen from these results that the wanted Br-ZrO2
impregnated with different weight loads of nickel catalysts had been prepared. There might be
some impurities in the catalyst, especially from the reactions in Eq.2.14, Eq.2.15 and Eq.2.16
where K2WO4 and WO2 was formed since the reactions are not complete. From Shurdumov et
al. the conversion of K2WO4 with K2CO3-WO3 (Eq.2.14) is only 61 % after 50 minutes with a
decreasing rate [31]. Also when reducing WO3 to WO2 with hydrogen other oxides may also
have formed such as WO2.96, WO2.90 and WO2.72 [32]. When reducing the WO3 to WO2 the
temperature is very important as well as the H2/Ar ratio.
The idea was to create the bronzes with a potassium content of x=0.25 in KxWO3 but other
values of x might also have formed. It can be seen from the XRD results that K 0.20WO3 and
K0.32WO3 also had formed during the calcination. For x=0.25 the precise amount of potassium
must be added and spread homogenously over the support. In other words it is reasonable to
assume that a variation of x-values have formed in the bronzes, both lower and higher than
0.25.
The XRD results also gave information on the structure of the catalyst. From Figure 49 and 50
in the appendix it can clearly be seen that the wanted hexagonal tungsten bronzes had been
prepared. For the structure of zirconia it was to obtain the tetragonal form and from Figure 59
it is confirmed that t-ZrO2 has been prepared.
The IWI method had to be repeated several times to impregnate the correct amounts of nickel
to the support due to the low pore volume. This could affect the structure and how well the
nickel was spread homogenously inside the pores of the support. This was mainly a problem
for the higher nickel loading catalysts, 10- and 15 %. However this could not be detected on
the XRD results and should be further studied with transmission electron microscopy or
scanning electron microscopy.
34
4.2 Experimental results and discussion
4.2.1 Part 1
In part 1 a steam to carbon ratio of 4 was used. The GHSV was higher compared to part 2
because of the much larger N2 and H2 flows which can be seen in the table 16 below.
Table 16 – Experimental information on Part 1 and Part 2.
4.2.1.1 GC results
S/C=4, Method 1
0.8
Permanent carbon gases produced %
0.7
0.6
0.5
5 wt%Ni (No alkali)
0.4
5 wt%Ni
0.3
10 wt%Ni
0.2
0.1 15 wt%Ni
0
0 1 2 3 4
Time (hours)
Figure 13- Permanent carbon gases produced, CO, CO2, CH4 and C2H4, of total carbon in with a S/C ratio 4. The
reactor temperature of the first 2 hours was 700°C and the last 2 hours was 800°C.
35
S/C=4, Method 2
0.8
0.7
0.6
0.5
5 wt%Ni (No alkali)
0.4
5 wt%Ni
0.3
10 wt%Ni
0.2
0.1 15 wt%Ni
0
0 1 2 3 4
Time (hours)
Figure 14- Permanent carbon gases produced, CO, CO2, CH4 and C2H4, of total carbon in with a S/C ratio 4. The
reactor temperature of the first 2 hours was 700°C and the last 2 hours was 800°C.
In Figure 13 and Figure 14 the amount of permanent gases produced, regarding carbon was
calculated. The amount of CO, CO2, CH4 and C2H4 produced was compared with the total
amount of C1 into the reactor. The conversion of carbon to permanent gases was low, less
than 1 %. It increased slightly when the temperature increased from 700°C to 800°C.
A reason for the low conversion could be the high GHSV. The GC was unable to detect such
low concentrations due to the high dilution of N2.
S/C=4, Method 1
100
90
80
H2 selectivity %
70
60
50 5 wt%Ni
40 10 wt%Ni
30
20 15 wt%Ni
10
0
0 1 2 3 4
Time (hours)
Figure 15- H2 selectivity, mol produced H2 / mol produced permanent gases. The reactor temperature of the first 2
hours was 700°C and the last 2 hours was 800°C.
36
S/C=4, Method 2
100
90
80
H2 selectivity %
70
60
50 5 wt%Ni
40 10 wt%Ni
30
20 15 wt%Ni
10
0
0 1 2 3 4
Time (hours)
Figure 16- H2 selectivity, mol produced H2 per mol produced permanent gases. The reactor temperature of the first 2
hours was 700°C and the last 2 hours was 800°C.
In Figure 15 and Figure 16 the H2 selectivity is shown. The selectivity is based on the amount
of hydrogen produced compared to the total amount of permanent gases produced. It can be
seen that the selectivity is almost 100 %, but it decreases slightly from 700°C to 800°C. The
reason for this is that at 700 °C almost no CO, CO2, CH4 or C2H4 is produced.
The amount of hydrogen produced is derived from the outlet of hydrogen measured by the GC
minus the inlet of hydrogen to the electrical furnace. The mass flow controller for hydrogen
and the GC measurements might have generated measuring errors.
S/C=4, Method 1
100
90
80
CO selectivity %
70
60 5 wt%Ni (No alkali)
50
40 5 wt%Ni
30 10 wt%Ni
20
15 wt%Ni
10
0
0 1 2 3 4
Time (hours)
Figure 17- CO selectivity. The reactor temperature of the first 2 hours was 700°C and the last 2 hours was 800°C.
37
S/C=4, Method 2
100
90
80
CO selectivity %
70
50
40 5 wt%Ni
30 10 wt%Ni
20
15 wt%Ni
10
0
0 1 2 3 4
Time (hours)
Figure 18- CO selectivity. The reactor temperature of the first 2 hours was 700°C and the last 2 hours was 800°C.
Figure 17 and Figure 18 shows the CO selectivity of the catalyst from part 1. The selectivity
was calculated from the amount CO produced over the amount CO, CO2, CH4 and C2H4
produced. There was no CO selectivity the first 2 hours because almost no permanent gases
were produced at 700 °C.
S/C=4, Method 1
100
90
80
CH4 selectivity %
70
50
40 5 wt%Ni
30 10 wt%Ni
20
15 wt%Ni
10
0
0 1 2 3 4
Time (hours)
Figure 19- CH4 selectivity. The reactor temperature of the first 2 hours was 700°C and the last 2 hours was 800°C.
38
S/C=4, Method 2
100
90
80
CH4 selectivity %
70
50
40 5 wt%Ni
30 10 wt%Ni
20
15 wt%Ni
10
0
0 1 2 3 4
Time (hours)
Figure 20- CH4 selectivity. The reactor temperature of the first 2 hours was 700°C and the last 2 hours was 800°C.
From Figure 19 and Figure 20 the CH4 selectivity is shown. As mentioned before at 700 °C
almost no permanent carbon gases were produced. The CH4 selectivity was calculated from
the amount of CH4 produced over the amount of CO, CO2, CH4 and C2H4 produced.
4.2.1.2 SPA results
S/C=4, Method 1
10
9
8
Tar reduction %
7
5
4 5 wt%Ni
3 10 wt%Ni
2
15 wt%Ni
1
0
700 800
Temperature (°C )
Figure 21- Tar reduction from SPA samples at 700 °C and 800 °C after the reactor. Two SPA samples were taken at
each temperature and the result in the graph is the average of these two.
39
S/C=4, Method 2
10
9
8
Tar reduction %
7
5
4 5 wt%Ni
3 10 wt%Ni
2
15 wt%Ni
1
0
700 800
Temperature (°C )
Figure 22- Tar reduction from SPA samples at 700 °C and 800 °C after the reactor. Two SPA samples were taken at
In Figure 21 and 22 the tar reduction from SPA analysis is shown. The tar reduction is the
amount of reduced tars compared to the total tar. From the results it can be seen that the tar
reduction is low but the trend is that it increases with temperature. 1-MN mainly reduced to
naphthalene but it was also reduced to other tars such as benzene, toluene, m/p xylene, indan,
indene, 2-methylnaphthalene and acenaphthene.
S/C=4, Method 1
120
Ratio 1-MN/Naphthalene
100
80
5 wt%Ni (No alkali)
60
5 wt%Ni
40
10 wt%Ni
20 15 wt%Ni
0
700 800
Temperature (°C )
Figure 23- Ratio of 1-MN over Naphthalene from SPA samples after the reactor. Two SPA samples were taken at
40
S/C=4, Method 2
140
120
100
60 5 wt%Ni
40 10 wt%Ni
20 15 wt%Ni
0
700 800
Temperature (°C )
Figure 24- Ratio of 1-MN over Naphthalene from SPA samples after the reactor. Two SPA samples were taken at
Since mainly naphthalene was found in the SPA samples the ratio of 1-MN over naphthalene
was calculated and is shown in Figure 23 and Figure 24. A high ratio indicates that the tar
reduction is low. The general trend is that the ratio decreases, but for Method 1, 5 wt% Ni,
Method 2, 5 wt% Ni and Method 2, 15 wt% Ni the ratio increases. Tar reduction is favoured
by an increase in temperature and the ratio of 1-MN over naphthalene should decrease at
higher temperatures.
The reason for a decrease in tar reduction and a higher ratio of 1-MN/naphthalene when the
temperature increased could be from errors collecting the SPA samples.
Because of the low amount of permanent gases produced and tar reduction it was hard to
analyse the results and compare the effect of alkali. The results from the GC look similar for
all the experiments where the catalysts show little or no activity. However, the trend is that
with an increasing temperature the tar reduction and amount of permanent gases produced
increased. The catalysts with highest nickel loading generated slightly more permanent gases
and had a higher CO selectivity.
The aim with this report was to study the effect of alkali at different S/C ratios and
temperatures and not to achieve as high tar conversion as possible but to study and compare
the effect of alkali more easily it was decided to conduct the experiments at better conditions
for the catalyst. To obtain better conditions it was decided to lower the GHSV.
4.2.2 Blank test

A blank test was performed as a reference by running the test rig at the same conditions
without a catalyst. From this reference the activity, GC results and SPA results could be
compared.
41
4.2.2.1 GC results
The results from the GC are found in Figure 25 and show the permanent carbon gases
produced.
Blank test
0.8
0.7
0.6
0.5
0.4
0.3 Blank test
0.2
0.1
0.0
0 1 2 3 4 5
Time (hours)
Figure 25- Permanent gases produced with blank test. The reactor temperature of the first 2.5 hours was 700 °C and
the last 2.5 hours 800 °C.
The results show a low permanent carbon gases produced. This is comparable to the results
from Part 1 which indicates that the GHSV is too high and the product gas flow is too diluted.
Due to the high GHSV the tar passes the catalyst bed without being reduced.
4.2.2.2 SPA results

SPA samples collected from the blank test is presented it Figure 26 and 27 below.
Blank test
10
9
8
Tar reduction %
7
6
5
4 Blank test
3
2
1
0
700 800
Temperature (°C)
Figure 26-Tar reduction from SPA samples on blank test.
42
Blank test
60
Ratio 1-MN/naphthalene
50
40
30
20 Blank test
10
0
700 800
Temperature (°C)
Figure 27-Ratio of 1-MN/naphthalene from SPA samples on blank test.
The tar reduction and ratio of 1-MN/naphthalene is similar to the results from Part 1.
4.2.3 Part 2
After obtaining the results in Part 1 it was decided to lower the flows in order to achieve a
higher tar reduction and thereby generate more permanent gases.
4.2.3.1 GC results
To obtain a lower GHSV a bleed off had to be installed after the alkali generator. From using
this bleed off new problems occurred. Now there was one inlet stream and two outlet streams,
which meant that the total flow in the system was not stable and fluctuated. This had to do
with the pressure drop over the catalyst bed and therefore the flow over the reactor was not
constant. A sketch of the system is shown in Figure 28 below.
43
Figure 28- Simple sketch of bleed off installed after the alkali aerosol generator.
One way to avoid this problem could be to introduce an inert with a known volumetric flow
rate, for example argon, to the GC and from this the flows in ml/min could be obtained for all
the flows. The known argon flow should be introduced and mixed with the main stream after
the reactor. Even if the bleed off was not used, more accurate values for permanent gases
produced and H2 selectivity could be acquired if this method was used. Another way to avoid
using a bleed of could be to install another alkali generator with a lower N2 demand compared
to the one used in the experiments.
However, due to time limitation this could not be tested and the amount of permanent carbon
gases produced and H2 selectivity could only be calculated for experiment 9 and 10 where no
alkali generator was used. The results are presented in Figure 29, 30, 31 and 32 below.
44
S/C=6, Method 1 and 2
1.4
1.2
1
0.8
Method 1, 10 wt%Ni
0.6 (No alkali)
0.4 Method 2, 10 wt%Ni
(No alkali)
0.2
0
0 1 2 3 4 5
Time (hours)
Figure 29- Permanent carbon gases produced, CO, CO2, CH4 and C2H4, of total carbon in with a S/C ratio 6 with no
alkali.

80
70
60
H2 selectivity %
50
40 Method 1, 10 wt%Ni
(No alkali)
30
Method 2, 10 wt%Ni
20
(No alkali)
10
0
0 1 2 3 4 5
Time (hours)
Figure 30- H2 selectivity, mol produced H2 / mol produced permanent gases with no alkali.
45
S/C=6, 20%Br-ZrO2 impregnated with Nickel
100
90
80
CO selectivity %
70
60
40 (No alkali)
20 (No alkali)
10
0
0 1 2 3 4 5
Time (hours)
Figure 31- Comparison in CO selectivity for Method 1 and 2, 10 wt%Ni with no alkali.

100
90
80
CH4 selectivity %
70
60
40 (No alkali)
20 (No alkali)
10
0
0 1 2 3 4 5
Time (hours)
Figure 32- Comparison in CH4 selectivity for Method 1 and 2, 10 wt%Ni with no alkali.
A nickel weight load of 10 % was used for these two experiments. Catalyst Method 1, 10 %Ni
had an experimental time of 4 hours, first 2 hours at 700 °C and last two hours at 800 °C.
Catalyst Method 2, 10 %Ni had an experimental time of 5 hours, first 2.5 hours at 700 °C and
last 2.5 hours at 800 °C.
The trend with an increase in the amount permanent gases generated at temperature increase
from 700 °C to 800 °C is seen in Figure 29. However the H2 selectivity is increasing with
temperature at a S/C ratio of 6 and at lower GHSV compared to results from S/C ratio of 4.
The CO selectivity is favoured for catalysts prepared from Method 1 and the CH 4 selectivity
is favoured from Method 2.
46
The CO selectivity was only around 20 % and seemed stable while the CH4 selectivity was a
bit higher, 30-40 %.
The tar reduction was still very low and therefore it was decided to change more parameters
for the remaining experiments. These experiments will be presented in a different way where
they will be compared one at a time.
From table 14 above in section 3.2.3.3 it can be seen that the 1-MN flow was changed from
0.1 to 0.03 g/min for experiments 11-17. This was done in order to increase the catalyst load
versus tar flow. Roberto Coll et al. studied the naphthalene catalytic steam reforming and they
concluded that lower naphthalene flows and a higher catalyst loading was necessary to
achieve measurable conversions [16]. The catalyst loading was also increased from 0.5 g to 1.0
g for experiments 13-17. The CO and CH4 conversions calculated from the GC analysis of
experiment 11 and 13 are presented in Figure 33 and 34 below.

100
90
80
CO selectivity %
70
60
50
20
10
0
0 1 2 3 4 5
Time (hours)
Figure 33- Comparison in CO selectivity for Method 1 and 2, 5 wt%Ni. The reactor temperature of the first 2.5 hours
was 700°C and the last 2.5 hours was 800°C.
47
100
90
80
CH4 selectivity %
70
60
50
20
10
0
0 1 2 3 4 5
Time (hours)
Figure 34- Comparison in CH4 selectivity for Method 1 and 2, 5 wt%Ni. The reactor temperature of the first 2.5 hours
was 700°C and the last 2.5 hours was 800°C.
In Figure 33 and 34 the 1-MN flow was reduced to 0.03 g/min. For catalyst Method 2, 5
wt%Ni the catalyst load was increased from 0.5 g to 1.0 g.
The reason for the short experimental time for Method 2, 5 wt%Ni is that with the increased
catalyst load the reactor got blocked from coke formation. The coke formation was visible
after the experiment when the catalyst was removed. There seemed to be a lot of coke
formation in every test and it was more obvious when the catalyst load was increased.
Another problem could have been the vaporisation of 1-MN. In the system 1-MN was
introduced to the primary stream after the preheater. It is known that 1-MN can be hard to
vaporise and therefore some 1-MN may enter the reactor in liquid state due to non-complete
evaporation. This might be solved by changing how 1-MN is introduced to the primary
stream. A solution could be to use a spray system, so that 1-MN is mixed with steam at high
temperatures which should affect the amount of vaporised 1-MN. This was however not
tested and should be analysed in the future.
The trend for CO selectivity is that it stabilizes at around 40 % at 800 °C and the same trend
can be seen for CH4 selectivity. It is difficult to say anything of how the increased catalyst
load affects the CO and CH4 ratio, but at 700 °C the results look similar to the experiment
with a lower catalyst load. From the GC it seemed that the amount of permanent gases
generated increased with a higher catalyst loading but the exact values could not be calculated
due to the bleed off system.
In experiment 14 the alkali concentration was reduced from 274.8 ppm to 137.5 ppm in
solution to see how this affected the CO and CH4 selectivity. The result is presented in Figure
35 and 36 below.
48
S/C=6, Method 1
100
90
80
CO selectivity %
70
60
50
30
20
10
0
0 1 2 3 4
Time (hours)
Figure 35- CO selectivity for Method 1, 10 wt%Ni with a lower alkali concentration. The reactor temperature of the
first 2.5 hours was 700 °C and then 800 °C
S/C=6, Method 1
100
90
80
CH4 selectivity %
70
60
50
30
20
10
0
0 1 2 3 4
Time (hours)
Figure 36- CH4 selectivity for Method 1, 10 wt%Ni with a lower alkali concentration. The reactor temperature of the
first 2.5 hours was 700 °C and then 800 °C.
The weight load of nickel for this experiment was 10 %. The results are similar to Figure 33
and Figure 34 which indicates that the reduction of alkali aerosols did not have any effect of
the catalyst selectivity toward CO and CH4.
Experiment 15 was only tested at a temperature of 800 °C to see how this affected the CO and
CH4 selectivity. The results are presented in Figure 37 and 38 below.
49
100
90
80
CO selectivity %
70
60
50
30
20
10
0
0 1 2 3 4
Time (hours)
Figure 37- CO selectivity for Method 2, 10 wt%Ni. The reactor temperature was 800 °C.

100
90
80
CH4 selectivity %
70
60
50
30
20
10
0
0 1 2 3 4
Time (hours)
Figure 38- CH4 selectivity for Method 2, 10 wt%Ni. The reactor temperature was 800 °C.
The CO and CH4 selectivity is more stable in this experiment, at around 50 % and between
30-40 % respectively.
The CO and CH4 for the last two experiments, 16 and 17, are presented in Figure 39 and 40
below. For experiment 17, a temperature increase to 900 °C was tested to see how this
affected the results.
50
100
90
80
CO selectivity %
70
60
50
20
10
0
0 1 2 3 4
Time (hours)
Figure 39- Comparison in CO selectivity for Method 1 and 2, 15 wt%Ni. The reactor temperature of the first hour
was 700°C, the second hour 800°C and the third hour 900 °C.

100
90
80
CH4 selectivity %
70
60
50
20
10
0
0 1 2 3 4
Time (hours)
Figure 40- Comparison in CH4 selectivity for Method 1 and 2, 15 wt%Ni. The reactor temperature of the first hour
was 700°C, the second hour 800°C and the third hour 900 °C.
The experiment with Method 1, 15 wt% had to be cancelled at around 1 hour since the reactor
got blocked again and this was a continuous problem with a higher catalyst load.
Coke formation was not the only thing that affected how long time it took before the reactor
got blocked. How the catalyst was loaded in the quarts tube had a large impact on this. The
design of the quarts tube could also be changed to avoid this problem. In these experiments
the tube had the same diameter above the catalyst bed but below the bed the diameter got
smaller. With this design the pressure increased over the catalyst bed and it might also have
some effect on the blocking of the reactor. For future tests maybe a quarts tube with the same
diameter all over should be tested.
51
From the results of Part 2 it can be seen that the CO selectivity increases with higher nickel
loading, lower GHSV and higher catalyst loading. The CO selectivity was also higher for
experiments with alkali aerosols compared to no alkali aerosols. However, changing the
parameters to obtain lower GHSV and higher catalyst load can also have affected this. This
trend can also be seen for the CH4 but in this case the selectivity is decreasing with the
amount of Ni load.
The GC results showed an increase in produced permanent gases with a lower GHSV,
however the exact amount could not be calculated. In the last experiment a temperature of 900
°C was also tested. At this temperature there was a large increase in the amount of permanent
gases produced indicating that the catalyst works better at higher temperatures. However, the
coke formed and the alkali aerosols also have catalytic effects which might contribute to
higher tar reduction and permanent gases produced.
4.2.3.2 SPA results

The SPA results for Part 2 with a S/C ratio of 6 is presented in Figure 41, 42, 43 and 44.
S/C=6, Method 1
50
45
40
Tar reduction %
35
25
20 5 wt%Ni
15 10 wt%Ni
10
15 wt%Ni
5
0
700 800
Temperature (°C )
Figure 41- Tar reduction from SPA results at temperatures of 700 °C and 800 °C.
52
S/C=6, Method 2
50
45
40
Tar reduction %
35
30
20 10 wt%Ni
15
10 15 wt%Ni
5
0
700 800 900
Temperature (°C )
Figure 42- Tar reduction from SPA results at temperatures of 700 °C, 800 °C and 900 °C.
S/C=6, Method 1
250
200
100 5 wt%Ni
10 wt%Ni
50
15 wt%Ni
0
700 800
Temperature (°C )
Figure 43- Ratio 1-MN/Naphthalene varying at temperatures of 700 °C and 800 °C.
53
S/C=6, Method 2
70
60
50
40
10 wt%Ni (No alkali)
30
10 wt%Ni
20
15 wt%Ni
10
0
700 800 900
Temperature (°C )
Figure 44- Ratio 1-MN/Naphthalene varying at temperatures of 700 °C, 800 °C and 900 °C.
From the SPA samples a clear trend can be seen that at higher temperatures, lower GHSV,
higher catalyst load and higher Ni weight load the tar reduction increases and the ratio of 1-
MN/Naphthalene decreases. Another more interesting trend that can be seen is that
experiments with catalyst from Method 2 seem to have a higher tar reduction and lower ratio
of 1-MN/naphthalene then catalysts from Method 1. This can be an indication to that a higher
purity of bronzes can be achieved if synthesised with Method 2 instead of Method 1.
At temperatures of 900 °C for catalyst Method 2, 15 wt%Ni it can be seen that almost 40 %
tar reduction is reached and the ratio of 1-MN/naphthalene is close to 2.
If compared with the results from S/C ratio 4 from Part 1 the tar reduction in Part 2 in general
are higher and the ratios of 1-MN/naphthalene are lower. However, a lot of parameters were
changed in Part 2 to lower the GHSV.
4.2.4 Alkali loss/gain before and after experiments in the catalyst and its
effect on the results.
Before and after measurements were made to see how the alkali varied in the catalyst. This
was an alternative solution to replace the surface ionization technique. The measurements
were done with atomic absorption spectrophotometry, AAS, and the instrument used was a
Perkin-Elmer 1100 B. The amount of alkali in the catalysts before the experiments was
calculated from the stoichiometric balances and the weight load of catalyst used in each
experiment. The amount of alkali after the experiments was measured with AAS. This was
only done for Part 2 of the experiments.
The results are presented in table 17 and Figure 45 and Figure 46 below.
54
Table 17- Mass of potassium in the catalyst before and after the experiments in Part 2.
Catalyst Before Experiment After experiment Difference

Mass K [mg] Mass K [mg] K loss [%]
Method 1, 10 wt%Ni 3.70 2.75 25.45
(No alkali)
Method 2, 10 wt%Ni 3.58 3.47 3.16
(No alkali)
Method 1, 5 wt%Ni 3.79 3.01 20.56
Method 2, 5 wt%Ni 7.74 6.37 17.71
Method 1, 10 wt%Ni 7.29 5.21 28.59
Method 2, 10 wt%Ni 7.27 6.61 9.17
Method 1, 15 wt%Ni 7.46 5.40 27.65
Method 2, 15 wt%Ni 6.92 6.47 6.53
S/C=6, Method 1
10
9
8
Mass potassium [mg]
7
6
5 5 wt%Ni
4 10 wt%Ni
3
15 wt%Ni
2
1
0
Before experiment After experiment
Figure 45- Mass in the catalyst before and after the experiments in Part 2.
S/C=6, Method 2
10
9
8
Mass potassium [mg]
7
6
5 5 wt%Ni
4 10 wt%Ni
3
15 wt%Ni
2
1
0
Before experiment After experiment
Figure 46- Mass in the catalyst before and after the experiments in Part 2.
55
The results from the AAS show a decrease of potassium in each catalyst for all the
experiments. However, the catalyst samples taken for AAS and measured after the experiment
might have contained carbon and also other compounds may have formed on the catalyst.
This might have affected the results in such way that a lower mass of potassium was
measured in the AAS compared to the actual mass.
A trend can be seen from the results that the alkali decrease is higher in catalyst prepared from
Method 1 compared to Method 2. What’s interesting about this is that from the SPA samples
the results from Method 2 showed a higher tar reduction and a lower 1-MN/naphthalene ratio.
This correlation indicates that catalysts prepared from Method 2 are more preferable than
Method 1. To better understand how the potassium loss affects the tar reduction and
deactivation of the catalyst, longer experiments at constant temperatures have to be
conducted. There was no tendency for deactivation in the experiments because of the short
experimental time, or none that could be observed.
Alkali in the aerosol also seemed to have a positive effect on the tar reduction properties seen
from the SPA results. From literature, alkali aerosol can act as a poison for nickel catalysts
and can contribute to the deactivation of the catalyst but because of the short experimental
time this was not observed. It seems that the structure of the hexagonal bronzes successfully
protects the potassium in the bulk from evaporating
In Part 2 the tar reduction was higher compared to the blank test and at 900 °C almost 40 %
tar reduction was reached. At low nickel weight loading almost no activity was observed but
for 10 wt% and especially 15 %wt the activity increased. It is known that for thermal cracking
to occur at temperatures of at least 1000 °C is needed which indicates that the catalyst is
active. Steam tar reforming is normally conducted at higher temperatures (850-1000 °C) and
for future work the tar steam reforming properties of the catalyst should be analysed at higher
temperature compared to those used in this study.
56
5 Conclusion
The characterisation of the catalysts gave information of that the wanted catalyst and structure
had been prepared for both methods. There was a difference in the purity of bronzes prepared
however and from characterisation it proved that Method 2 was more favourable since it
showed a higher yield of the K2WO4 produced from Eq.2.16 compared to Method 1. From the
XRD analysis it could be seen that the support was prepared with the hexagonal structure of
the bronzes and the tetragonal structure of zirconia. This concludes that the catalysts had been
prepared correctly. From the BET results it was found that the Surface area and pore volume
of the catalyst was rather low as expected. The Surface area varied between 40 and 46 m 2/g
and was slightly higher for Method 2. Hence the purity of the bronzes prepared affects the
surface area positively.
In the experimental results from Part 1 where a S/C ratio of 4 was used, there was almost no
tar reformation. It is therefore recommended to use a lower GHSV if these experiments were
to be continued. Some trends could be detected where the permanent carbon gases produced
and the tar reduction increased with temperature and the ratio 1-MN/naphthalene decreased
with temperature despite the high GHSV. With a lower GHSV, as in Part 2, the trends became
clearer. By installing a bleed off the GHSV could be reduced. The negative effect of installing
the bleed off was that the flow over the reactor was not stable and the volumetric flow rate
after the reactor could not be calculated. This could in the future be solved by using another
alkali generator with a lower N2 demand or by introducing a known internal standard to the
GC. When the GHSV decreased the catalyst yielded higher tar reduction and permanent gases
produced. With less N2 dilution the contact time of 1-MN with the catalyst increased and also
the micro-GC’s was able to detect the different compounds.
By increasing the catalyst load for the experiments it the amount of permanent gases
generated increased. With the increased catalyst loading the reactor had a tendency to get
blocked from coke formation which was visible after the experiment. The coke formed and
alkali aerosols could also contribute to the increase of permanent gases produced and the
higher tar reduction.
From the test conducted at 900 °C the catalyst showed much more activity with a large
increase in tar reduction and lower 1-MN/naphthalene ratio. The amount of produced
permanent gases also increased a lot, however this could not be calculated.
A clear trend was found that the tar reduction increased and the 1-MN/naphthalene ratio
decreased with a higher nickel loading. It was also found that catalyst from Method 2 showed
a higher tar reduction and lower 1-MN/naphthalene ratio. This also indicates that a higher
purity was obtained in the bronzes for Method 2 compared to Method 1 that was proposed
from the XRD results.
The results from the AAS showed that the mass of alkali in the catalyst before the experiment
decreased between 3-29 % when measured after the experiment. Of interest is that catalysts
from Method 1 showed a significantly higher decrease in mass of alkali compared to Method
57
2. This indicates that a more stabled support was synthesised from Method 2 where the
potassium was protected better, compared to Method 1. Catalysts prepared by Method 2 also
showed higher tar reduction properties from experiments conducted in Part 2.
Introducing alkali aerosol to the reactor also had a positive effect on the tar reduction.
However, the experimental time was short and longer tests have to be conducted to investigate
the deactivation time from alkali poisoning and coke formation.
58
6 Further improvements
Based from the results a few changes are suggested for future testing of the catalyst. A low
GHSV was favourable for the tar reduction and for future tests it could be decreased even
more. A higher catalyst loading compared to 1-MN flow could also be tested. A higher tar
reduction and conversion could make it easier to compare how the alkali affected the catalyst,
even if a high tar reduction was not the main goal in this study. Since the catalyst showed low
tar steam reforming properties at low reactor temperatures and nickel loadings, the catalyst
should be tested at higher temperatures and lower GHSV for future tests.
To avoid the reactor from being blocked by coke forming it is suggested that the design of the
reactor tube should be changed so that the diameter of the tube is the same all over so that less
pressure build up occur over the catalyst bed. To avoid the large amount of coke formation
formed a spray system could be installed for the 1-MN inlet to the primary stream. This would
generate a better mixing of the streams and hopefully more 1-MN will be vaporised before
reaching the catalyst bed.
59
7 References
[1] A. V. Bridgewater, Chem. Eng. J. 2003, 91, 16.
[2] H. A. M. Knoef, Handbook: Biomass Gasification, Biomass Technology Group, 2005, p.
[3] K. Engvall, H. Kusar, K. Sjöström and L. J. Pettersson, Top. Catal. 2011, 54, 949-959.
[4] P. McKendry, Bioresour. Technol. 2001, 83, 37-46.
[5] C. Higman and M. v. d. Burgt, Gasification, Gulf Professional Publishing, 2008, p.
[6] S. Karimipour, R. Gerspacher, R. Gupta and R. J. Spiteri, Fuel 2013, 103, 308-320.
[7] A. Godwin, Chukwu. in Study of transportation of GTL products from alaskan North
slope (ANS) to markets, Vol. University of Alaska Fairbanks, United States, 2002, p. 317.
[8] C. Li and K. Suzuki, Renewable and Sustainable Energy Reviews 2009, 13, 594-604.
[9] T. A. Milne and R. J. Evans, National Renewable Energy Laboratory 1998, 204.
[10] D. C. Elliott, Pyrolysis Oils from Biomass, American Chemical Society, 1988, p.
[11] E. G. Baker, M. D. Brown, D. C. Elliott and L. K. Mudge, Pacific Northwest Laboratory
1988, 14.
[12] P. Basu, Biomass Gasification and Pyrolysis: Practical Design, Elsevier Inc., 2010, p.
[13] L. Devi, K. J. Ptasinski and F. J. J. G. Janssen, Biomass Bioenergy 2003, 24, 16.
[14] I. Aigner, U. Wolfesberger and H. Hofbauer, Environmental Progress & Sustainable
Energy 2009, 28, 372-379.
[15] S. A. Nair, K. Yan, A. J. M. Pemen, E. J. M. van Heesch, K. J. Ptasinski and A. A. H.
Drinkenburg, Ind. Eng. Chem. Res. 2004, 43, 1649- 1658.
[16] R. Coll, J. Salvadó, X. Farriol and D. Montané, Fuel Process. Technol. 2001, 74, 19-31.
[17] J. Long, H. Song, X. Jun, S. Sheng, S. Lun-Shi, X. Kai and Y. Yao, Bioresour. Technol.
2012, 116, 278-284.
[18] J. A. Moulun and F. Kapteijn, Carbon 1995, Volume 33, 11.
[19] D. Sutton, B. Kelleher and J. R. H. Ross, Fuel Process. Technol. 2001, 73, 19.
[20] Z. A. El-Rub., E. A. Bramer and G. Brem, Ind. Eng. Chem. Res. 2004, 43, 9.
[21] D. Dayton, National Renewable Energy Laboratory 2002, 33.
[22] E. Salaices in Catalytic steam gasification of biomass surrogates: A thermodynamic and
kinetic approach Vol. The University of Western Ontario, London, Ontario, Canada, 2010.
[23] A. Hussain and L. Kihlborg, Acta Cryst. 1975, 9.
[24] A. Hussain, Acta Chemica Scaninavica 1978, 6.
[25] P. G. Dickens and M. S. Whittingham, Transaction of the Faraday Society 1964, 61, 6.
[26] S. Raj, T. Sato, S. Souma, T. Takahashi, D. Sarma, P. Mahadevan, J. Campuzano, M.
Greenblatt and W. McCarroll, Physical Review B 2008, 77.
[27] D. G. Barton, S. L. Soled and E. Iglesia, Top. Catal. 1998, 6, 87-99.
[28] W. Ji, J. Hu and Y. Chen, Catal. Lett. 1998, 53, 15-21.
[29] A. M. Abdullah, T. Debnath, C. H. Rüscher and A. Hussain, Journal of Scientific
Research 2012, 4.
[30] M. M. R. Shakil, T. Debnath, C. H. Rüscher and A. Hussain, Journal of Bangladesh
Chemical Society 2012, 25, 8.
[31] G. K. Shurdumov, Z. V. Shurdumova and Z. A. Cherkesov, Russian Journal of Inorganic
Chemistry 2009, 54, 137-140.
[32] M. I. Zaki, N. E. Fouad, S. A. A. Mansour and A. I. Muftah, Thermochim. Acta 2011,
523, 90-96.
[33] G. Stefanic, S. Popovic and S. Music, Thermochim. Acta 1995, 259, 225-234.
[34] W. D. Nesse, Introduction to Mineralogy, 2011, p.
[35] N. S. Charles, Chemical engineering series, McGraw-Hill inc., United States of America,
1980, p.
60
[36] J. R. Anderson and K. C. Pratt, Introduction to characterisation and testing of catalysts,
Academic Press, 1985, p.
[37] J. F. Richardson and J. H. Harker, Coulson and Richardson's Chemical Engineering:
Particle Technology and Separation Processes, Elsevier Ltd., 2002, p.
[38] H. Kawano, International Journal of Mass Spectrometry and Ion Physics 1981, 50, 33.
[39] H. Kawano, Prog. Surf. Sci. 2008, 83, 1-165.
[40] R. L. Myers and W. L. Fite, Environmental Science & Technology 1975, 9.
[41] K. O. Davidsson, K. Engvall, M. Hagström, J. G. Korsgren, B. Lönn and J. B. C.
Pettersson, Energy Fuels 2002, 16, 9.
[42] U. Jäglid, J. G. Olsson and J. B. C. Pettersson, J. Aerosol Sci 1996, 27, 11.
[43] A. Daniele, M. Monti and A. Baiker, J. Catal. 1983, 83, 323-335.
61
8 Appendix
Catalyst characterisation, XRD
Figure 47 – Preparation of K2WO4 from Method 1
62
Figure 48 – Preparation of K2WO4 from Method 2
63
Figure 49 – Preparation of K0.25WO3 from Method 1
64
Figure 50 –Preparation of K0.25WO3 from Method 2
65
Figure 51 – Preparation of 20%Br- ZrO2 from Method 1
66
Figure 52- Preparation of 20%Br- ZrO2 from Method 2
67
Figure 53 – Preparation of 20%Br- ZrO2 with 5wt% Nickel from Method 1
68
Figure 54 – Preparation of 20%Br- ZrO2 with 10wt% Nickel from Method 1
69
Figure 55 - Preparation of 20%Br- ZrO2 with 15wt% Nickel from Method 1
70
Figure 56- Preparation of 20%Br- ZrO2 with 5wt% Nickel from Method 2
71
72
73
Figure 59 – Preparation of ZrO2 from Zr(OH)4
74

Material DRX

Uploaded by

Copyright:

Available Formats

Material DRX

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Material DRX

Uploaded by

Copyright:

Available Formats

Catalytic tar reforming in

Supervisor: Angélica V. González, PHD

Examiner: Klas Engvall, professor

Master of Science Thesis

Royal institute of Technology, KTH

Department of Chemical Technology

Stockholm, Sweden 2014

1.1 Aim and delimitations

AAEM – Alkaline and alkali earth metal

AAS – Atomic absorption spectrophotometer

GHSV – Gas hourly space velocity

HTB – Hexagonal tungsten bronzes

ITB – Intergrowth tungsten bronzes

IWI – Incipient wetness impregnation

PAH – Polycyclic aromatic hydrocarbons

S/C – Steam to carbon

SPA – Solid Phase Adsorption

TPO- Temperature programmed oxidation

TPR – Temperature programmed reduction

TTB – Tetragonal tungsten bronzes

𝐵𝑖𝑜𝑚𝑎𝑠𝑠 + 𝑔𝑎𝑠𝑖𝑓𝑦𝑖𝑛𝑔 𝑎𝑔𝑒𝑛𝑡 + 𝑕𝑒𝑎𝑡 𝐶𝑂, 𝐶𝑂2 , 𝐻2 𝑂, 𝐻2 , 𝐶𝐻4 + 𝑜𝑡𝑕𝑒𝑟 𝑕𝑦𝑑𝑟𝑜𝑐𝑎𝑟𝑏𝑜𝑛𝑠

Drying process - Evaporation of moisture

2.1.1 Tar and tar reforming

𝑀𝑖𝑥𝑒𝑑 𝑃𝑕𝑒𝑛𝑜𝑙𝑖𝑐 𝐴𝑙𝑘𝑦𝑙 𝐻𝑒𝑡𝑒𝑟𝑜𝑐𝑦𝑐𝑙𝑖𝑐 𝐿𝑎𝑟𝑔𝑒𝑟

Figure 2 – Tar formation scheme, where PAH is polycyclic aromatic hydrocarbons

Table 1- Classification of tars.

Product class Tars

Tar can be decomposed in several ways: [8]

Thermal cracking: 𝑝𝐶 𝐻 𝑕𝑒𝑎𝑡 𝑞𝐶 𝐻 𝑟𝐻 (Eq. 2.9)

Steam reforming: 𝐶 𝐻 𝑛𝐻 𝑂 𝑛 𝐻 𝑛𝐶𝑂 (Eq. 2.10)

Dry reforming: 𝐶 𝐻 𝑛𝐶𝑂 ( )𝐻 𝑛𝐶𝑂 (Eq. 2.11)

Carbon formation: 𝐶 𝐻 𝑛𝐶 ( ) 𝐻 (Eq. 2.12)

Figure 3 – Primary and secondary method of tar reforming

2.1.2 Alkali compounds and its effect on biomass gasification

2.2.1 Choice of catalyst

2.2.1.1 The active phase

Figure 5 – Structure of hexagonal tungsten bronzes

2.2.2 Catalyst preparation

Zirconia can be obtained by thermal decomposition of amorphous zirconium hydroxide.

𝑟 𝑂𝐻 𝐻𝑒𝑎𝑡 𝑟𝑂 𝐻 𝑂 (Eq. 2.17)

2.2.3 Instruments used for characterisation of the catalyst

Figure 7 – Diffraction of X-rays from two parallel planes of atoms

2.3.1 GC for measurement of gases, hydrocarbons and low molecular

2.3.1.1 Instrumental set-up micro-GC

Micro- GC Detected compounds

2.3.2 Surface ionization technique for measurement of alkali levels

2.3.2.1 Instrumental set-up surface ionization

Catalyst number Method Nickel wt %

3.1.1 Preparation of support

Method Mass K2WO4 Mass WO3 Mass WO2 Mass K0.25WO3

𝑟 𝑂𝐻 𝐻𝑒𝑎𝑡 𝑟𝑂 𝐻 𝑂 (Eq. 2.17)

Method Mass K0.25WO3 Mass ZrO2 Total mass 20%Br-ZrO2

3.1.2 Nickel IWI method

Compound 5 wt% nickel 10 wt % nickel 15 wt % nickel

20%Br- ZrO2 Pore Volume [cm3/g] Surface area [m2/g]

3.2.1 Experimental set-up

Figure 9 – Detailed drawing of the test rig

Calibration standard Components

3.2.3 Activity evaluation