Fischer-Tropsch Syn thesis on Supported Cobalt- Based Catalysts: Influence of Various Preparation Methods and Supports on Catalyst Activity and Chain Growth Probability Thesis for obtaining the degree of Doktor der Naturwissenschaften (Dr. rer. nat.) of the Faculty of Chemistry Ruhr-Universität Bochum submitted by Diplom-Chemiker Martin Kraum Bochum 1999
158
Embed
Fischer Tropsch synthesis on supported cobalt based ... · PDF fileFischer-Tropsch Syn thesis on Supported Cobalt- Based Catalysts: Influence of Various Preparation Methods and Supports
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Fischer-Tropsch Syn thesis
on Supported Cobalt- Based Catalysts:
Influence of Various Preparation Methods and Supports
on Catalyst Activity and Chain Growth Probability
Thesis
for obtaining the degree of
Doktor der Naturwissenschaften (Dr. rer. nat.)
of the Faculty of Chemistry
Ruhr-Universität Bochum
submitted by
Diplom-Chemiker
Martin Kraum
Bochum 1999
II
Submitted at : 04. October 1999
Examination at : 05. November 1999
Board of examiners:
Chairman : Prof. Dr. H. Sander
Referee : Prof. Dr. M. Baerns
Co-Referee : Prof. Dr. M. Muhler
3rd Examiner : Prof. Dr. H. Sander
III
Für meine Eltern
Ich Danke Euch Für Alles !
Nicht weil es schwer ist wagen wir es nicht,
sondern weil wir es nicht wagen,
ist es schwer.
Seneca
IV
Some parts of the work here described were done between January 1996 and June
1996 in the laboratories of the chair "Technische Chemie" at Ruhr-Universität Bo-
chum. From July 1996 to December 1998 the work was carried on at the Institut für
Angewandte Chemie Berlin-Adlershof e.V. in fulfilment of the requirements for the
Ph.D. degree.
I wish to express my deep gratitude to my advisor, Professor Dr. M. Baerns, for his
encouragement and support, and for his willingness to entrust so much of the devel-
opment of the project to my judgement.
Further, I wish to appreciate Dr. O.V. Buyevskaja for her everlasting readiness for
help, for lots of fruitful discussions and for the very good co-operation.
I thank Dr. N. Steinfeldt for his kinetic studies on Fischer-Tropsch synthesis and Dr.
J.P. Müller for his development of the plasma-induced preparation technique for co-
balt catalysts. For their interpretation of TPR, XRD and XPS results, I have to thank
Dres. H. Bernd, M. Schneider and M. Pohl.
Many thanks to Mrs. R. Dambowsky for the ICP-OES analysis and to Mrs. S. Evert
and Mrs. K. Struwe for their support by carrying out the TPR measurements.
Furthermore, I thank all members of the Institut für Angewandte Chemie for the good
atmosphere and co-operation that was very helpful for the success of this thesis. A
special thanks goes to Mr. E. Ostrowski and his crew for technical support.
This work was supported by the European Commission (Contract no.: JOF3-CT95-
0016). The contract was administrated by Ruhr-Universität Bochum; the experimental
work was conducted at Institut für Angewandte Chemie Berlin-Adlershof e.V..
V
Abstract
1. Objective
As state-of-the-art in FISCHER-TROPSCH synthesis cobalt catalyst (Co-Ref) supported
on titania a chain growth probability α of 0.91 and a turn-over-frequency; i.e. the mol
of converted carbon monoxide divided by the mol of active metal per second), of
17.7·10-3 s-1 was reported. (Treac = 200 °C, ptot = 20 bar, H2:CO ratio of 2:11). The goal
of the present thesis was the development of a cobalt catalyst which reveals a higher
activity, i.e., an improvement of the turn-over-frequency (TOF), with an equally high
chain growth of ≈ 0.90.
2. Methods
Catalysts
The activity of cobalt catalysts used in Fischer-Tropsch synthesis is closely related to
the accessible cobalt surface area which depends on cobalt dispersion (DCo). Cobalt
dispersion can be affected by the preparation procedure. To increase Co dispersion
various preparation techniques were applied: incipient wetness-, precipitation-,
spreading and plasma- induced techniques. Furthermore, it was known from lit-
erature that the type of applied cobalt precursor influences dispersion and hence ac-
tivity and selectivity.. Based on this knowledge, six different cobalt precursors (cobalt
-acetate, -oxalate, -(II) + (III) acetyl acetonate, -EDTA and -hydroxide) were used in-
stead of the usually applied cobalt nitrate. Additional, three more catalysts were pre-
pared supported on CeO2, ZrO2 and TiO2 (pure rutile type) as alternate support mate-
rial to titania (Degussa P25; mixture of rutile and anatase) as it is known also that the
support influences the cobalt dispersion.
Characterisation
It was anticipated that the preparation technique and the kind of cobalt precursor in-
fluence the bulk and surface composition of the catalyst, the reducibility as well and
the dispersion of cobalt on the support. These possible effects were studied by
means of XRD, XPS, TPR and CO-pulsing over the catalyst. The total quantity of co-
balt of the catalyst sample was determined by ICP-OES. The interaction of the reac-
tants (H2 + CO) with the catalytic surface was examined by DRIFT and a transient
adsorption technique.
Catalytic Evaluation
The catalytic tests were carried out in a fixed-bed reactor at pre-set reaction condi-
tions: Treac = 200 °C, ptot = 20 bar, H2:CO:N2 = 12:6:2 bar, GHSVSTP = 1200 h-1. This
1 E. Iglesia, S.L. Soled, R.A. Fiato, J. Catal. 137 (1992) 212
VI
procedure allows an easy comparison of the obtained catalytic data (XCO, TOF, α)with the state-of-the-art cobalt catalyst (Co-Ref). Furthermore, the Fischer-Tropsch
reaction was carried out in a slurry phase to study the influence of the various reac-
tion conditions on carbon monoxide conversion and product distribution. The rate of
carbon monoxide consumption and methane formation was determined for a se-
lected catalyst in a gradientless recycle reactor (Berty- type).
3. Results
The catalytic results obtained at standard conditions in a fixed-bed reactor are given
in Table A1. An improvement of TOF was obtained on all catalysts. The most active
catalyst sample was SPR-OXA: XCO = 32.3 % and α = 0.81 as compared to Co-Ref
(XCO = 14.7 %, α = 0.83). The data in Table A1 show that TOF is affected by the ap-
plied cobalt precursor, the preparation technique and added promoter. Furthermore,
the kind of support material influenced the catalytic performance as well. On ceria
supported catalyst an improvement of XCO = 23.7 % was achieved in comparison to
Co-Ref.
Table A1: Overview of carbon monoxide conversion, selectivity towards C5+ fraction,
α−value, TOF and TOFnom of some selected catalysts applied in the present thesis
(average for steady-state conditions, Treac = 200 °C, ptot = 20 bar, H2:CO:N2 = 12:6:2,
GHSV = 1200 h-1)
Catalyst Cobalt precursor XCO
[%]
SC5+
[wt%]
α
[-]
DCored i
[%]
TOF j
[103s-1]
TOFnom k
[-]
Co-Ref a nitrate 14.7 80.0 0.83 6.1 17 1.0
IW-OXA-NH3 a,b oxalate 12.8 75.9 0.81 5.4 18 1.1
PR-EDTA-Ru c,d EDTA 15.8 81.5 0.84 6.5 26 1.5
IW-ACAC3 a acetyl acetonate 23.6 67.9 0.71 9.8 43 2.5
a Co-TY = (converted CO / metal s); metal = total amount of cobaltb Site-TY = (converted CO / surface metal s); surface metal = total amount of reduced cobalt as de-
rived from hydrogen chemisorptionc 473 K, 2000 kPa, H2 / CO = 2.1, XCO = 50-63%
2 State of the Art 19
Cobalt Supported on Zeolites
Zeolites, especially ZSM-5, are a preferred support when targeting to gasoline pro-
duction, because:
· its pore structure assists shape selectivities
· the acidic surface supports reactions like oligomerisation, cracking and aromati-
sation
· it is insensitive against cooking
· it is stable under FT conditions [108]
JONG and CHENG used ZSM-5 zeolites as support for cobalt based catalyst [109].
They prepared catalyst by incorporation of cobalt in the zeolite synthesis gel as well
as by precipitation/impregnation of cobalt oxide on the zeolites; in both cases highly
dispersed cobalt particles were obtained. The chemical nature of the obtained cobalt
particles is different as revealed by TPR-examinations. The main cobalt species for
the co-precipitated catalysts was substituted cobalt within the ZSM-5 framework;
some additional extra-framework Co3O4 existed only in small amounts. For the im-
pregnated catalyst the predominant species was cobalt silicate assemblies attached
to the ZSM-5 framework. These assemblies were difficult to reduce so that tem-
peratures above 720 °C were necessary to obtain metallic cobalt; the zeolites frame
work was, however, hardly affected by these high temperatures. The low carbon
monoxide conversions which were reported by the authors (about 4 %) can be ex-
plained by the low reducibility of these species.
In a more detailed study the ZSM-system was examined by S. BESSEL [110-112] with
paying attention to the channel size. Therefore different ZSM-zeolites, namely, ZSM-
5, ZSM-11, ZSM-12 and ZSM-34 were applied. From this work it was concluded that
the kind of ZSM applied influenced mainly the activity of the catalytic system (Tab.
2.3). Moreover, the activity could be linked to the channel size of the zeolite. The
highest CO conversion was obtained over Co supported on ZSM-12 which consists
of a 12 membered ring system, followed by the 10 membered rings of ZSM-5 and
ZSM-11, while on the 8 membered rings of ZSM-34 the lowest carbon monoxide
conversion was obtained. No significant differences occurred in methane selectivity
and α-value. This result was explained by the absence of any electronic cobalt-
support effect. The increase in CO-conversion was ascribed to a higher dispersion of
cobalt species, i.e., an increasing formation of smaller cobalt crystals with increasing
dimensions of the zeolite channels.
2 State of the Art 20
Tab. 2.3: CO-Conversion, methane selectivity and α-value for various ZSM-catalysts
(H2:CO = 2:1, 240 °C, 20 bar, GHSV = 1000 h-1) [110]
Catalyst XCO [%] S(CH4) [%] α [−]
ZSM-5 60 21 0.81
ZSM-11 61 20 0.82
ZSM-12 79 19 0.79
ZSM-34 45 18 0.82
Cobalt Supported on Nb 2O5
Niobia was applied as catalytic support by SCHMAL and co-workers [113]. TPR-ex-
periments resulted in the well-known, two-peaked spectrum structure as previously
reported for alumina and silica supported catalysts. The catalytic tests were carried
out at a preset reaction temperature of 220 °C, a H2/CO ratio of 1.4 and total pres-
sure between 0.1 and 3 MPa. With the increasing pressure a shift to higher hydro-
carbons was expect but an unusual deviation form the Schulz-Flory distribution was
observed (Tab. 2.4): the range C13-18 increased whereas C5-12 decreased. This result
was ascribed by the authors to the formation and propagation of chains on two types
of sites or chain growth can be attributed to the increasing readsorption probability of
products at higher pressure. Finally, the authors assumed that a strong cobalt-
support interaction took place at the high pressure experiments. The reducing feed
gas might penetrate into gaps between the Co-NbOx interface, thereby allowing the
formation of Cox-NbOy on the support surface. This modification should lead to an
altered distribution between Co2+ and Co3+ species and the appearance of the new
species, which increases sharply the selectivity within the C13-18 fraction.
Tab. 2.4: Overview over the obtained selectivities on Co/Nb2O5 catalysts at different
reaction pressures [113].
MPa Selectivity (wt%)*
C1-4 C5-12 C13-18 C19-26 C27+
0.1 4.11 8.69 18.24 40.70 4.25
3.0 5.24 4.14 41.03 36.80 1.50
*(220 °C, H2:CO=1.43)
2 State of the Art 21
2.6.2. EFFECT OF SUPPORT MATERIALS ON THE PERFORMANCE OF COBALT
BASED CATALYSTS
REUEL and BARTHOLOMEV studied the influence of support and Co dispersion on ac-
tivity and selectivity of supported cobalt systems. Co/Al2O3, Co/SiO2 and Co/TiO2
catalysts with varying metal content (3 and 10 wt%) were prepared from cobalt nitrate
applying the incipient wetness technique. [114,115]. In the course of this work they
established that the TOFCO is a function of support, dispersion, metal loading and
preparation. By catalytic tests at atmospheric pressure and 225°C it was found that
the activity decreases in the following order: Co/TiO2 > Co/SiO2 > Co/Al2O3 > Co/C >
Co/MgO (all 3 wt% Co). On all catalysts no complete reduction could be obtained;
the catalyst loaded with 3 wt% Co was reduced to 14% only, the 10 wt% catalyst to
47%. Within one set of catalysts the specific activity in CO hydrogenation decreased
with increasing dispersion. The product distribution was also influenced by the above
mentioned parameters. The formation of hydrocarbons can be correlated with dis-
persion and extent of reduction, i.e., low boiling hydrocarbons and a higher CO2/H2O
ratio was observed on catalysts having a higher dispersion and a lower extent of re-
duction. In a later study [116] the researchers concluded that the dispersion effect
can be minimised if only highly reduced catalysts (>90%) are examined. Additionally
it should be noted that the reported results were obtained on systems with different
metal loadings.
A survey of the catalytic data reported above is given in Tab. 2.5. One can conclude
that the applied support influences the reaction performance as well as the product
distribution. For cobalt catalysts (10 wt%) supported on Al2O3, C, MgO and TiO2,
which were evaluated under comparable reactions conditions, the reported TOF-val-
ues vary between 2.2 and 38⋅10-3 s-1. Furthermore, the extent of reduction and the
dispersion was also affected.
2.6.3. EFFECT OF PROMOTERS ON THE PERFORMANCE OF COBALT BASED
CATALYSTS
Lanthana Promoted Systems
The promoter influence of La2O3 on a cobalt catalyst (20 wt%) prepared from cobalt
nitrate applying the impregnation technique was examined within a La/Co atomic ra-
tio from 0.0 to 0.75 by HADDAD et al. [117-119]. At La/Co ratios below 0.5 the amount
of reducible Co increased from 30 to 50 % with increasing La content; no effect on
the reducibility of cobalt silicates was observed. Above a La/Co ratio of 0.5 the
Co2SiO4 species became easier to reduce. The presence of La appeared to moder-
ate the formation of strong Co-support interactions leading to better reducibility of the
Co oxide phase and to a large number of exposed Co0 atoms. From catalytic tests it
2 State of the Art 22
was derived that the presence of La did not influence the overall activity because the
estimated TOF* values were of the same order of magnitude independent of the cho-
sen La/Co ratio. On the other hand, the α-value increased from the undoped catalyst
(α = 0.57) to the doped one (α = 0.71, La/Co ratio of 0.75). The influence of La as a
promoter for alumina- supported FTS catalysts on their catalytic performance was
examined by VADA et al. [120].
The addition of La with a ratio of La/Co = 0.1 led to a decreasing value of XCO in
comparison to the unpromoted catalyst; on the other hand the formation of high-
boiling hydrocarbons was favoured as expressed by a higher α-value.
On the catalysts described above an increasing chain growth probability was noticed.
These findings can be explained by the presence of La. It is proposed that La3+ may
enhance olefin readsorption near the La3+-Co interface which is responsible for the
higher chain growth probability. ADACHI and FUJIMOTO supported these findings [121].
Ruthenium Promoted Systems
The promoter influence on carbon monoxide conversion and C5+ selectivity on a
Co/TiO2 catalyst was examined by IGLESIA [12]. In this case Ruthenium was added to
the previously described catalyst. The research group obtained a 3 times higher
turnover rate in comparison to the undoped catalyst; additionally, SC5+ increased from
84.5 to 91.1 %. These findings can be explained by a structural promotion effect of
Ru [13], i.e., the promoter acts stabilising on the catalysts phase composition and on
the previously applied cobalt oxide. This structural promotion resulted in a lower re-
duction temperature and in preventing the agglomeration of CoOx species on the
support material. Furthermore, an electronical promotion was observed as shown by
the higher TOF values obtained. It seems that the intimate contact of ruthenium and
cobalt increases the number of exposed metal atoms that are involved in the Fischer-
Tropsch reaction.
* TOF calculated based on hydrogen chemisorption; reaction conditions: Treac = 220 °C, p = 1 atm, H2/
CO = 2:1
2 State of the A
rt23
Tab. 2.5: O
verview of physico-chem
ical properties, reaction conditions and catalytic
results for various supportsreference
[148]
[114.115]
[114.115]
[114.115]
[136]
[114,115]
[114,115]
[92]
[92]
[104]
[111]
[111]
[111]
[92]
α
n.m.
n.m.
n.m.
n.m.
0.79
n.m.
n.m.
n.m.
n.m.
0.94
0.81
0.81
0.78
n.m.
S(C5+)
[wt%]
16.1
35.7
16
6.2
38.2
42.2
53.7
34
9
84.5
n.m.
n.m.
n.m.
14
S(C1)
[wt%]
52.3
32
53
55
8.3
29
16
13
49
7
22
21
26
42
TOF
[103s-1]
45
12
2.2
0.13
n.m.
7.5
38
n.m.
70.9
17.7
n.m.
n.m.
n.m.
6.07
XCO
[%]
9.7
5.6
4.7
7.4
44.1
7.5
5.6
5
n.m.
48.7
32.4
59.6
33.9
n.m.
GHSV
[h-1]
n.m.
500
150
200
268
500
900
1500
1800
n.m.
1000
1000
1000
1800
p
[bar]
1
1
1
1
5.4
1
1
n.m.
1
20
20
20
20
1
H2:CO
2
2
2
2
1
2
2
1.2
2
2.05
2
2
2
2
Treac
[°C]
260
200
200
300
268
200
200
210
250
200
240
240
240
250
Red.
[%]
89
34
47
13
n.m.
92
47
n.m.
34
n.m.
38
100
52
64
DCo
[%]
3.8
9.9
36
1.9
n.m.
10
4.5
n.m.
7.6
2.2
22
6
18
6.3
Precur-
sor
nitrate
nitrate
nitrate
nitrate
nitrate
nitrate
nitrate
Car-
bonate
nitrate
nitrate
nitrate
nitrate
nitrate
nitrate
Support
Nb2O5
Al2O3
C
MgO
MnO
SiO2
TiO2
CeO2
Al2O3
TiO2
Y
ZSM-5
Mordenite
Al2O3
Co
[wt%]
5
10
10
10
10
10
10
11
12
12
7.7
7.8
8.8
20
2 State of the Art 24
Furthermore, Ru facilitated the reduction of cobalt catalyst supported on silica as de-
rived from TPR experiments and an increase of the number of exposed cobalt metal
atoms on the catalyst surface was noticed. The authors concluded that Ru acts as a
structural promoter for Co by increasing the reducibility and dispersion of cobalt.
Similar results were obtained by the use of Pt as promoter by SCHANKE et al.
[122,123] and KAPOOR et al. [124,125] as well as for alkali promotion as shown by
BLEKKAN et al. [126].Starting from cobalt carbonyls JÄÄSKELÄNIEN added Ru as pro-
moter to the catalyst precursor [127,128]. On these catalysts higher site-time yields in
comparison to undoped, impregnated catalysts was achieved; on the other hand this
type of catalysts deactivates within a time of 10 h; a Co-Ru/SiO2 catalysts lost about
40% in activity.
On alumina supported cobalt catalysts the chain growth probability stayed nearly
constant by promoting with ruthenium as shown by KOGELBAUER and co-workers
[129], but an increase of carbon monoxide conversion was observed.
Cobalt Catalysts Promoted with Metals
ALI et al. introduced Zr as a promoter for cobalt- based catalysts supported on silica
[130] with a weight content of Zr in the range of 0.7 to 8.5 wt% applying the kneading
technique, i.e., the mechanical mixture of previously prepared Co/SiO2 catalyst with
ZrO2, and the incipient wetness procedure. The addition of zirconium led in both
cases to an increase in activity and chain growth probability. It appeared that the ef-
fect of Zr promotion depends on the method of preparation and on the Zr/Co ratio.
IGLESIA et al. studied Re as a promoter for a Co/TiO2 catalyst; the addition of 0.8 wt%
Re led to an increase of cobalt dispersion from 2.2 to 5.3 % without any effect on
FTS turnover rate, apparently by forming Re oxide species that anchor CoOx clusters
and avoid sintering during the catalyst oxidation procedure [103].
MgO was used as promoter by NIEMELÄ and KRAUSE [131]; they obtained an increase
in activity due to the added magnesia as well as a decrease in selectivity towards
methane and carbon dioxide. This shift in the product distribution was caused by a
chemical modification of the active cobalt sites; the way how magnesia interacts with
surface cobalt is still a point of interest. Table 2.5 gives an overview over the catalytic
data obtained; it is visible that no direct link between dispersion and carbon monox-
ide conversion exists.
2 State of the Art 25
Tab. 2.6: Cobalt dispersion, degree of reduction and carbon monoxide conversion
obtained on cobalt catalysts promoted with MgO (ptot = 0.5MPa, Treac = 235 °C,
catalyst) and cobalt acetate (IW-ACE catalyst) were applied. The influence of the ap-
plied solvent was examined at two catalysts; instead of water a ammonia solution
(IW-OXA-NH3 catalyst) as well as acetone (IW-NIT-AC catalyst) prepared from co-
balt nit rate) was used.
The spr eading technique is based on the mechanism of the chemical transport reac-
tion. Under the chosen conditions the spreading of the cobalt precursors was as-
sumed, as formerly reported for vanadia [152-156], which should lead to higher co-
balt dispersions; 4 different catalysts (SPR-OXA, SPR-NIT, SPR-CoTiO3 and SPR-
Co3O4) were prepared by this method.
The precipitation of cobalt as hydroxide by variation of the pH-value (PR-8 and PR-
12 catalysts) and EDTA complex (PR-EDTA catalyst) was used in order to test if a
prevention of the sorption of supports pore system with the cobalt precursor led to a
higher cobalt dispersion.
Beside the conventional preparation methods the usefulness of plasma-inducedpreparation techniques were tested in order to obtain an increase of cobalt dispersionand a controlled deposition of cobalt onto the chosen support. Therefore three dif-ferent techniques were applied, namely plasma induced decomposition, plasmasputtering and plasma activation [158]. For the plasma induced decomposition tech-nique catalysts prepared by incipient wetness technique were applied. The coatedcatalyst precursor was then decomposed by means of an oxygen-plasma. It was as-
3 Objectives and Methods 32
sumed that the moderate decomposition conditions (low temperature ≈ 60°C, pres-
sure ≈ 0.1 bar) prevent the migration of surface cobalt. Another possibility for the ap-
plication of the plasma technique was plasma sputtering; pure cobalt was used ascathode. An argon plasma sputtered cobalt atoms from the cathode that precipitatedon the support material sited above the anode. It was supposed that this procedurewould lead to thin cobalt layer on the support surface. Furthermore cobalt acetylacetonate was added to the support as a powder. This mixture was continuouslystirred by use of a vibrational membrane and the cobalt precursor was decomposedwithin an oxygen plasma-field, so that cobalt atoms adsorbed on the support.
3.2.2. CHARACTERISATION OF CATALYSTS
The catalyst bulk structure was examined by means of x-ray diffraction (XRD) to in-
vestigate the influence of different cobalt precursors on the formation of cobalt oxide
species, i.e. of CoO and Co3O4, respectively. Furthermore, the existence of cobalt ti-
tanate could be established. Additionally, the bulk composition was determined by
the use of ICP-OES (Inductively Coupled Plasma-Optical Emission Spectrometry).
The reduction behaviour was studied by the use of Temperature Programmed Re-
duction (TPR). Based on TPR experiments more detailed information can be gath-
ered on the structure of the surface cobalt. It is well known that small particles are
more difficult to reduce than the larger ones. Likewise, the extent of reduction and the
reduction temperature are key points in the assessment of the feasibility of the newly
prepared catalyst as it is known that a low reduction temperature can be assigned to
easily accessible surface cobalt. The adsorption behaviour of the reference catalyst
was examined by Temperature Programmed Desorption (TPD) of CO in order to in-
vestigate the surface processes. The oxidation behaviour was asses by Temperature
Programmed Oxygenation (TPO), also. Cobalt dispersion was determined by the use
of CO-pulse measurements under the assumption that bridge and double bonding of
carbon monoxide can be neglected. Then, the percentage of accessible cobalt can
be directly derived from the amount of adsorbed carbon monoxide.
The surface composition and the oxidation state of the supported elements of the
catalysts were examined by means of X-ray Photoelectron Spectroscopy (XPS).
The application of the Temporal Analysis of Products (TAP) reactor system allowed a
principal insight on the interaction of H2 and CO on the surfaces of the catalysts. In
addition to these experiments DRIFT and pseudo in-situ XPS were carried out to ex-
amine the nature of the adsorbed species.
3 Objectives and Methods 33
3.2.3. CATALYTIC EVALUATION
Fixed-bed-reactor: The reference catalysts Co-Ref and Co/Ru-Ref were tested under
fixed-conditions (H2:CO = 2, ptot = 20 bar, GHSV = 1200 h-1). The obtained carbon
monoxide conversion (XCO) and chain growth probability (α-value) obtained on the
reference catalysts was set as standard so that the derived data of the prepared
catalyst (as mentioned within section 3.2.1) samples were set in comparison. This
allows an easy assessment if an improved catalyst was prepared.
Slurry reactor: The catalyst IW-ACAC2 was transferred into a slurry reactor with the
aim to check the product distribution and activity by changing reaction variables and
conditions.
3.2.4. KINETIC EVALUATION
The rate of CO consumption and methane formation was obtained applying a Berty-
type gradientless recycle reactor. The advantage of this type of reactor is that the
rate constant can be directly estimated from the consumption of carbon monoxide
under the assumption that only differential amounts of the reactant is involved in the
catalytic reaction, i.e., low conversion levels only. The experiments were carried out
on IW-ACAC3 catalyst by varying reaction temperature, partial pressure of CO and
H2 in the feed-gas and residence time.
4 Experimental 34
4 Experimental
The applied characterisation techniques and experimental conditions are described
in detail. Afterwards the preparation recipes of the new catalytic materials are given.
Furthermore, specifications of the employed catalytic reactors, the analytical system
as well as the evaluation of the catalytic results are presented. Used chemicals, their
suppliers and purity are listed in the attachment.
4.1. CHARACTERISATION OF CATALYSTS
4.1.1. XRD- INVESTIGATION
The XRD-analysis was carried out by transmission powder diffractomery (Stoe STA-
DIP). A CuKα1 radiation was used and the 2 θ range from 10 to 70° was examined.
The crystalline phases were identified by use of the JCPS data bank.
4.1.2. TPR- EXPERIMENTS
TPR (Temperature Programmed Reduction, Oxidation and Desorption) experiments
were performed in a DSC-apparatus (Setaram). The gaseous compounds were ana-lysed by a quadrupole mass spectrometer.
The catalyst was pretreated in a flow of oxygen before TPR-examination (1 h at400 °C; flow rate: 30 ml/min). After the pretreatment, the catalyst bed was cooleddown to room temperature and heated up linearly to 400 °C with a ramp rate of 10K/min; afterwards the temperature was kept constant at 400°C for 30 min. A flow ofdiluted hydrogen (5 % H2 in He) was passed over the catalyst bed during TPR analy-sis.
Another set of TPR experiments was carried out using a power-input controlled oven(Eurotherm 4472) which allowed linear heating up to 400 °C; the effluent concen-tration and hence, the hydrogen consumption was measured by means of a thermalconductivity detector (TCD, educt gas: 5 % H2 in Ar; flow rate: 45 ml/min).
4.1.3. TPO- AND TPD- EXPERIMENTS
TPO and TPD experiments (Temperature Programmed Oxidation and Desorption)
were performed in a DSC-apparatus (Setaram). The gaseous compounds were ana-lysed by a quadrupole mass spectrometer.
The sample was reduced in a flow of hydrogen (400 °C, 2 h, 30 ml/min) before ex-perimentation (for TPO as well as TPD). For TPO-experiments a flow of diluted oxy-gen (5 % O2 in He) was used. For TPD-examinations the catalyst surface was firstcovered with carbon monoxide (5 % CO in He). Subsequently, a flow of helium was
4 Experimental 35
passed through the reactor. The flow rate for all gas mixtures mentioned above was30 ml/min.
After pretreatment the catalyst bed was cooled down to room temperature andheated up linearly to 600 °C with a ramp rate of 20 K/min.
4.1.4. CO-PULSE EXPERIMENTS
Cobalt dispersion was determined from the amount of CO that remained on the sur-face during subsequent pulsing of CO over the reduced catalyst. A defined amount of
catalyst (pressed, crushed and sieved to a fraction of 255-350 µm) was filled into a
quartz tube and incorporated in a temperature controlled oven. The reactor outletwas connected to a Thermal Conductivity Detector (TCD). The catalyst was reduced
in a flow of hydrogen by 400 °C for 2 h (flow rate H2: 45 ml/min). Afterwards the sam-ple was purged with He at 400 °C for one hour and finally cooled down to room tem-perature. Carbon monoxide was pulsed at 23 °C over the reduced catalyst until theintensities of the CO pulses as measured by TCD remained constant.
For determining Co dispersion from CO pulse experiments it was assumed that:
· one carbon monoxide molecule adsorbs on one cobalt site (i.e., no bridgedbonding was accounted for)
· the formation of any cobalt carbonyls can be neglected
The cobalt dispersion was calculated based on the total amount of cobalt on thecatalyst as derived by ICP measurements.
100% Co Mol
CO MolD
avaibletotally
catalyst on adsorbedCo ⋅= (4.1)
Additionally the cobalt dispersion was determined based on the amount of reducedcobalt as derived from TPR-examinations:
100% Co Mol
CO Mol redD
)0(Co state reducded the in avaible
catalyst on adsorbedCo ⋅= (4.2)
4.1.5. XPS MEASUREMENTS
For determining the surface concentration of cobalt and titania, the previously oxi-
dised catalysts were studied by XPS (Fisons ESCALAB 220i-XL). The measurements
were performed applying an AlKα (1486.3 eV) X-ray source. The pressure in the
analysis chamber was below 10-8 mbar. Before taking the XP spectra in the analysis
chamber, it was possible to pretreat the sample under different atmospheres inside a
reaction chamber at elevated temperature. A transfer of the sample from the reaction
chamber to the analysis chamber without contact to air was also possible.
4 Experimental 36
The assignment of the measured binding energies to different species and oxidationstate was made by use of published reference data [159]. The correction of peak lo-cation was made applying the Ti 2p3/2 signal at 458.8 eV for Ti. The experimentswere carried out under charge compensation (Flood Gun).
An examination of the Ru-Region was not possible, due to the low amount of Ru pre-sent in the samples (0.1 wt%) as well as to the always present background noise ofcarbon which overlaps the Ru-region.
4.1.6. PSEUDO IN-SITU XPS MEASUREMENTS
Pseudo in-situ XPS measurement is defined as a examination of a sample within areaction chamber attached to the XPS apparatus. The sample can be transferred tothe analysis chamber without any contact to air and can be examined directly afterthe reaction was stopped.
First the fresh, unreduced reference catalyst was examined. Afterwards the catalystwas oxidised in a flow of oxygen (30 ml/min) at a temperature of 400 °C for 3h andthen reduced in a flow of hydrogen at 400 °C for 3h in the reaction chamber of thespectrometer. The chamber was evacuated, the sample was transferred into theanalysis chamber without contact to ambient air for XPS-spectra recording. After-wards a flow of H2:CO with a ratio of 10:1 was passed over the catalyst outside theanalysis chamber. An excess of H2 was used to avoid the formation of higher hydro-carbons that would soil the reaction chamber. Fischer-Tropsch reaction took place for12 h at a reaction temperature of 200 °C at a pressure of 18 bar. After that time onstream the reaction chamber was evacuated again and the catalyst was character-ised.
4.1.7. ICP- OES
The catalyst was dissolved in a mixture of HNO3 and HF (ratio: 1:5) and treated sub-sequently in a microwave oven (8 bar operating pressure) in order to ensure a com-plete dissolution of the sample. Subsequently the sample was analysed in the ICP-apparatus (Perkin-Elmer: Optima 3000 XL).
4.1.8. TEM- MEASUREMENTS
The TEM examinations were carried out in a Hitachi H-8100 electron microscope; theenergy of the electron beam was 100 keV. A resolution of 0.5 nm and a magnificationof 3105 was used. For the element analysis energy dispersive x-ray analysis (EDX)was applied.
4 Experimental 37
4.1.9. DRIFT- MEASUREMENTS
The cobalt catalyst was reduced in a flow of hydrogen for 3 h at a temperature of400 °C. There, a background spectrum of the fresh catalyst was taken as reference.A gas mixture (5% CO, 10% H2, 85% N2) was passed through the catalyst bed. Thereaction temperature was varied in a range between 180 °C to 220 °C. The totalpressure amounted to 1 bar. The time difference between each spectrum amountedto 1 min. The reference spectrum was subtracted from every DRIFT-spectrum in or-der to obtain a difference spectra.
4.1.10. TAP- REACTOR- SYSTEM
The pulse experiments applying the TAP-reactor system (Temporal Analysis ofProducts) were carried out under vacuum conditions (110-8 mbar) and the pulse size
was approximately 21014 molecules/pulse.
All catalysts were reduced in-situ in a flow of hydrogen at 400 °C. Afterwards, the re-actor was evacuated; the pulse experiments were carried out at a reaction tem-perature of 200 °C and mixtures of H2/He (1:1) and CO/Ne (1:1) were used. Thesample was reduced in a flow of hydrogen at 400°C for 3 h under atmospheric pres-sure. CO/Ne or H2/Ne (21014 molecules per pulse) were pulsed over the catalyst bedunder variation of temperature (180-220°C) and products were detected by a quad-rupole mass spectrometer.
Heat of Adsorption
The adsorption enthalpy was determined by means of the TAP reactor system. As-
suming that CO diffusion was in the range of Knudsen-Diffusion, the heat of adsorp-
tion for CO on the doped catalyst can be derived by equitation (4.3) as described by
GLEAVES et al. [160].
0,d
0,aadad
i
iCO
k
kln
RT
H
RT
EE
t
ttln +
∆−=
−=
− (4.3)
The term tCo describes the mean residence time of CO in the reactor system. ti cor-responds to the mean residence time of an inert gas molecule. By plotting ln(tco-ti)/ti)vs. 1/T (the so-called ARRHENIUS-plot), the heat of adsorption of CO on Co/Ru-Refcatalyst can be determined.
4.2. PREPARATION OF CATALYSTS
The preparation of the catalysts is described. The goal was to obtain supportedcatalysts containing 12 wt% of cobalt and in the case of the doped catalysts 0.1 wt%of Ru (or alkali, respectively).
4 Experimental 38
4.2.1. SUPPORT PRETREATMENT
Titania (Degussa P25) was calcined at 560°C for 16 h under ambient conditions be-fore exposure to the cobalt precursor. This pretreatment of titania leads to an enrich-ment of the rutile fraction (> 70 % rutile and < 30 % anatase), which is the more ac-tive modification of titania as had been shown by IGLESIA’S work [161].
The other supports (ceria, zirconia, and Bayer-titania) were calcined under ambient
conditions at 560°C for 12 h.
4.2.2. CATALYST PRECURSOR TREATMENT
After coating of the support with a cobalt precursor applying incipient wetness-,spreading- and precipitation techniques all catalyst precursors were decomposedand transformed into cobalt oxide in a flow of oxygen at 400°C for 4 h (heating line-arly from room temperature to 400°C; ramp rate 6 K/min; flow rate: 30 ml/min). Thedecomposition of the nitrogen containing precursors led to NO2 and H2O and of thehydrocarbon containing precursors to CO2 and H2O respectively.
4.2.3. OVERVIEW OF AL PREPARED CATALYST
In the following table 4.1 all prepared catalyst are listed sorted by the applied prepa-ration technique along with the used support material and if indicated a change of thestandard preparation technique.
4.2.4. INCIPIENT WETNESS TECHNIQUE
Preparation of Catalyst: Co-Ref, IW-ACE and IW-OXA
Co nitrate (0.33 mmol/ml, Co-Ref), cobalt acetate (0.01 mmol/ml, IW-ACE), or cobaltoxalate (0.01 mmol/ml), respectively, was dissolved in deionized water. The solutionof cobalt precursor was added to the support until visible wetness was obtained; thematerial was subsequently dried at 130 °C for 4 h. This procedure was repeated untilapprox. 12 wt% of cobalt was deposited on the support.
Variation of Cobalt Precursor: IW-ACAC2 and IW-ACAC3
Co(II)-acetyl acetonate (0.4 mmol/ml, IW-ACAC2) and Co(III)-acetyl acetonate (0.25
mmol/ml, IW-ACAC3) was dissolved in CH2Cl2. After each impregnation step the
catalyst precursors were dried in a rotary evaporator (70 mbar, 70 °C bath tempera-
ture). The impregnation steps for IW-ACAC2 and IW-ACAC3 were repeated two and
three times, respectively.
4 Experimental 39
Tab. 4.1: Overview of all prepared catalysts, their names applied cobalt precursorsand support materials
Catalyst Technique Precursor
Co
Support Remarks
Co-Ref impregnation nitrate TiO2
Co/Ru-Ref impregnation nitrate TiO2 promoter: Ru
IW-ACAC2 impregnation (II) acetyl acetonate TiO2 solvent: CH2Cl2
Doping of IW-ACAC3 and Co-Ref with Ruthenium: IW-ACAC3-Ru and Co/Ru-Ref
The above described IW-ACAC3 as well as Co-Ref catalyst was doped with ruthe-
nium. Ru(III) chloride (1.38⋅10-5 mol/gprecursor) was dissolved in CH2Cl2 and added to
the uncalciend catalyst precursor used. The impregnated catalyst was dried at 70 °Cfor 2 h in a rotary evaporator as well.
Variation of Solvent: IW-NIT-AC and IW-OXA-NH3
The procedure described above was modified by change of solvent for the cobaltprecursor solution. Acetone (IW-NIT) or ammonia (IW-OXA) was used instead ofwater, otherwise the procedure was carried out in the same manner. Acetone wasevaporated at a pressure of 70 mbar at 50 °C (bath temperature) in a rotary evapo-rator. IW-OXA-NH3 was dried in an oven at 130 °C for 2 h.
Variation of Decomposition Procedure: IW-NIT-Step
Another impregnated cobalt catalyst was prepared based on cobalt nitrate. Thepreparation procedure was varied in the way that the applied cobalt nitrate was de-composed in a flow of pure oxygen after each impregnation step instead of drying,only. This procedure was repeated five times and the loading was increased from2.4 wt% (1. step) over 7.4 wt% (3. step) until the desired amount of approx. 12 wt%(5. Step) was achieved.
Variation of Support Material: IWC-NIT, IWZ-NIT and IWB-NIT
CeO2, ZrO2 and Bayer-titania (pure rutile type) were chosen as supports instead oftitania [162,163,164,165]. The supports were impregnated with an aqueous solutionof Co(NO3)2 (0.33 mmol/ml) until visible wetness and dried at 110 °C. This procedurewas repeated until the desired Co loading of 12 wt% was reached.
4.2.5. SPREADING OF COBALT PRECURSORS
Spreading of Cobalt Oxalate: SPR-OXA
Catalyst SPR-OXA was prepared from cobalt oxalate (3.12 g) and titania (10.00 g) bymechanical mixing. The mixture was heated from room temperature to 250°C with aramp-rate of 10 K/min under an inert atmosphere (N2). After 3 h at 250 °C the tem-perature was raised to 400°C; the decomposition of cobalt oxalate begins at 343 °C[166].
Doping of SPR-OXA with Ruthenium: SPR-OXA-Ru
The above described SPR-OXA catalyst was doped with Ru after spreading of the
precursor. Ru(III) chloride (1.38⋅10-5 mol/gprecursor) was dissolved in CH2Cl2, added to
the catalyst sample and dried at 80 °C for 2 h in an oven.
4 Experimental 41
Spreading of Cobalt Oxide and Cobalt Titanate: SPR-Co 3O4 and SPR-CoTiO 3
The catalysts SPR-Co3O4 and SPR-CoTiO3 were prepared from cobalt oxide and co-balt titanate as cobalt precursors. The calcined support material titania (10.00 g foreach catalyst) was mechanically mixed with cobalt oxide (5.12 g) and cobalt titanate(3.29 g), respectively. The mixture was heated from room temperature to 400 °C witha ramp-rate of 10 K/min under an inert atmosphere (N2), the final temperature washeld for 12 h.
4.2.6. PRECIPITATION
Precipitation of Co(OH) 2 : PR-8 and PR-12
For preparing PR-8 catalyst, Co(NO3)2 was dissolved (0.33 mmol/ml) under stirring indeionized water containing titania (1.1 mmol/ml) at 60 °C; then a saturated NaHCO3
(1 mmol/ml) solution was dropped into the slurry. The pH-value was adjusted withaqueous HNO3 until pH=8 was achieved. After two hours of stirring the solid catalystprecursor, containing cobalt hydroxide precipitated on titania was separated by filtra-tion from the solution; washed five times with water (each time: 3ml (H2O)/g catalyst)and then dried in air at 130 °C for 12 h.
In order to examine the influence of the pH-value on the cobalt deposition a catalystwas prepared by precipitation of Co(OH)2 on the support material at a pH-value of 12(PR-12).
Doping of Precipitated Co(OH) 2: PR-12-Na
The catalyst PR-12 was doped with 0.1 wt% Na as a citrate, since alkaline com-pounds are known to suppress cobalt migration during calcination [167].
Precipitation of Cobalt-EDTA-complex: PR-EDTA
To prepare Co-EDTA catalyst, Co(NO3)2 (0.33 mmol/ml) was dissolved in distilledwater to which titania was added and heated to 60 °C. After 1 h of vigorous stirringthe necessary amount of 0.03 mol EDTA was added and the pH-value was adjustedto 9 with an aqueous 25 vol% -NH3-solution. The water was evaporated and the re-sulting solid material was subsequently dried under ambient pressure at 130 °C overnight.
Doping of PR-EDTA with Ruthenium: PR-EDTA-Ru
After drying of the precipitated complex (see above) ruthenium chloride, dissolved inCH2Cl2, was added to the catalyst which was subsequently dried in a rotary evapo-rator (40 hPa, bath temperature: 70 °C) consequently.
4 Experimental 42
4.2.7. PLASMA INDUCED PREPARATION
Plasma Decomposition: PD-NIT, PD-ACE, PD-ACAC2
Three different kinds of cobalt precursors, cobalt acetyl acetonate, cobalt acetate and
cobalt oxalate were used for preparing the catalyst precursors applying the incipient
wetness technique. After coating of the support (TiO2) and subsequent drying at
110 °C, the impregnated support was treated in an oxygen-plasma. The plasma was
generated by a micro-wave source with a power of 120 W. The decomposition took
place at approx. 60 °C. A scheme of the apparatus is given in Fig. 4.1.
Plasma Sputtering: PS-Co
Titania (pressed and sieved into a fraction of 250-355 µm) was placed into the
plasma apparatus. Solid cobalt metal was used as cathode. Small amounts of cobaltwere scraped off by the atomic, ionic and radical oxygen species within the plasmafield. These cobalt species then precipitated on the support (see Fig. 4.1).
impregnated catalyst
anode
kathode
fitting
glas cylinder
rotor
pla
sma
filed
Fig. 4.1: Scheme of the apparatus applied for plasma decomposition
Vibrational Plasma Sputtering: PL-100W and PL-150W
Two catalysts were prepared by the means of a novel plasma preparation technique.
The support was sieved into a fine fraction of < 125 µm; in this case titania was used
but every other support material may also be coated by the described method.
4 Experimental 43
This fine powder was placed on a membrane stimulated by a frequency generator, sothat the powder was mixed and turned around by the vibrations (seeFig. 4.2). Themembrane itself was fixed within the plasma field that allows an even coating of thesupport material. As cathode a cobalt cylinder was used. The cobalt was sputteredby an argon-plasma under a pressure of 65 mbar. Two catalysts were prepared ap-plying a plasma power of 100W [PL-100W] and 150W [PL-150W], respectively.
Plasma Activation: PL-AT and PL-AP
For plasma activation, the cobalt precursor (cobalt (III) acetyl acetonate) was mixed
with the sieved support instead of applying impregnated precursor as described for
the plasma decomposition technique. The cobalt precursor was then decomposed in
an oxygen-plasma under the conditions described above. For the catalyst PL-AT a
tablet of cobalt precursor was used, for the catalyst PL-AP cobalt acetyl acetonate
was added as a powder (see Fig. 4.2).
anode
kathode
pla
sma
filed
membrane
catalyst precursor
Fig. 4.2: Plasma apparatus for vibrational and activation technique.
4.3. CATALYTIC TESTING
The feed gas was supplied by three PC-controlled mass-flow-controllers (MFC 1 toMFC 3), which provided carbon monoxide, hydrogen and nitrogen up to 50 bar totalpressure. A flow sheet of the testing equipment is shown in (see Fig. 4.3). Behind themass-flow controller’s three valves (V1 to V3) were mounted which were attached tothe personal computer and shut down in case of a malfunction of the equipment. The
4 Experimental 44
gaseous streams were passed through high-grade steel tubes (type: V4A) with anouter diameter of 6 mm. The feed gas passed a safety valve (SV1); it was purifiedand mixed in an Oxisorb- and a Hydrosorb-cartridge (PF1 and PF2). The Oxisorb-cartridge removed eventually existing oxygen and the feed gas was dried by the Hy-drosorb-cartridge. Behind another safety valve (SV2), a 3-way-valve (3V4) was in-stalled to lead the^23456 789ijo gas mixture through the catalytic reactor or over abypass-tube.
The fixed-bed reactor could be replaced by either a slurry or a Berty-type reactor.The temperature was monitored by axial movable thermocouples (type: PT-100) overthe whole reactor length (TC2 and TC3). A maximum axial temperature gradient of3°C was measured under steady-state conditions; thus, the reactor was nearly iso-thermal. The reactor was heated applying a heating tape; its electrical power inputwas controlled by a personal computer and the temperature was monitored via ther-mocouple TC4.
After the reactor a 3-way valve (3V5) led the stream either to the condenser C1 or toC2. The condenser C1 had a volume of 1.5l, allowing catalytic tests overnight orweekend. The pressure of the gaseous mixture was released by a back pressureregulator PC1 and led to a Siemens gas chromatograph (see below) under atmos-pheric pressure for product analysis; the pre-pressure of the analyse-gas was moni-tored by PI3 in order to guarantee a constant flow through the GC. The pre-pressurewas adjusted by two valves V7 and V8. The analysed gas was then passed to theexhaust.
The pressure within the testing-rig was registered by two pressure gauges mountedbefore the reactor (PI1) and after the condensation line (PI2). This construction al-lowed a rapid detection of a blockage within the reactor, condensers or tubes.
All the tubes between the safety valve SV2 and the pressure regulator PC1 were
heated by means of heating tapes; the temperature of the lines was kept constant at
200 °C during the catalytic tests.
4 Experimental 45
Fig. 4.3: Flow-diagram of the Fischer-Tropsch testing-rig
4 Experimental 46
4.3.1. FIXED-BED REACTOR
The tubular fixed-bed was made of high-grade steel (V4A) with a length of 8 cm and
an inner diameter of 1.5 cm. The reactor was heated with a heating tape; the electri-
cal power input was controlled by a personal computer. 8 g of catalyst, diluted by in-
ert material in a ratio of catalyst to inert material = 4:1, was placed into the fixed-bed
reactor. Cooling of the reactor was not necessary because of its large surface-to-vol-
ume ratio, the small amount of catalyst and the relatively low catalytic activity.
The catalysts were tested at a reaction pressure of 20 bar and a space velocity of
1200 h-1. The feed gas consisted of hydrogen and carbon monoxide at a ratio of 2:1,
nitrogen was added as internal standard (p(H2):p(CO):p(N2) = 12:6:2). The catalysts
were reduced in a flow of hydrogen for 4 h at 400 °C before reaction start-up.
Fig. 4.4: Scheme of the fixed-bed reactor.
4.3.2. SLURRY REACTOR
The slurry-reactor (Autoclave Engineers, 0.3 l) was applied as an alternative to the
fixed-bed reactor in the testing rig [168,169]. The impeller was driven by a magnetic
drive. The reactor was heated by a heating tape with temperature control by a per-
sonal computer; the temperature inside the slurry was monitored by a thermocouple.
As liquid phase tetracosan was used and the reaction was carried out in a semi-con-
tinuous (gas continuous, liquid batch) manner. To the molten wax, 10 g of previously
reduced IW-ACAC3 catalyst (particle size < 50 µm) was added. The reaction pres-
sure was adjusted to 20 bar and a H2 to CO ratio of 2:1 by a GHSV of 1200 h-1 was
used.
4 Experimental 47
4.3.3. BERTY REACTOR
A Berty Reactor with internal recycle (Autoclave Engineers) was used for kineticmeasurements[170]. A scheme of the reactor is given in Fig.4.5. The reactor waselectronically heated. One thermocouple was placed below the catalyst bed and an-other one was above the catalyst in the upper part of the reactor. The temperaturegradient between both thermocouples was 1 K during all experiments. The reactiontemperature was adjusted to 202 and 218 °C, respectively.
The internal impeller was driven by a magnetic drive with a speed of 1500 rpm in or-der to guarantee a gradientless reaction.
As catalyst IW-ACAC3 (10g, fraction 500 – 1000 µm) was used. Before the catalytic
tests the sample was reduced in-situ at 400 °C with pure hydrogen (100 ml/min). Theratio of H2:CO was varied between 1.3 and 5.4; furthermore the GHSV was variedbetween 1200 and 2000 h-1.
Fig. 4.5: Scheme of the Berty- type reactor [171].
4.3.4. ANALYSIS OF PRODUCTS
A Siemens gaschromatograph (SiCHROMAT 2) was applied for FT product analysisequipped with a Thermal Conductivity Detector (TCD) and two Flame Ionisation De-tector’s (FID). The products obtained during FTS were analysed on-line as well asoff-line. By on-line analysis carbon monoxide, hydrogen, nitrogen, carbon dioxide andC1 to C6 hydrocarbons were separated by means of two capillary columns; a GS-Q(later on Al2O3-column) and a Molsieve 5a with a length of 30 m and a inner diameter
4 Experimental 48
of 0.53 mm was used. The permanent gases were detected by an TCD, for the hy-drocarbons a FID was applied.
Condensed high boiling hydrocarbons were dissolved in CS2 and analysed off-line.The solution was mixed with iso-propanol as internal standard and directly injected tothe column (DB-1 capillary column; length: 60 m; i.d. 0.32 mm; later on a CP-PONA:length 50 m; i.d. 0.23 mm), which allows to separate the hydrocarbons up to C50. Thetemperature programs applied during on-line and off-line analysis are given in Tab.4.2.
Tab. 4.2: Temperature program applied for on-line and off-line analysis
on-line Analysis off-line Analysis
Temp.
Start [°C]
Temp.
End [°C]
Ramp
[K/min]
Duration
[min]
Temp.
Start [°C]
Temp.
End [°C]
Ramp
[K/min]
Duration
[min]
50 50 0 8 60 270 10 21
50 180 20 6.5 270 270 0 159
180 180 0 32.5
4.4. DETERMINATION OF XCO, S(CN), α, TOF AND TONNOM
The carbon monoxide conversion, XCO, can be expressed as follows:
100% n
nnX
inletCO
outletCO
inletCO
CO •−= •
••(4.2)
where: inletCOn•
= molar flow entering the reactor
outletCOn•
= molar flow leaving the reactor
The yield (YCi) is defined as the amount of the desired hydrocarbon fraction (Ci)
formed during the reaction, taking the stoichiometry factors (ν) into consideration, i.e.
in the case of Ci = propane the value for ν(C3) amounted = 1 and for ν(CCO) = 3.
100% n
nY
Ci
CO
CO
CiCi •
νν
⋅=Α
•
•(4.3)
The selectivity towards Ci that corresponds to the fractional amount of Ci formed from
4 Experimental 49
CO converted can be calculated according to:
)CO(XY
S CiCi = (4.4)
Later the selectivity is given in wt%. These values can be derived by dividing SCi withthe molecular weight of Ci related to the sum of all products formed.
The chain growth probability was derived from the slope of the straight line whichwas obtained by plotting the weight concentration of Ci divided by the carbon number
n against n (see section 2.4). The described α – value is based on the hydrocarbonswith a carbon number >10.
In order to allow an easier comparison of the obtained data's the turn-over-frequencyas well as a normalised TOF is given later in the discussion of the catalytic results.The TOFnom is the quotient of the derived TOF for the tested catalyst (TOFcat) and theTOF obtained for the reference catalyst Co-Ref. This number allows an easy com-parison if the newly prepared catalyst got a better performance than the referencecatalyst; all numbers greater then 1 indicate a higher catalytic performance than onCo-Ref. The equations used for the calculation of the TOF values are given in (4.5)and (4.6).
][s n 100
X(CO) n TOF 1-
Co
CO
⋅⋅=
•(4.5)
with nCo = Co atoms as derived from CO-pulse experiments
Ref-Co
catnom TOF
TOF TOF = (4.6)
4.5. KINETIC EXPERIMENTS
Experiments were performed in a gradientless reactor (Berty-type). The partial pres-sure of nitrogen was used as internal standard to guarantee an accurate calculationof the molar flow of carbon monoxide; for detailed descriptions please refer to theappendix.
The advantage of the gradientless recycle reactor is that the rate of carbon monoxideconsumption can be derived directly out of the quotient of mole stream of carbon
monoxide at the reactor inlet ( Α
•
COn ) and outlet ( Ω
•
COn ). The CO consumption is de-
fined as:
4 Experimental 50
−=
Ω•
Α•
CO
CO
CO
n
n1X (4.7)
The rate of CO consumption related to the mass of catalyst was calculated as fol-lows:
mXnR
cat
COCOCO
⋅=−$
(4.8)
The rate of formation of compound k was determined under the assumption that theconcentration of [nk] is zero with the start of the reaction and no consecutive reac-tions with nk took place by:
mnR k
k
=
The number of moles of H2O could be estimated from the oxygen balance.
nnn H2OCOCO
$ += (4.9)
As CO2 formation was not observed in all experiments, carbon dioxide was not ac-counted for in C-and O- balance.
Furthermore, it was assumed that oxygenated hydrocarbons were not formed during
the reaction. The higher hydrocarbons were not taken into account in calculation of
partial pressures, since the number of moles of these products were low especially at
low conversion.
5 Results and Discussion 51
5 Results and Discussion
The characterisation results obtained for all catalysts are described. First, for com-
parison the data of characterisation studies on the reference catalysts (Co-Ref and
Co/Ru-Ref) will be shown. Thereafter the physico-chemical findings obtained on the
prepared samples will be presented arranged by the applied preparation technique.
The characterisation results of all catalysts will be discussed afterwards. Next, the
catalytic results arranged in the above described order are given and discussed. The
chapter closes with the presentation and discussion of the results obtained during
slurry tests and kinetic measurements.
5.1. CHARACTERISATION OF CATALYSTS
5.1.1. CHARACTERISATION OF REFERENCE CATALYSTS
XRD
No difference could be observed in the XRD pattern of the doped and undoped cata-
lyst. Three phases were detected: Co3O4 [172] and titanium dioxide in its two modifi-
cations anatase [173] and rutile [174]. The ratio of the latter two phases amounted to
3.4 and was not influenced by added ruthenium. Other possible phases like TiOX,
Co3O2, CoO, CoTiO2 and Co2TiO4 were not detected (see Fig. 5.1 ).
10 20 30 40 50 60 70
(o)(o)
Intensity / a.u.
(o)
(o)
2 θ
Fig. 5.1: XRD pattern of oxidised cobalt reference catalysts Co/Ru-Ref obtainedwithin a 2θ range from 10° to 70°. (o) =Co3O4
5 Results and Discussion 52
TPR
The reduction of the doped (Co/Ru-Ref) and undoped (Co-Ref) catalyst was investi-
gated by means of TPR within a temperature range from 25 to 500°C (reducing
agent: H2 (5%) in He). The TPR plots showed in both cases two maximum (see Fig.
5.2). The two peaks are located at 328 °C and 395 °C for the undoped and at 200 °C
and 340 °C for the doped catalyst, respectively.
TPO
Only one maximum appeared during TPO on both reduced samples (see Fig. 5.3)
which can be assigned to the re-oxidation of Co0 to cobalt oxide (Co3O4). The pres-
ence of Ru also influenced the oxidation temperature that decreased from 303 °C for
the Co-Ref catalyst to 280 °C for the doped catalyst (Co/Ru-Ref).
TPD
In the TPD-profile (see Fig. 5.4) of the doped catalyst two maximum at 130 °C and
227 °C and a broad shoulder at 340 °C were detected which correspond to three dif-
ferent adsorption states of carbon monoxide. For the catalyst Co-Ref only one de-
sorption maximum was detected at a temperature of 107 °C. At a temperature of
240°C the desorption of CO2 was observed on both catalysts.
100 200 300 400 500
395°C
II328°C
I
340°C
II200°C
I
Co/Ru-Ref
Co-Ref
H2-Consumption / a.u.
T / °C
Fig. 5.2: TPR-profiles of doped (Co/Ru-Ref) and undoped (Co-Ref) cobalt referencecatalyst (ramp rate: 10K/min, flow rate: 30 ml/min, reducing agent: H2 (5%) in He )
5 Results and Discussion 53
100 200 300 400 500
H2-Consumption / a.u.
Co/Ru-Ref
Co-Ref
303 °C
280 °C
T / °C
Fig. 5.3: TPO-profiles of doped (Co/Ru-Ref) and undoped (Co-Ref) reference catalyst(ramp rate: 20 K/min, flow rate: 30 ml/min, oxidising agent: O2 (5%) in He)
100 200 300 400 500
241°C
237°C
Co/Ru-Ref
Co-Ref
Co/Ru-Ref
Co-Ref
T / °C
100 200 300 400 500
340°C
227°C130°C
107°C
CO - desorption / a.u.CO 2-desortion / a.u.
T / °C
a) b)
Fig. 5.4: TPD-profiles of a) CO2 and b) CO desorption for reference catalysts Co-Refand Co/Ru-Ref (ramp rate: 20 K/min, flow rate: 30 ml/min, surface covered with CObefore experimentation)
5 Results and Discussion 54
Pseudo in-situ XPS Examination
Pseudo in-situ XPS measurements were made in order to follow the changes in the
surface composition after oxidation in O2, reduction in H2 and reaction with a syngas
feed (i.e. the sample can be transferred between the reaction and analysis chamber
without any contact to air; that allows a so-called pseudo in-situ XPS examination di-
rectly after oxidation, reduction and reaction, respectively).
In Fig. 5.5 the XP-spectra of the carbon region are plotted (binding energies between
275 to 300 eV). The fresh catalyst, i.e., an unused sample right after the de-
composition of the catalysts precursor, showed a carbon peak at 285.0 eV and a not
identified peak at 292.3 eV which might be a surface carbon species which combine
with oxygen from the catalysts structure. On the reduced catalyst no distinct peaks
are visible with the exception of a very broad shoulder, which indicates that most of
the carbon was removed from the catalysts surface due to preceding oxygen treat-
ment. The catalyst used within the XPS- apparatus shows two peaks in the carbon
region at 283.6 eV and 279.7 eV (see Fig. 5.5), the latter may be assigned to a car-
bide carbon species.
In the spectra of the fresh catalyst the presence of both Co0 and Co3O4 was ob-
served as shown in Fig. 5.6 [159]. The amount of Co3O4 decreased after reduction as
indicated by a significant decrease in the intensity of the shake-up satellite at
786.7 eV. The examination of the pseudo in-situ reduced catalyst (Flow of hydrogen
under atmospheric pressure; 400 °C, 3 h) pointed out, that not all available cobalt
was reduced to its metallic state because still Co3O4 was present. After the Fischer-
Tropsch reaction took place only Co0 was detected. The Co3O4 was totally reduced.
Within the Ti- region (450 to 470 eV) no change during oxidation, reduction and FT
reaction was recognised; that indicates that the support material was not affected
during the FTS.
Dispersion of Co
The cobalt dispersion amounted to 1.70 % for Co-Ref and to 1.65 % for the doped
catalyst. Therefore, doping the catalyst with ruthenium had no significant effect on
the cobalt dispersion. IGLESIA et al. determined a Co dispersion of 2.3 % [12].
5 Results and Discussion 55
300 295 290 285 280 275
292.3 eV
285.0 eV
283.6 eV 279.5 eVreference catalyst afterFT reaction
reference catalyst afteroxidation and reduction
fresh catalyst
Intensity / a.u.
binding energy (eV)
Fig. 5.5: XP-spectra measured in the carbon region on the fresh, reduced and used
(FTS within the reaction chamber) reference catalyst Co-Ref.
810 800 790 780 770
802.3 eV
795.0 eV
786.7 eV
779.8 eV
781.0 eV796.1 eV
777.6 eV792.5 eVreference catalyst after
FT reaction
reference catalyst afteroxidation and reduction
fresh catalyst
Intensity / a.u.
binding energy (eV)
Fig. 5.6: XP-spectra measured in the Co region on the fresh, reduced and used (FTS
within the reaction chamber) reference catalyst Co-Ref.
5 Results and Discussion 56
TEM-Results
The TEM patterns show that different, distinguishable regions on the sample exist.
Small spots are located as separate spheres on the support material. These spheres
consisted mainly of cobalt as revealed by the elementary analysis applying EDX
(Tab. 5.1).
Tab. 5.1: Results of elementary analysis of Co/Ru-Ref catalysts determined by EDX
examinations
Component Sample Substrate Sphere
[wt%] Aa) [wt%] Ba) [wt%] Ca) [wt%] Da) [wt%]
Ti 53.5 22.4 49.4 13.4 24.1
O 36.3 75.7 50.3 2.9 32.7
Co 9.1 1.8 0.02 78.2 40.5
Ru 0.9 0.0 0.1 5.3 2.6
a) refer to Fig. 5.7
Fig. 5.7: TEM photography of Co/Ru-Ref (resolution 0.5 nm, magnification 3·105)
DRIFT Adsorption Spectra
The DRIFT-spectra measured during exposure of a gaseous mixture of CO, H2 and
N2 to Co/Ru-Ref catalyst under atmospheric pressure at 200°C indicated the forma-
tion of an alkene-species (band at 862 cm-1, Fig. 5.8), beside gaseous carbon dioxide
5 Results and Discussion 57
(band at 1542 cm-1) already after one minute of reaction. This result was to be ex-
pected by suggesting the previously reported mechanistic models in which an alkene
species plays a key-role in the chain propagation. After a short period of time (6 min.)
bands of adsorbed alkane-species (2783 cm-1) in addition to adsorbed CO
(2029 cm-1) were also visible. An evidence for bridge-bonded carbon monoxide was
not detected. The assignment of the resonance frequencies are in good agreement
with results obtained by FREDRIKSEN et al. [175] and Price et al. [176].
Tab. 5.2: Assignment of the bands from in-situ DRIFT experiments over Co/Ru-Ref
The variation of the preparation procedure when using cobalt nitrate did not effect the
cobalt surface ratio. The obtained Co/Ti values were between 0.39 (Co-Ref) and 0.41
(IW-NIT-Step).
The total cobalt content of the prepared catalysts was determined by ICP-OES (see
Tab. 5.3) The total Co content ranges from 10.4 to 12.1 %. On the reference catalyst
a DCo of 1.6 % and a DCored of 6.1 % was obtained. An increase in cobalt dispersion
up to 1.8 % (DCored = 7.8 %) was achieved on the catalyst IW-ACE and IW-NIT-AC.
On the catalysts IW-ACAC3-Ru and IW-NIT-Step the value of DCo amounted to of
1.6 % but the DCored was to 8.5 and 6.3 %, respectively. Furthermore, it should be
mentioned that Ru addition to IW-ACAC3 resulted in an 31 % higher dispersion in
comparison to the undoped sample. For the other catalysts the dispersion was be-
tween 1.0 and 1.4 %.
Tab. 5.3: Overview of Co/Ti ratio, bulk composition, cobalt dispersion, degree of re-
duction and DCored of impregnated catalysts
Catalyst Co/Ti
(XPS)
Co [wt%]
(ICP)
DCo [%]a)
(CO-pulse)
Red. [%]
(TPR)
DCored [%]b)
(CO-pulse)
Co-Ref 0.39 12.0 1.6 73.7 6.1
IW-ACAC2 0.40 11.1 1.2 76.0 5.0
IW-ACAC3 0.78 11.7 1.1 88.7 9.8
IW-ACAC3-Ru 0.58 10.4 1.6 81.8 8.5
IW-ACE 0.44 11.0 1.8 76.9 7.8
IW-NIT-AC 0.41 12.1 1.8 76.9 7.8
IW-NIT-Step 0.40 10.7 1.6 74.6 6.3
IW-OXA 0.44 10.3 1.0 56.5 2.3
IW-OXA-NH3 0.54 11.8 1.4 74.0 5.4
a) DCo = mol CO pulsed / mol Co on catalyst b) DCored = mol CO pulsed / mol Co in metallic state
Summary of Characterisation Results Obtained on Impregnated Catalysts
Cobalt oxide (Co3O4) was present on all catalysts as derived from XRD. Only in the
case of IW-ACAC3 an additional crystalline phase, namely CoTiO3, was observed; by
doping of IW-ACAC3 with Ru the formation of a cobalt titanate was not observed
anymore.
On catalysts prepared from organic and complex precursors only one peak for the
reduction of cobalt oxide was obtained (TPR). The use of different solvents starting
5 Results and Discussion 68
from the same cobalt precursor resulted in a lower reduction temperature in compari-
son to catalysts prepared from an aqueous cobalt precursor solutions (TPR).
An raise in cobalt dispersion (DCored) was obtained applying cobalt acetate (DCored =
7.8 %) and (III) acetyl acetonate (DCored = 9.8 %) instead of cobalt nitrate (DCored =
6.1 %) for impregnation as derived from CO-pulse experiments; further an enrich-
ment of surface cobalt (described by Co/Ti ratio) was obtained (XPS).
5.1.3. CHARACTERISATION OF COBALT BASED CATALYSTS SUPPORTED ON
CERIA, ZIRCONIA AND TITANIA (RUTILE TYPE)
XRD
When ceria was used as support no crystalline cobalt oxide neither Co3O4 nor CoO
beside the interference pattern of crystalline CeO2 was detected by XRD examina-
tions (see Fig. 5.19). On zirconia supported catalyst, crystalline cobalt oxide Co3O4
was formed during precursor decomposition and the peaks were very narrow. For
IWB-NIT catalyst, only titania in its rutile modification and crystalline Co3O4 were pre-
sent.
10 20 30 40 50 60 70
IWC-NIT
IWB-NIT
IWZ-NIT
(O)(O)(O)(O)(O) (O)
Inte
nsi
ty /
a.u
.
2 θ
Fig. 5.19: XRD pattern obtained within a 2θ range from 20° to 70° of oxidised cata-
lysts prepared ex cobalt nitrate supported on ceria (IWC-NIT), titania-rutile type (IWB-
NIT) and zirconia (IWZ-NIT); (O) = Co3O4
5 Results and Discussion 69
TPR-results
The TPR-profiles of cobalt supported on ceria and zirconia were similar to that ob-
tained for the reference catalyst. The profile was composed of one shoulder and one
maximum. As plotted in Fig. 5.20, the shoulder for the catalyst IWC-NIT was more
distinctive than for the IWZ-NIT catalyst. The TPR-profile indicates that amorphous
cobalt oxide must be present on ceria (IWC-NIT) because of the well defined re-
duction peaks for Co3O4. Furthermore, the catalyst IWC-NIT was more easily to re-
duce; the reduction peak maximum was located at 365 °C. For the catalyst supported
on zirconia a temperature of 395 °C was required to obtain a reduced catalyst.
The cobalt catalyst IWB-NIT which was supported on Bayer-titania (rutile type)
showed only one reduction maximum which was located at 265 °C. This temperature
was about 130 °C lower than on the reference catalyst Co-Ref as described in
section 5.1.1.
20 40 60
331°C392°C
380°C
T
/ °
C
time on stream / min
H 2
- co
nsum
ptio
n /
a.
u.
IWZ-NIT
0
100
200
300
400314°C
251°C
IWB-NIT
IWC-NIT
Fig. 5.20: TPR-profiles of catalysts supported on various materials (ramp rate: 10
K/min, reducing agent: H2 (5%) in He, flow: 30 ml/min)
XPS, ICP and CO-pulse examinations
As shown in Tab. 5.4 on all supports an enrichment of cobalt on the samples surface
was detected in comparison to the Co-Ref catalyst (Co/Ti = 0.39). The highest
Co/support ratio was observed for the zirconia supported catalysts (0.59). As shown
5 Results and Discussion 70
by means of ICP examinations in all cases the desired amount of approx. 12 wt%
cobalt was achieved.
An improvement of cobalt dispersion was achieved on the IWC-NIT catalyst (see
Tab. 5.4). On this sample the DCO and DCored amounted to 2.4 % and 7.7 %, respec-
tively. The IWB-catalyst (pure rutile type of titania) showed a little higher DCO of 1.9 %
in comparison to Co-Ref catalyst prepared on Degussa P25 in which a mixture of ru-
tile and anatase was present. However, the DCored value derived was to 4.2 % in
comparison to 6.1 % for the reference catalyst due to the low amount of reduced co-
balt. On the zirconia supported IWZ-NIT catalyst only a cobalt dispersion value of
1.0 % was obtained.
Tab. 5.4: Overview of Co/support ratio, cobalt content, cobalt dispersion, degree of
reduction and DCored for cobalt catalysts supported on different carriers.
Catalyst Co/Sup ratio
(XPS)
Co [wt%]
(ICP)
DCo [%]
(CO-pulse)
Red. [%]
(TPR)
DCored [%]
(CO-pulse)
Co-Ref 0.39 12.0 1.7 73.7 6.1
IWB-NIT 0.49 11.8 1.9 54.7 4.2
IWC-NIT 0.52 12.5 2.4 68.8 7.7
IWZ-NIT 0.59 12.0 1.0 76.1 4.2
a) DCo = mol CO pulsed / mol Co on catalyst b) DCored = mol CO pulsed / mol Co in metallic state
Summary of Characterisation Results Obtained on Cobalt Based Catalyst Sup-ported on Ceria, Titania (rutile) and Zirconia
Co3O4 was present on all catalysts with the exception of IWC-NIT (XRD). Therefore,
amorphous cobalt oxide must be formed on IWC-NIT during calcination because
from TPR studies it can be concluded that Co3O4 was on the catalyst surface. The
temperature, necessary for the reduction of cobalt oxide, was lower on all catalyst in
comparison to Co-Ref.
Only on IWC-NIT an improvement of DCored (7.7 %) was obtained compared to Co-
Ref (6.1 %).
5 Results and Discussion 71
5.1.4. CHARACTERISATION OF SPREADED CATALYSTS
The results obtained on the catalysts SPR-Co3O4, SPR-CoTiO3, SPR-OXA and SPR-
OXA-Ru are given afterwards (for explanation of the applied catalyst name refers to
section 4.2.3).
XRD
The XRD-pattern of the SPR-CoTiO3 and SPR-Co3O4 catalyst is plotted in Fig. 5.21.
Titania was present in both its modifications i.e., rutile and anatase in both cases. On
the catalysts prepared from Co3O4 and CoTiO3 no other crystalline phase beside
those of the applied precursors was observed; that means that no phase transforma-
tion of Co3O4 to CoTiO3 or reverse, occurred.
On the SPR-OXA catalyst as well as on SPR-OXA-Ru no phenomena worth men-
tioning occurred. Only titania and cobalt oxide (Co3O4) were present. However, the
addition of Ru did not effect the phase composition of the catalyst.
TPR
As presented in Fig. 5.22 the TPR-profiles of pure Co3O4 are given along with the
supported cobalt oxide (SPR-Co3O4) as well as pure CoTiO3 beside SPR-CoTiO3.
The TPR- pattern of the pure cobalt oxide consists of two maximum located at
346 °C and 384 °C in contrast to the SPR-Co3O4 catalyst which consists of only one
reduction peak (378 °C). Furthermore the reduction maximum of the latter was about
6 K lower in comparison to pure cobalt oxide. An opposite behaviour was observed
for SPR-CoTiO3; for the supported cobalt titanate a 17 K higher reduction tempera-
ture was necessary.
The reducibility of SPR-OXA catalyst is illustrated in Fig. 5.23 in comparison to the
doped catalysts SPR-OXA-Ru. In the spectrum of the SPR-OXA only one reduction
peak located at 384 °C exists. For the doped catalyst, two peaks were detected. The
presence of ruthenium leads to a decreasing reduction temperature, the major re-
duction peak was shifted to 352 °C. A small reduction peak was visible at 203 °C.
5 Results and Discussion 72
20 30 40 50 60 70
CoTiO3CoTiO3
CoTiO3CoTiO3
Co3O4
Co3O4
Co3O4
SPR-Co3O4
SPR-CoTiO3In
ten
sity
/
a
.u.
2 θ
Fig. 5.21: XRD-spectra of oxidised SPR-CoTiO3 and SPR-Co3O4 catalysts obtainedwithin a 2θ range from 20° to 70°.
100 200 300 400 500
384°C346°C
407°C
378°C
390°C
SPR-CoTiO3
pure Co3O4
pureCoTiO3
SPR-Co3O4
H 2
- c
onsu
mpt
ion
/
a.u.
T / °C
Fig. 5.22: TPR-profiles of pure Co3O4 and CoTiO3 and spreaded catalysts SPR-
Co3O4 and SPR-CoTiO3 (ramp rate: 10 K/min, reducing agent: H2 (5%) in He).
5 Results and Discussion 73
0 10 20 30 40 50 60
SPR-OXA-Ru
SPR-OXA
H2-
cons
umpt
ion
/
a.u.
time on stream / min
0
100
200
300
400352 °C
384 °C
203 °C
Fig. 5.23: TPR-plot of doped and undoped SPR-OXA catalyst (ramp rate: 10 K/min,
reducing agent: H2 (5%) in He, flow: 30 ml/min)
XPS-, ICP-OES-and CO-pulse measurements
On the catalysts starting from pure Co3O4 or CoTiO3 a decreasing surface content of
cobalt was observed (see Tab. 5.5).The Co/Ti ratio amounted to 0.07 (SPR-CoTiO3)
and 0.25 (SPR-Co3O4), respectively. On the other side the use of cobalt oxalate led
to an increasing Co/Ti ratio (0.53) in comparison to Co-Ref catalyst (0.39); this find-
ing was similar to the catalysts prepared from organic cobalt precursors by means of
the impregnation technique. On the doped catalyst SPR-OXA-Ru an increasing Co/Ti
ratio of 0.53 to 0.56 compared to SPR-OXA was observed.
An increase of cobalt dispersion on the catalysts prepared by the spreading method
was not observed. On SPR-Co3O4 and SPR-CoTiO3 the lowest DCo of all prepared
catalyst was obtained with DCo values of 0.3 % and 0.2 %. On SPR-OXA the cobalt
dispersion was to 1.0 % and a slight increase to 1.2 % was observed after the addi-
tion of ruthenium as promoter. However, that was no improvement in comparison to
the catalyst CO-Ref (1.6 %). But when the cobalt dispersion was calculated based on
the amount of reduced cobalt a raise in DCored to 7.7 % (SPR-OXA) and 8.3 % (SPR-
OXA-Ru) compared to the reference catalyst (Co-Ref) was obtained.
5 Results and Discussion 74
Tab. 5.5: Overview of Co/Ti ratio, cobalt content, cobalt dispersion, degree of reduc-
tion and DCored for catalysts prepared by spreading of cobalt precursor
Catalyst Co/Ti
(XPS)
Co/Ti
(ICP)
Co [wt%]
(ICP)
DCo [%]a)
(CO-pulse)
Red. [%]
(TPR)
DCored [%]b)
(CO-pulse)
Co-Ref 0.39 0.16 12.0 1.6 73.7 6.4
SPR-Co3O4 0.25 0.14 10.0 0.3 93.6 4.7
SPR-CoTiO3 0.07 0.14 10.3 0.2 80.0 1.0
SPR-OXA 0.53 0.16 11.5 1.0 87.1 7.7
SPR-OXA-Ru 0.56 0.16 11.5 1.2 85.5 8.3
a) DCo = mol CO pulsed / mol Co on catalyst b) DCored = mol CO pulsed / mol Co in metallic state
Summary of Characterisation Results Obtained on Catalysts Prepared bySpreading
XRD examinations revealed no special feature than the expected pattern of CO3O4
and CoTiO3 beside titania.
In the case of SPR-OXA the addition of Ru influenced the reduction behaviour of the
catalyst; a 32 K lower reduction temperature was observed on SPR-OXA-Ru com-
pared to SPR-OXA. Additionally, a second reduction maximum at 203 °C was ob-
tained (TPR). Furthermore, an improved cobalt dispersion of 7.7 % (SPR-OXA) and
8.3 % (SPR-OXA-Ru) in comparison to 6.1 % (Co-Ref) was achieved (CO-pulse).
5.1.5. CHARACTERISATION OF PRECIPITATED CATALYST
Five catalysts were prepared applying the precipitation method, namely PR-8, PR-12,
PR-12-Na, PR-EDTA, and PR-EDTA-Ru. The data obtained during characterisation
measurements are given afterwards (for explanation of the applied catalyst name
refers to section 4.2.3).
XRD
As an example for the phase composition of the precipitated catalysts the XRD
spectra of catalyst PR-12 is shown in Fig. 5.24. Besides titania in its both modifica-
tions crystalline cobalt oxide (Co3O4) was present. In comparison to catalyst prepared
applying the incipient wetness technique with cobalt nitrate as source material the
XRD-peaks are broader, which indicates the presence of large cobalt clusters on the
catalyst surface. The catalysts, PR-EDTA, PR-8, PR-12-Na and PR-12-K were also
examined, but no peculiarities were observed; so that these XRD-plots are not given.
5 Results and Discussion 75
20 30 40 50 60 70
(O)
(O)
(O)
PR-12
Inte
nsi
ty
/
a.u
.
2 θ
Fig. 5.24: XRD-profile of the oxidised, precipitated catalyst PR-12 catalysts obtained
within a 2θ range from 20° to 70°; (o) = Co3O4
TPR
In Fig. 5.25 the TPR-profile for the preticipated catalyst PR-12 along with that of the
catalyst PR-EDTA are plotted. Two shoulders at 348 °C and 400 °C (the shoulder
was obtained after 8 min within the isothermal zone at the end of the TPR run) and
one maximum at 400°C can be detected for PR-12. By doping the catalyst with
sodium no significant influence on the reduction behaviour was observed. The
slightly lower reduction peaks at 346 °C and 397 °C were within the margins of
errors. The change of the pH-value to 8 shows in contrast to PR-12 catalyst a
reduction maximum located 13 K lower. Further was no additional shoulder within the
isothermal zone detected (please refer to Tab. 5.6 for TPR-results). On catalyst PR-
EDTA only one reduction maximum was observed. This is in accordence with results
obtained for other catalyst (e.g. IW-ACE, IW-ACAC2, IW-ACAC3) prepared from
organic cobalt precursors. For the doped catalyst PR-EDTA-Ru, the reduction
maximum was at 307 °C.
5 Results and Discussion 76
10 20 30 40 50 60
400°C
400°C8 min. isothermal
348°CPR-12
H2-
upta
ke /
a.u.
time on stream / min.
0
100
200
300
400
311°CPR-EDTA
T / °C
Fig. 5.25: TPR-Profile of PR-12 und PR-EDTA catalyst as a function of time onstream (ramp rate: 10 K/min, reducing agent: H2 (5%) in He, flow: 30 ml/min)
Tab. 5.6: Overview of resolved reduction maximum for precipitated catalysts as de-
rived from TPR experiments (temperature range: 25- 400 °C, ramp rate: 10 K/min,
reducing agent: H2 (5%) in He, flow: 30 ml/min)
Catalyst Reduction maximum [°C]
PR-8 337(m), 384(m)
PR-12 346(s), 400 (m), 400(si)
PR-12-Na 346(s) 397 (m), 400 (si)
PR-EDTA 311(m)
PR-EDTA-Ru 307(m)
(m) = maximum, (s) = shoulder, (si) = shoulder within the isothermal zone
XPS, ICP and CO-pulse-measurements
An overview of surface and bulk composition is given in Tab. 5.7 for the precipitated
catalysts. Based on results obtained from precipitation at pH 8 or 12 it can be ascer-
tained that an increase of the pH-value leads to an increase of the cobalt amount (ex-
pressed as Co/Ti) from 0.57 and 0.83 on the catalyst surface. The use of Na as pro-
moter resulted in a significant decrease of the Co/Ti surface ratio from 0.83 to 0.67.
5 Results and Discussion 77
The same effect was noticed by doping of PR-EDTA catalyst with ruthenium but in
that case the difference between the two samples is minimal and amounted to 0.58
(PR-EDTA) and 0.53 (PR-EDTA-Ru), respectively.
Tab. 5.7: Overview of Co/Ti ratio, bulk composition, cobalt dispersion, degree of re-
duction and DCored of precipitated catalysts
Catalyst Co/Ti
(XPS)
Co [wt%]
(ICP)
DCo [%]a)
(CO-pulse)
Red. [%]
(TPR)
DCored [%]b)
(CO-pulse)
Co-Ref 0.39 12.0 1.7 73.7 6.1
PR-12 0.83 10.4 1.7 75.7 7.0
PR-12Na 0.67 10.7 0.2 93.5 3.1
PR-8 0.57 11.1 1.3 71.1 4.5
PR-EDTA 0.58 10.3 1.3 79.3 6.3
PR-EDTA-Ru 0.53 11.2 1.2 81.5 6.5
a) DCo = mol CO pulsed / mol Co on catalyst b) DCored = mol CO pulsed / mol Co in metallic state
As reported previously the chosen pH value influenced the Co/Ti ratio. In the same
manner DCo was effected; the cobalt dispersion increased from 1.3 % (PR-8) to 1.7 %
(PR-12) and was in the same magnitude than determined for the Co-Ref catalyst
(please refer to Tab. 5.7). The addition of Na and Ru, respectively, resulted in a de-
creasing cobalt dispersion. Similar as reported for XPS results this effect was more
strongly marked on the PR-12 catalyst.
Summary of Characterisation Results Obtained on Precipitated Catalysts
On the precipitated catalysts was the only crystalline phase beside titania in its both
modification; but the broad interference peaks indicate that larger cobalt clusters than
on impregnated catalysts were formed (XRD).
The necessary reduction temperature was for all catalysts prepared from Co(OH)2
higher than on Co-Ref. On PR-EDTA catalysts an about 70 K lower reduction tem-
perature was achieved compared to Co-Ref; An increase in DCo was noticed also.
5.1.6. CHARACTERISATION OF CATALYST APPLYING PLASMA INDUCED
PREPARATION
The plasma technique was applied in order to introduce a novel preparation tech-
nique to the FTS. The catalysts PD-Nit, PD-ACAC2, PD-ACE were prepared by the
5 Results and Discussion 78
decomposition of previously impregnated catalyst within an oxygen plasma. Cobalt
metal was sputtered for PS-Co catalyst. PL-AT, PL-PP, PL-100W and PL-150W
catalysts were prepared by a decomposition of a mechanical mixture of titania and
cobalt acetyl acetonate. The physico-chemical properties for the catalysts are de-
scribed afterwards (for explanation of the applied catalyst name refer to section
4.2.3).
XRD
XRD measurements on PD-ACAC2, PD-ACE and PD-Nit resolved only small
amounts of Co3O4 beside species of cobalt precursor as can be derived out of the in-
tensity of the cobalt oxide reference peaks; the amount of crystalline cobalt was to
only 0.7 % related to the area of the interference peaks of titania.
From the XRD examination the presence of cobalt oxide (see Fig. 5.26) could not be
proved on the PS-Co catalyst because the specific interference maximum were not
visible. An other point of interest is that the typical distribution of titania phases, e.g.;
rutile and anatase, had changed; on the plasma catalyst rutile is the only detectable
phase. For the catalysts PL-100W, PL-150W, PL-AT and PL-PP not worth mention-
ing occurrences were detected; neither a change within the rutile – anatase ratio nor
the absent of Co3O4 as crystalline phase.
10 20 30 40 50 60 70
(O)
(O)(O)(O)
(O)
(O)(O)
PS-Co catalyst
= expected Co3O
4 signals
Inte
nts
ity /
a.u
.
2 θ
Fig. 5.26: XRD-spectra of the plasma prepared catalyst PS-Co obtained within a 2θrange from 20° to 70°; (o) = Co3O4
5 Results and Discussion 79
TPR
All catalysts of the PD series showed reduction peaks known for transformation of
Co3O4 to CoO (between 310 °C and 320 °C) and for CoO to Co0 (between 380 °C
and 400 °C). The catalyst prepared from cobalt nitrate (PD-NIT) showed an addi-
tional peak at 230 °C. The catalysts prepared from organic cobalt precursors showed
additional peaks at 250 °C (PD-ACE) and 230 °C (PD-ACAC) also.
The TPR profiles of PL-100W and PL-150W catalyst are presented in Fig. 5.27. This
plot consists of one maximum at 238 °C and a adjacent shoulder at 345 °C. On PL-
150W a maximum at 243 °C was obtained but the baseline was not reached and until
the end of the TPR run a hydrogen consumption was visible. The TPR plots for PL-
AT and PL-PP showed only one reduction maximum located at 379 °C and 376 °C,
respectively.
Tab. 5.8: Overview of resolved reduction maximum for plasma- induced catalysts asderived from TPR experiments (temperature range: 25- 400 °C, ramp rate: 10 K/min,reducing agent: H2 (5%) in He, flow: 30 ml/min)
Catalyst Reduction maximum [°C]
PD-ACAC 230 (m), 314 (s), 392 (m)
PD-ACE 250 (m), 310 (m), 399 (m)
PD-NIT 230 (m), 317 (s), 386 (m)
PL-100 W 238 (m), 345(s)
PL-150W 243 (s), 320 - 400(si)
PL-AT 379 (m)
PL-PP 376 (m)
(m) = maximum, (s) = shoulder, (si) = shoulder within the isothermal zone
5 Results and Discussion 80
10 20 30 40 50 60
PL-150W
H2-
cons
umpt
ion
time on stream / min
0
100
200
300
400
T / °C
238 °C
243 °C
PL-100W
Fig. 5.27: TPR-Profile of PL-100W and PL-150W (ramp rate: 10 K/min, reducing
agent: H2 (5%) in He, flow: 30 ml/min)
ICP examination and CO-pulse experiments
The change of preparation procedure effected on the one hand the amount of loaded
cobalt on the support, on the other hand the cobalt dispersion. First, the cobalt con-
tent was more than 5 times higher comparing to PL-100W with PL-AT. Furthermore,
an increasing DCo from 0.4 % to 1.8 % was achieved and this was similar to the ref-
erence catalyst (see Tab. 5.9).
Tab. 5.9: Overview of bulk composition, cobalt dispersion, degree of reduction and
DCored obtained for plasma catalysts
catalyst Co [wt%]
(ICP)
DCo [%]
(CO-pulse)
Red. [%]
(TPR)
DCored [%]
(CO-pulse)
Co-Ref 12.0 1.7 73.7 6.1
PL-100W 1.0 0.4 16.2 1.2
PL-150W 1.2 0.3 33.2 0.9
PL-AT 5.5 1.8 71.1 7.0
PL-PP 6.5 0.6 75.7 3.1
a) DCo = mol CO pulsed / mol Co on catalyst b) DCored = mol CO pulsed / mol Co in metallic state
5 Results and Discussion 81
5.1.7. DISCUSSION OF CHARACTERISATION RESULTS
First, the characterisation results obtained for the reference catalysts Co-Ref and
Co/Ru-Ref will be discussed; subsequently the findings on the new catalysts ar-
ranged by the applied preparation technique will be commented.
Reference Catalyst
Bulk composition of reference catalysts
As shown by the XRD-results the phase composition of the reference catalyst (Co-
Ref) and the doped catalyst (Co/Ru-Ref) was identical. However, a change of phase
composition was not expected because of the low amount of added Ru. Neverthe-
less, the formation of CoRuO4 cannot completely be excluded because such small
quantities might be below the detection limit.
Reducibility of reference catalysts
The TPR profiles of the doped and undoped catalysts consist of two separate peaks.
For both samples, the maximum can be assigned to the stepwise reduction of Co3O4
(maximum 1) via CoO (maximum 2) to Co0 as described by stoichiometric formulae
(5.1) and (5.2). This result agrees with TPR-examinations reported in literature [5].
OH CoO 3 H OCo 2243 +→+ (5.1)
OH 3 Co 3 H 3 CoO 3 20
2 +→+ (5.2)
As revealed by the TPR-results (see Fig. 5.2 in section 5.1.1) the addition of small
amounts of Ruthenium to the cobalt reference catalyst resulted in a decreasing re-
duction temperature of 55 K in comparison to Co-Ref (reduction maximum at
395 °C). This may be explained by interactions between cobalt and ruthenium in-
duced by the higher mobility of Ru oxides as well as by the formation of Co-Ru ox-
ides [12]. This assumption is supported by the fact that Co2RuO4 forms a spinel
isostructural to Co3O4 [177]. On the other hand one can assume that in the presence
of Ru the reduction of Co is accelerated by a spill-over process, i.e., hydrogen ad-
sorbed on Ru has a higher adsorption enthalpy than on cobalt. The adsorbed hydro-
gen (Ru-H2) dissociate and lead to activated hydrogen (H-Ru-H) which then can dif-
fuse into the neighbouring cobalt oxide [178]. This assumption is supported by TAP
data on hydrogen interaction with the surface of Co-Ref and Co/Ru-Ref catalyst. On
the Ru-doped reference catalyst (Co/Ru-Ref) a longer mean residence time in com-
parison to the undoped catalyst was derived; this finding can be correlated with a
stronger interaction of Co/Ru-Ref catalyst with hydrogen than on Co-Ref catalyst or
in turn a stronger adsorption of H2 Co/Ru-Ref was due to the promoter. However, the
obtained TAP data were not sufficient to prove the latter assumption entirely; further
5 Results and Discussion 82
studies will be necessary.
Oxidation behaviour of reference catalysts
The observed decrease in re-oxidation temperature (as derived by TPO experiments
given in section 5.1.1, Fig. 5.3) for the doped catalyst Co/Ru-Ref (280 °C) in com-
parison to the undoped catalyst Co-Ref (303 °C) might also be due to an oxygen ac-
tivation over Ru, since on Ru oxygen spill-over was reported, also [178].
Adsorption behaviour of CO on reference catalysts
For TPD experiments the reduced catalysts surface was covered with carbon mon-
oxide and in the case of Co/Ru-Ref catalyst 3 desorption peaks (130 °C, 227 °C,
340 °C) and one desorption peak (107 °C) for Co-Ref was obtained as shown before
in Fig. 5.4. Although the existence of bridged bonded carbon monoxide beside a
linearly one is conceivable only linearly adsorbed carbon monoxide was present on
the catalysts surface as shown by DRIFT (see Fig. 5.8). Additionally, this finding jus-
tifies the applied CO-pulse method for the determination of cobalt dispersion DCo and
DCored because each adsorbed CO molecule can be counted to one cobalt atom.
One explanation for the peaks at 130 °C and 227 °C is that two different adsorption
sites are present on the catalyst. The first one can be assigned to oxidised cobalt and
the second one to reduced cobalt. This assumption is in good agreement with results
obtained by CHOI et al. [179]. The observed shoulders at 340 °C on Co/Ru-Ref cata-
lyst and at 410 °C for the undoped catalysts originate from surface carbon species
which combine with oxygen from the catalysts structure, instead from residual gases
such as CO and CO2 at elevated temperature. It can be assumed that this carbon
species was formed during the BOUDOUARD reaction. Another hint for the BOUDOUARD
reaction was given by TAP data. During the pulsing of CO over the reduced refer-
ence catalysts CO2 was found.
Surface composition of reference catalysts
As shown by XPS oxidation and reduction of the reference catalyst does not affect
the binding energies of TiO2 and Co3O4 within the Ti- and Co- region, or more precise
the distance between the 2p1/2 and the 2p
3/2 photoemission lines, for the reference
catalyst (see Fig. 5.6 in section 5.1.1. Pseudo in-situ XPS Examination). Therefore, it
can be concluded that no accountable formation of bimetallic mixed oxides took
place, as was reported to occur on supports like SiO2 [180] and Al2O3 [181]. During
pseudo in-situ XPS measurements the formation of a carbide species was observed.
Probably this CoC2 was formed in the Fischer-Tropsch reaction. The presence of co-
balt carbide was reported by CHAUMETTE et al. [182] on a Co/TiO2 catalyst. This result
indicates that under high pressure condition cobalt carbide might be involved in the
Fischer-Tropsch synthesis. The carbide mechanism can be seen as a parallel reac-
5 Results and Discussion 83
tion pathway beside the CO-insertion mechanism.
According to TEM studies an accumulation of cobalt within separated spheres were
detected (see Fig. 5.7). This result may explain the small cobalt-dispersion and sup-
port the studies of PUSKAS et al. [183] carried out on promoted cobalt catalysts in
which a sphere formation was correlated with a low cobalt dispersion. From the low
amount of Co found in the substrate by EDX it can be concluded that only a small
amount of cobalt migrated into the support material; this explains also the enrichment
of the catalysts surface as derived by XPS and expressed by the increasing Co/Ti ra-
tio.
Catalysts Prepared by Incipient Wetness Technique
Bulk and surface composition of impregnated catalysts
As revealed by XRD the formation of a cobalt titanate (CoTiO3) was observed on IW-
ACAC3 catalyst (see Fig. 5.14 in section 5.1.2). In addition the highest surface Co/Ti
ratio of all impregnated catalysts was obtained on IW-ACAC3 catalyst (Co/Ti = 0.78)
in comparison to Co-Ref catalyst (Co/Ti = 0.39) as derived by XPS (see Tab. 5.3).
Further, it should be mentioned that this enrichment of surface cobalt was observed
for all organic cobalt precursors in comparison to a catalyst made from inorganic
salts like cobalt nitrate. This effect is significant and can be related to the applied
precursor; comparable results were obtained according to a study by NIEMELÄ et al.
[25]. However, the mechanism of how the cobalt precursor interacts with the support
material is not solved up to now.
One can assume that a high Co/Ti ratio should also lead to a high cobalt dispersion
but the opposite is the case with the exception of IW-ACE. As derived from TEM ex-
periments for Co-Ref catalyst the cobalt oxide particles are arranged as spheres on
the support. During XPS experiments only uppermost located species will be de-
tected (5 to 10 atomic layers); therefore a high Co/Ti ratio indicates that compara-
tively large particles are present which in turn resulted in a low cobalt dispersion.
Similar conclusions were drawn by NIEMANTSVERDRIET [184] and LINDNER and PAPP
[185]. A reason for the high cobalt dispersion on IW-ACE might be the sorption of
Co2+ ions on titania in an acetic solution according to NICHOLSON [186], which is not
the case for other cobalt salts.
Reducibility of impregnated catalysts
For the organic cobalt precursors as well as for the stepwise prepared catalysts ex
cobalt nitrate (IW-NIT-Step) only one TPR peak was resolved within a temperature
range from 332 °C for IW-ACE catalyst to 397 °C for IW-ACAC3 catalyst (see Fig.
5.16 and Fig. 5.17). One explanation could be that only CoO was on the supports
surface. But this assumption is not supported by the XRD results in which exclusive
5 Results and Discussion 84
Co3O4 was detected. Therefore, another explanation can be given following the
studies carried out by WANG and CHEN [92]. In this work only one broad peak was
obtained during the reduction of pure Co3O4. The authors assigned it to the stepwise
reduction of cobalt oxide over Co3+ → Co2+→ Co0 and the broad peak results from an
overlap of both reduction steps.
Furthermore the reduction behaviour of the undoped and promoted IW-ACAC3 cata-
lyst should be discussed; on IW-ACAC3 catalyst only one reduction maximun located
at 397 °C in contrast to IW-ACAC3 catalyst on which two peaks at 183 °C and 319 °C
were obtained (see Fig. 5.17). The decreasing reduction temperature can be as-
cribed to the presence of Ru only and could be assigned with an easier reduction of
cobalt in the vicinity of Ru. This result led to the conclusion that bimetallic effects that
can certainly be ascribed to the close contact of the two metals affect the reduction
process. Similarly, a decreasing reduction temperature after doping with ruthenium
was also observed by thermogravimetric measurements carried out by IGLESIA et al.
[12,187] on a cobalt catalyst made from cobalt nitrate.
Catalysts Prepared by Incipient Wetness on Ceria, Zirconia and Titania (rutile)
Bulk and surface composition of cobalt catalysts supported on Ceria, Zirconia and
Titania (rutile)
The XRD results for the IWC-NIT (Fig. 5.20 in section 5.1.3) catalyst indicated that no
crystalline cobalt oxide was present on the supports surface because no interference
peaks according to Co3O4 were detected. On the other hand the presence of cobalt
oxide could be derived from TPR studies; therefore the cobalt oxide must be present
in an amorphous state. It seems that strong interactions between the support
material (ceria) and the cobalt nitrate supressed the formation of crystalline cobalt
oxide or reduced it to such small amounts that they were not detectable by XRD.
These interactions cannot be excluded for zirconia and titania (rutile-typ) supported
catalysts but it can be assumed that on these catalysts the SMSI effects was not that
marked. On the other hand the high cobalt dispersion of the IWC-NIT catalyst
(DCo = 2.4 %) in comparison to Co-Ref catalyst (DCo = 1.6 %) can be correlated with
the amorphous surface cobalt. Within an amorphous oxide no near order is effective,
i.e., the cobalt particles were not arranged in a well-defined crystalline structure.
Therefore, a great number of defect sites should be present which in turn affect the
number of accesible cobalt [188].
Reducibility of cobalt catalyst supported on Ceria, Zirconia and Titania (rutile)
The lower reduction temperature of IWB-NIT catalyst (265 °C) in comparison to Co-
Ref catalyst (395 °C) can be explained by reduced SMSI effect [189] due to the
different phase composition (see Fig. 5.20). A similar support effect on dispersion
5 Results and Discussion 85
and reduction behaviour was reported by BARTHOLOMEW et al. [190] and LAPIDUS et
al. [191] on a nickel catalyst and by PONEC and NONNEMAN [192] on a Rh catalyst,
respectively.
Catalysts Prepared by Spreading of Cobalt Precursors
Bulk and surface composition of catalyst by spreading of cobalt precursor
The catalysts prepared from pure Co3O4 and CoTiO3 showed no exceptional obser-
vations by XRD; no transformation of Co3O4 to CoTiO3 and reversed was noticed
(see Fig. 5.21 in section 5.1.4). Further, it should be mentioned that no well dis-
persed cobalt on the catalysts could be achieved; on SPR-Co3O4 the value of DCo
amount to 0.2 and for SPR-CoTiO3 to 0.3. This might be due to a neglectable migra-
tion tendency of the pure phases. The use of cobalt oxalate resulted in a better dis-
persed system as derived by CO-pulse for SPR-OXA and was to DCo = 1.0 (see Tab.
5.5). An explanation might be that cobalt oxalate melt at a temperature of 263 °C so
that the melting can fill the supports pore system [166].The addition of Ru leads to a
small increase in cobalt dispersion.
Reducibility of catalysts prepared by spreading of cobalt precursor
In the case of SPR-CO3O4 the necessary reduction temperature decreased from
384 °C to 378 °C compared with pure Co3O4 (see Fig. 5.22). This effect was previ-
ously reported for Co/SiO2 catalysts [86] and may be subscribed to a greater surface
area of the supported cobalt oxide than for pure Co3O4. An explanation for the 17 K
higher reduction temperature (407 °C) on SPR-CoTiO3 in comparison to pure cobalt
titanate was not found, yet; further examination will be necessary.
The addition of Ru to SPR-OXA catalyst led to a similar decrease in reduction tem-
perature as reported for IW-ACAC3-Ru before catalysts, instead of one maximum at
384 °C two reduction peaks were obtained at 203 °C and 352 °C; this finding can
also related to the spill-over as well as bimetallic effect.
Catalysts Prepared by Precipitation
Bulk and surface composition of catalysts prepared by precipitation
From XRD studies no special features were obtained.
The observed high Co/Ti ratio of 0.83 on PR-12 can be explained by a distinct
hydroxylation of the support. It can be assumed that the precipitated cobalt precursor
was surrounded by Ti-OH groups which explains the high amount of surface cobalt,
because the migration into the supports lattice was extremly hindered; on the other
hand the hydroxylated titania species were responsible for a suppression of cobalt
clusters aggregation which leads to the high cobalt dispersion in comparison to the
5 Results and Discussion 86
PR-8 catalyst.
As derived from ICP-OES in no case the desired amount of 12 wt% on cobalt was
achieved, instead the Co contentent amounted from 10.1 to 11.0 wt%. This finding
can be explained by an incomplete precipitation of cobalt ions onto the support (see
Tab. 5.7 in section 5.1.5).
Reducibility of catalysts prepared by precipitation
The TPR-profiles for the cobalt hydroxide based catalyst showed additional peaks
within the isothermal region (400 °C) at the end of the TPR run up to now were not
noticed for the other new catalyst (see Tab. 5.6 and Fig. 5.25 in section 5.1.5). For
catalyst PR-12 the shoulder at 348 °C and the maximum at 400 °C could be related
to the reduction of Co3O4 to Co0. The second shoulder within the isothermal zone can
be assigned to the reduction of not totally decomposed Co(OH)2. The reduction of
cobalt hydroxide can be described as follows:
OH 2 Co H )OH(Co 20
22 +→+ (5.4)
Catalyst Prepared by Plasma- Induced Techniques
Surface composition for catalysts prepared by plasma- induced techniques
The catalyst prepared by sputtering of cobalt metal onto the support did not lead to
the desired improvement in DCo nor to the desired amount of cobalt (only 0.2 wt%) on
the support. This result led to the conclusion that the sputtered plasma technique has
to be modified in a way that a longer covering time or a higher plasma power should
be used.
The catalysts PL-AT and PL-PP (see Tab. 5.9 in section 5.1.6) which were mingled
with cobalt (III) acetyl acetonate and then decomposed by an oxygen plasma
achieved an acceptable metal loading (5.5 wt% for PL-AT and 6.5 wt% for PL-PP)
cobalt dispersion (DCo(PL-AT) = 1.8 %, DCo(PL-PP) = 0.6). Up to now no comparable
preparation methods were reported in the literature and it can be assumed that due
to the mixing of support and precursor an optimal layer thickness was received. This
layer was thin enough in order to guarantee a total decomposition of the precursor
and also thick enough to obtain the highest cobalt loading of all plasma prepared
catalyst
Reducibility of catalysts prepared by plasma- induced techniques
On the TPR – spectra of catalysts from the PD (decomposition of cobalt precursors
within an oxygen plasma) series more than the two typical reduction peaks for the
stepwise reduction of Co3+ via Co2+ (between 310 to 320 °C) to Co0 (between 380 to
5 Results and Discussion 87
400 °C) were detected (see Tab. 5.8). During TPR studies on PD-NIT catalyst an ad-
ditional peak at 230 °C was obtained which can be assigned to the decomposition of
NO3 to NO2. It may be assumed that only the first layer of the impregnated precursor
was decomposed and so the expected goal to obtain a well dispersed cobalt surface
layer was not reached. This result leads to the conclusion that the plasma-decom-
position is not useful for preparation of FTS catalysts in the case of impregnated
samples, because the oxygen plasma did not penetrate the whole coating of the co-
balt precursor
5.2. CATALYTIC EVALUATION
The catalytic results obtained during FTS tests in a fixed-bed reactor on the differ-
ently prepared catalyst samples are reported. The data are arranged by the applied
preparation technique.
5.2.1. REFERENCE CATALYST
The results on syngas conversion and product distribution for the reference catalyst
Co-Ref are listed in Tab. 5.10 as a function of time on stream. After an initial period
of 18 h the CO conversion amounted to about 7 % and reached a nearly constant
value of approximately 14 to 15 % after 70 to 140 h on stream. While CO conversion
increased the selectivity to methane dropped from approx. 15 % to a nearly constant
value of 10 %. Concomitantly, the selectivity towards the C5+ fraction increased to
79 % under pseudo steady–state conditions. The chain growth probability amounted
to 0.84 after the initial start-up procedure. A typical Schulz-Flory plot in for estimation
of the α-value is given in Fig. 5.28 acting for all catalysts.
For the Ru-doped catalyst, a very similar reaction performance was observed. After
24 h t.o.s. a carbon monoxide conversion of 8.1 % was obtained and raised over
12.6 % to 16.3 %. Later, constant XCO of approx. 8 % was observed for over 40 h.
The methane selectivity behaved in the same way and decreased with a drop in XCO
to 10 % under steady-state conditions. Although the reaction temperature was raised
to 210 °C after 136 h t.o.s. the prior XCO of 16 % could not be obtained anymore; this
indicates catalyst deactivation. As shown in Tab. 5.11 and in contrast to the unpro-
moted catalyst the formation of higher-hydrocarbons was favoured. This trend was
also expressed in a higher α-value of 0.90.
Summary of catalytic results obtained on reference catalysts
The level of carbon monoxide conversion was on both catalysts in same order of
magnitude about 15 %. The addition of Ru to the reference catalyst Co-Ref resulted
in an increasing α-value from 0.84 to 0.91 on coast of long-term stability. Co/Ru-Ref
5 Results and Discussion 88
catalyst deactivated after 102 h t.o.s. compared to Co-Ref; on the later XCO stayed
constant over the whole run.
0 5 10 15 20 25 30-4
-3
-2
-1
0
1
2
3
a = 0.84
ln (
mn/
n )
n
Fig. 5.28: Schulz-Flory plot for Co-Ref catalyst
Tab. 5.10: XCO, α and selectivities of hydrocarbons depended on t.o.s. obtained for
On the IW-OXA catalyst, only a very small carbon monoxide conversion of 5.2 % was
measured which was obtained after 23 h t.o.s. and did not change until the end of the
run. The methane selectivity was always higher than 16 % that led to a low SC5+ of
58.3 %. By changing the solvent during the impregnation procedure (IW-OXA-NH3
catalyst) a change in the activity level was observed; the average XCO amounted to
12.8 %, which was constant during 43 h. At the end of catalysts evaluation a drop in
conversion to 6.9 % was noticed. An α-value of 0.81 was derived from the Schulz-
Flory-Plot with a corresponding S5+ of 75.9 %.
Summary of catalytic results obtained on impregnated catalysts
In comparison to Co-Ref catalyst an significant improvement of carbon monoxide
conversion and TOF was achieved on IW-ACAC3 (XCO = 23.6 %, TOFnom = 2.5) and
IW-ACE (XCO = 26.4 %, TOFnom = 1.8) catalysts. In the case of IW-ACAC3 the doping
of catalyst with ruthenium resulted in an increasing value of XCO = 29.3 % and α =
0.80 (see Tab. 5.14).
The applied precursor cobalt (II) acetyl acetonate and oxalate failed because neither
a raise in carbon monoxide conversion (or TOF) nor chain growth probability was
achieved. Further, no improvement of the α-value was noticed with the exception of
IW-NIT-Step; on this catalyst the chain growth probability amounted to 0.90 com-
pared with Co-Ref (α = 0.83).
5 Results and Discussion 92
Tab. 5.14: Overview of carbon monoxide conversion, selectivity towards C5+ fraction,
α−value and TOF values for impregnated catalyst compared to Co-Ref (average for
steady-state conditions, Treac = 200 °C, GHSV = 1200 h-1, H2: CO: N2 = 12:6:2 bar,
ptot=20 bar)
Catalyst XCO [%] SC5+ [wt%] α [-] TOF [103s-1] a) TOFnom[-] b)
Co-Ref 14.7 80.0 0.83 17 1.0
IW-ACAC2 7.3 75.3 0.79 13 0.7
IW-ACAC3 23.6 67.9 0.71 43 2.5
IW-ACAC3-Ru 29.3 80.4 0.80 42 2.5
IW-ACE 26.4 66.9 0.74 32 1.8
IW-NIT-AC 19.7 77.2 0.83 21 1.3
IW-NIT-Step 16.8 83.2 0.90 23 1.3
IW-OXA 5.2 58.3 0.68 12 0.7
IW-OXA-NH3 12.8 75.9 0.81 18 1.1
a) TOF = nCO · XCO / 100 · nCo b) TOFnom = TOFcat / TOFCo-Ref
5.2.3. CATALYTIC EVALUATION OF IMPREGNATED CATALYSTS SUPPORTED ON
CERIA, ZIRCONIA AND TITANIA (RUTILE TYPE)
The reaction performance of the IWC-NIT catalyst was difficult to access because its
conversion did not a achieve a constant level. After 24 h on stream a CO conversion
of 17.3 % was obtained which dropped to 14.2 % after 72 h. After a steady-state pe-
riod of 36 hours with a XCO of 23.7 % the activity finally decreased to 19.5 %. It is
worth mentioning that during the steady-state period a lower methane selectivity of
8.8 % was noticed then in the time with a lower activity. The determined chain growth
probability was to 0.81. On the catalyst IWZ-NIT only a low conversion of 9.4 % was
achieved combined with an undesirable methane formation of 24.2 % over a period
of 121 h. Due to the high SCH4 only a low α-value of 0.68 was reached. The formation
of hydrocarbons >15 was observed in only small amounts. After a short start time of
16 h on IWB-NIT catalyst reached a XCO of 10.5 % which retain during 80 h. A drop in
conversion to 8.5 % was noticed after 100 h on stream. The methane selectivity
amounted to 12.3 % and SC5+ was 79.3 % with a corresponding chain growth prob-
ability of 0.78. For a detailed overview of the obtained product distribution refer to the
appendix.
5 Results and Discussion 93
Summary of catalytic results obtained on impregnated catalysts
The TOFnom values of IWC-NIT and IWZ-Nit are in the same order of magnitude of ≈1 as derived for Co-Ref catalyst (see Tab. 5.15); on IWB-NIT a lower TOFnom of 0.6
was achieved.
The carbon monoxide conversion varied between 9.7 to 23.4 %. Therefore the sup-
ports can be lined up in the following ascending order according to the obtained car-
1 Eniricerche 2 Institut Francais du Petrole 3 Stiftelsen for Industriell og Teknisk Forskning4 Ruhr-Universität Bochum / Institut für Angewandte Chemie Berlin-Adlershof e.V.
5 Results and Discussion 100
5.2.8. DISCUSSION OF CATALYTIC RESULTS FOR THE NEW CATALYSTS
First, the obtained catalytic data of the new catalysts will be compared with the refer-
ence catalyst. Following the influences of cobalt oxide precursor, the Ru promoter,
the preparation conditions and the resulting cobalt dispersion on the catalyst activity
and selectivity is discussed.
Comparison of Catalytic Data of the New Prepared Catalyst with Co-Ref
Tab. 5.22: Overview of carbon monoxide conversion, selectivity towards C5+ fraction,α−value and TOF values of all tested catalyst compared to Co-Ref (average forsteady-state conditions, Treac = 200 °C, GHSV = 1200 h-1, H2: CO: N2 = 12:6:2 bar, ptot
= 20 bar)
Catalyst XCO [%] SC5+ [wt%] α [-] TOF [103s-1] a) TOFnom[-] b)
Co-Ref 14.7 80.0 0.83 17 1.0
SPR-CoTiO3 2.1 60.3 0.65 2 0.1
IWB-NIT 10.5 79.3 0.77 11 0.6
IW-OXA 5.2 58.3 0.68 12 0.7
IW-ACAC2 7.3 75.3 0.79 13 0.7
IWZ-NIT 9.7 61.0 0.68 19 1.1
IW-OXA-NH3 12.8 75.9 0.81 18 1.1
IWC-NIT 23.7 80.3 0.81 19 1.1
IW-NIT-Step 16.8 83.2 0.90 23 1.3
IW-NIT-AC 19.7 77.2 0.83 21 1.3
PR-EDTA 14.4 81.0 0.83 26 1.5
PR-EDTA-Ru 15.8 81.4 0.84 27 1.5
IW-ACE 26.4 66.9 0.74 32 1.8
IW-ACAC3 23.6 67.9 0.71 43 2.5
IW-ACAC3-Ru 29.3 80.4 0.80 42 2.5
SPR-OXA-Ru 31.3 81.8 0.83 50 2.9
SPR-OXA 32.3 83.0 0.81 60 3.5
a) TOF = nCO · XCO / 100 · nCo b) TOFnom = TOFcat / TOFCo-Ref
5 Results and Discussion 101
In Tab. 5.22 the catalytic results obtained on the new prepared catalysts in compari-
son to the reference catalyst is given. The catalyst prepared by spreading of cobalt
oxalate over titania (SPR-OXA) achieved the highest improvement of turn-over-fre-
quency that was 3.5 times higher than on Co-Ref catalyst. The addition of ruthenium
to the catalyst SPR-OXA led to a slight decrease in XCO; but on this catalyst an α-
value of 0.83 was observed which is in the same order of magnitude as for Co-Ref
catalyst.
The catalysts IW-ACAC3 and IW-ACAC3-Ru can also count to the better catalysts.
since the TOF was improved by a factor of 2.5 compared to the reference catalyst.
Furthermore, a slight increase in activity was obtained on PR-EDTA and PR-EDTA-
Ru; a TOFnom of 1.5 was estimated.
All other catalyst achieved a TOF value nearly in the same order of magnitude as on
Co-Ref; i.e. IW-NIT-AC, IW-NIT-Step, IWZ-NIT, IWC-NIT and IW-OXA-NH3, or even
worse like IW-OXA, IWB-NIT and SPR-CoTiO3.
Comparison of Catalytic Data with Literature
Since cobalt FT catalysts supported on titania was not intensively studied up to now
only a few catalytic data obtained by IGLESIA et al. [104] and REUL and BARTHOLOMEW
[114] can be cited (see Tab. 5.23). In addition some data obtained for catalysts sup-
ported on silica and alumina is given.
Tab. 5.23: Comparison of literature data with Co-Ref and SPR-OXA catalyst
This effect can be attributed to a change in the power of a the SMSI effect on the dif-
ferent sup port materials. The SMSI effect can be described as follows: electrons
from the cobalt crystallites were withdrawn by the support leading to a more metallic
performance of the cobalt species; This state is known to have a high specific activity
and selectivity for high-molecular-weight hydrocarbons in carbon monoxide hydro-
genation as previous reported by VANNICE [204]. BARTHOLOMEV et al. came to a com-
parable conclusion on nickel catalysts supported on alumina, silica and titania [190].
The pure rutile support showed a disadvantageous effect on the catalyst activity. This
result is in good agreement with a catalytic test carried out on a catalyst with an
amount of 16 % of rutile. On this catalyst a lower carbon monoxide conversion was
observed [199]. From this result the conclusion can be drawn that an optimum com-
position of both the phases, rutile and anatase, is necessary in order to get a catalyst
which reached a high CO conversion.
Dispersion Influence
In Fig. 5.29 the carbon monoxide conversion of all new catalysts is plotted versus the
cobalt dispersion (DCored). It is obvious that a dependence of carbon monoxide con-
version on cobalt dispersion exists because an increase in XCO goes along with a
raise in DCored. Since DCored expressed the amount of accessible cobalt an increas-
ing number of active sites can be ascribed to an increasing X(CO). The obtained in-
fluence of cobalt dispersion on XCO was also reported by REUEL and BARTHOLOMEV
[114]. However, there are some exceptions, namely IW-ACAC3 and IW-NIT-AC; on
this catalysts a Dcored of 9.8 % and 7.7 % and according values of XCO to 23.6 % and
19.7 % were obtained; in comparison to catalysts with a similar cobalt dispersion like
IWC-NIT (DCored = 7.7 %; XCO = 23.7 %) the obtained XCO is significant lower.
5 Results and Discussion 105
0 2 4 6 8 10 120
10
20
30
40X
CO
/
%
DCo
red / %
Fig. 5.29: Dependence of carbon monoxide conversion (XCO) on cobalt dispersion(DCored). ( = examined catalysts)
Catalysts prepared by Plasma- induced technique
The improvement in activity of the PL-PP in comparison to IW-ACAC3 catalyst is
minimal and goes along with a higher selectivity towards the undesirable methane
and C2-C4 hydrocarbon fraction (see Tab. 5.20).
A unique low SCH4 of 1.3 achieved on the PL-AT catalyst can not be explained up to
now and such a low amount of methane was not reported in literature so far. In order
to ensure the obtained data the catalyst was prepared again and the results of the
reproduced samples PP-REP1 and PP-REP2 is given in Tab. 5.24 [205]. First, on the
latter samples a higher amount of cobalt (8.1 wt% and 12.0 wt%) and a lower cobalt
dispersion was noticed (0.3 % and 0.7 %) in comparison to PL-PP (Co = 5.5 wt% and
DCo = 1.8 %). Beside the deviation within the physico-chemical properties the cata-
lytic data differ also. The carbon monoxide conversion was for both catalysts about
30 % with an according methane selectivity of approx. 6 %. It became clear that the
results obtained on PL-PP were not reached in any point nor are the results obtained
on PL-REP1 and PL-REP2 even close. One reason might be, that for the reproduc-
tion preparation uncalcined titania (Degussa P25) was applied. However, the calci-
nation of titania is very important because during the pretreatment the desired ratio of
anatase/rutile (30/70) was set. Further, the variation of the total cobalt content as well
as dispersion is responsible for the lower activity; the latter was discussed above. It
5 Results and Discussion 106
seems that the catalysts prepared by plasma- induced techniques are hard to re-
produce; therefore, further studies on this technique should be carry out until a recipe
was found which lead to “same” catalyst all the time because that is essential for in-
dustrial application.
Tab. 5.24: Overview of cobalt content, cobalt dispersion, XCO, α and selectivities to-wards hydrocarbons for the catalyst PL-PP compared to the catalysts PP-REP1 andPP-REP2 (Treac = 238 °C, ptot = 20 bar, H2:CO:N2 = 12:6:2, GHSV = 1500 h-1)