Preparation and Characterization of Cu/ZnO
Catalysts for Methanol Synthesis
vorgelegt von
Diplom-Chemiker
Stefan Zander (geb. Kißner)
aus Berlin
Von der Fakultät II - Mathematik und Naturwissenschaften
der Technischen Universität Berlin
zur Erlangung des akademischen Grades
Doktor der Naturwissenschaften
Dr. rer. nat.
genehmigte Dissertation
Promotionsausschuss:
Vorsitzender: Prof. Dr. Reinhard Schomäcker
Berichter/Gutachter: Prof. Dr. Robert Schlögl
Berichter/Gutachter: Prof. Dr. Thorsten Ressler
Berichter/Gutachter: Prof. Dr. Martin Muhler
Tag der wissenschaftlichen Aussprache: 07.08.2012
Berlin, 2013
D 83
Abstract
i
Abstract
In this work, systematic investigations of the preparation of Cu/ZnO-based methanol synthesis
catalysts are presented. The catalyst precursors were prepared by co-precipitation, followed by
aging, filtrating, washing and drying according to the proven, but not completely understood
industrial preparation method. Subsequent calcination and reduction led to the catalytically
active Cu/ZnO/X catalyst, X being a refractory oxide acting as structural promoter. The
investigations focus on the chemistry of the zincian malachite precursor that was identified as
the material yielding the best catalytic performance with the aim of identifying the role of the
different synthesis parameters on its formation mechanism and properties and of establishing
structure-performance-relationships that explain the role of the synthesis conditions and the
structural promoter phase X on the final catalytic activity.
Co-precipitation (Cu:Zn = 70:30) was performed in a pH- and temperature-controlled (338 K)
manner and enabled homogeneous distribution of the metal ions in the amorphous initial
precipitate which transformed into crystalline zincian malachite during aging. This aging step
was found to be critical with regard to the incorporation of Zn into zincian malachite and was
investigated by in-situ methods. Therefore, it had to be decoupled from the prior co-
precipitation step using co-precipitation with continuous spray-drying. As a function of aging
pH, two different aging mechanisms were found that explain the effect of the synthesis
conditions in the early stages of preparation on the structural properties of the precursor and
later the resulting catalyst. Low pH-values (5.0-6.5) trigger a direct co-condensation
mechanism, while at high pH values (7.0-8.0) a transient sodium zinc carbonate phase was
observed upon crystallization of the zincian malachite precursor phase. The Zn incorporation
into the zincian malachite precursor phase was higher at low pH values. Temperature was found
to accelerate both pathways at a given pH value. Based on these results, the setting of the
synthesis parameters in the applied catalyst preparation method can be rationalized. They have
been optimized to yield maximal Cu,Zn substitution in zincian malachite which in turn is a
precondition for final nanostructuring of the catalyst.
Also in conventional batch synthesis of Cu/ZnO catalysts the application of different pH-values
in the range of pH 6.0-9.0 during co-precipitation was observed to influence the precursor
chemistry. Application of pH values ≥ 6.5 led to higher phase fraction of zincian malachite at
the expense of the undesired Zn-rich by-phase aurichalcite. As a consequence, more Zn was
inserted into zincian malachite after aging, leading to smaller CuO domain size in the calcined
catalyst. For pH-values in the basic regime, formation of two clearly different substituted
Abstract
ii
zincian malachite phases was found indicating inhomogeneous Zn distribution in the precursor
material. The pH-dependent switch of the aging mechanism observed during the previously
described in-situ experiments is a likely explanation for the differences in Zn incorporation.
However, the highest Cu surface area, which is a prerequisite for an efficient catalyst, was
obtained for catalysts prepared at pH 8.5. Unfortunately, we were not able to track back this
observation directly to the synthesis pH in a simple synthesis parameter–structure–performance
relationship. The batch process is probably more complex as variation of the parameter pH may
induce numerous changes in the precursor material that can lead to different and partially
compensating effects for the resulting catalyst.
ZnO is known to act as a spacer for the single Cu particles in the Cu/ZnO catalyst and to enable
the widely studied Cu-ZnO synergy which beneficially affects the activity. MgO was
investigated to act as a substitute for ZnO following the substituted malachite preparation
approach. At the same Cu content
(80 mol%)
, the geometric influence turned out to be even
better compared to ZnO but the synergetic effect of Cu and ZnO during methanol synthesis
from CO
2
/CO/H
2
was lacking. By subsequent impregnation with ZnO both geometric and
synergetic effects were combined in a Cu/MgO/ZnO catalyst which exhibited a higher activity
than Cu/ZnO and Cu/MgO. Thus, the geometric and synergetic effects of the oxide components
have been separated during synthesis. Interestingly, if the feed gas was changed to CO/H
2
,
Cu/MgO was by far most active.
The effect of Ga
2
O
3
as a promoter in the Cu/ZnO/(Ga
2
O
3
) system was investigated by preparing
a sample series with increasing Ga concentration. Ga contents up to 3 mol% were incorporated
in the zincian malachite precursor despite the charge mismatch and changed the characteristics
of the sample dramatically. After calcination, some of the Ga was incorporated in the ZnO.
After reduction, the Cu surface area was increased by 100% and the methanol synthesis activity
by 60% compared to the binary Cu/ZnO reference system. Higher Ga contents led to
segregation and inhomogeneous microstructure of the resulting catalyst. The functionality of Ga
promotion was found to critically depend on the homogeneous distribution of Ga. The best
distribution was achieved by incorporation into the zincian malachite precursor phase and a
linear correlation of the (Zn,Ga) content in this phase with the catalytic activity of the final
catalyst was observed.
Zusammenfassung
iii
Zusammenfassung
Systematische Untersuchungen der Präparation von Cu/ZnO-basierten Methanolsynthese-
Katalysatoren sind Gegenstand dieser Arbeit. Die Katalysatorpräkursoren wurden gemäß dem
etablierten aber nur unzureichend verstandenen industriellen Syntheseweg hergestellt, der aus
den Schritten Co-Fällung, Altern, Filtern, Waschen und Trocknen besteht. Calcinierung und
Reduktion führen schließlich zum aktiven Cu/ZnO/X Katalysator, wobei X ein
temperaturbeständiges Oxid darstellt, welches als struktureller Promotor fungiert. Die
Untersuchungen richten sich auf die Chemie des Zink-Malachit-Präkursors, der letztendlich zu
einer hohen katalytischen Aktivität führt. Dabei sollen die Einflüsse verschiedener
Syntheseparameter auf die Bildung und Eigenschaften des Zink-Malachits und Mikrostruktur-
Aktivitäts-Korrelationen untersucht werden, um die katalytische Aktivität mittels
Syntheseparameter und Promotorphase X erklären zu können.
Durch pH- und temperaturkontrollierte (338 K) Co-Fällung (Cu:Zn 70:30) wird eine homogene
Verteilung der Metallionen im anfänglich amorphen Fällungsprodukt erreicht, welches durch
Altern zu Zink-Malachit kristallisiert. Die Alterung wird als entscheidender Schritt für die
Einlagerung von Zink-Ionen im Zink-Malachit angesehen und wurde mit Hilfe von in-situ
Methoden untersucht. Dafür war eine Entkopplung von der vorausgehenden Co-Fällung nötig,
was durch Co-Fällung und kontinuierliche Sprühtrocknung realisiert wurde. Zwei verschiedene
Alterungsmechanismen wurden, abhängig vom pH-Wert, beobachtet, welche den Einfluss der
Syntheseparameter in den frühen Präparationsschritten auf strukturelle Eigenschaften des
Präkursors und letztlich des Katalysators erklären können. Kleine pH-Werte (5.0-6.5) führen zu
direkter Co-Kondensation, während für höhere pH-Werte (7.0-8.0) eine vorübergehende
Natrium-Zink-Carbonat-Phase beobachtet wurde, bevor die Kristallisation der Zink-Malachit-
Phase einsetzte. Bei kleinen pH-Werten konnte eine erhöhte Substitution von Cu-Ionen durch
Zn-Ionen im Zink-Malachit erreicht werden. Erhöhung der Temperatur führte bei gegebenem
pH-Wert zu einer Beschleunigung beider Mechanismen. Basierend auf diesen Erkenntnissen
kann die Einstellung der Parameter bei dem vorliegenden Präparationsprozess vorgenommen
werden. Ziel dabei ist eine möglichst große Einlagerung von Zink im Zink-Malachit, was
wiederum eine Vorbedingung für die spätere Nano-Strukturierung ist.
Der pH-Wert spielt auch während der Co-Fällung eine entscheidende Rolle für die
Eigenschaften des Cu,Zn-Präkursors (70:30) und wurde in zwei Serien von Batch-Synthesen im
Bereich von 6.0-9.0 variiert. Wurden pH-Werte ≥ 6.5 verwendet, konnte nach dem Altern ein
hoher Phasenanteil von Zink-Malachit neben der unerwünschten Zink-reichen Nebenphase
Zusammenfassung
iv
Aurichalcit erzielt werden. Demzufolge konnte mehr Zink in Zink-Malachit eingebaut werden,
was zu einer kleineren Kristallitgröße von CuO nach der Calcinierung führte. Für pH-Werte im
basischen Bereich (≥ 7.5) wurden zwei deutlich verschieden substituierte Zink-Malachit-Phasen
gebildet, die Zink-Verteilung im Präkursor war demnach inhomogen. Eine Erklärung für den
Unterschied der Zink-Substitution liefert möglicherweise das Ergebnis des in-situ Alterungs-
Experiments, wonach abhängig vom pH-Wert zwei verschiedene Mechanismen ablaufen
können. Eine Voraussetzung für effektive Katalysatoren ist letztendlich die Cu-Oberfläche, die
für bei pH 8.5 präparierte Proben am größten war. Leider konnte dieses Ergebnis nicht mit Hilfe
einer Parameter-Struktur-Aktivitäts-Korrelation beschrieben werden. Der gesamte Batch-
Prozess scheint daher sehr komplex zu sein, da die Variation des Parameters pH-Wert weitere
Veränderungen im Präkursor hervorrufen kann, die zu verschiedenen und möglicherweise
kompensierenden Effekten für den resultierenden Katalysator führen.
ZnO fungiert als Abstandshalter für einzelne Cu-Partikel in Cu/ZnO Katalysatoren und
ermöglicht außerdem die Cu-ZnO-Synergie, welche die Aktivität positiv beeinflusst. Die
Verwendung von MgO anstelle von ZnO wurde untersucht, wobei die Präparation der Cu/MgO-
Katalysatoren analog zu der von Cu/ZnO erfolgte. Bei gleichem Cu-Gehalt (80 mol%) wurde
für MgO ein besserer geometrischer Einfluss festgestellt, jedoch war kein vergleichbarer Effekt
der Synergie während der Methanolsynthese aus CO
2
/CO/H
2
messbar. Nachfolgende
Imprägnierung mit ZnO führte zu einer Kombination der geometrischen und synergetischen
Effekte in Gestalt eines Cu/MgO/ZnO Katalysators, der aktiver war als Cu/ZnO und Cu/MgO.
Daher können geometrischer und synergetischer Effekt der Oxidkomponenten sequentiell
während der Synthese eingeführt werden. Interessanterweise war Cu/MgO bei der
Methanolsynthese aus CO/H
2
mit Abstand am aktivsten.
Der Effekt von Ga
2
O
3
als Promotor im Cu/ZnO/(Ga
2
O
3
)-System wurde durch eine Probenserie
mit ansteigender Ga-Konzentration untersucht. Bis zu 3 mol% Ga
3+
konnten trotz der
abweichenden Ladung in den Zink-Malachit-Präkursor eingebaut werden und führten zu einer
drastischen Änderung der Probeneigenschaften. Nach der Calcinierung konnte ein Teil des Ga
im ZnO nachgewiesen werden. In den reduzierten Proben wurde, verglichen mit der Cu/ZnO-
Referenzprobe, die Cu-Oberfläche um 100% und die Aktivität in der Methanolsynthese um 60%
erhöht. Höhere Ga-Gehalte führten zu Segregation und inhomogener Mikrostruktur des
resultierenden Katalysators. Die beste Elementverteilung wurde erzielt, wenn Ga vollständig im
Zink-Malachit-Präkursor eingebaut war. Es wurde eine lineare Korrelation zwischen dem
(Zn,Ga)-Gehalt in Zink-Malachit und der katalytischen Aktivität gefunden.
Danksagung
v
Danksagung
Bedanken möchte ich mich besonders bei Herrn Prof. Dr. Robert Schlögl für die Möglichkeit, die
vorliegende Arbeit in der Abteilung Anorganische Chemie des Fritz-Haber-Instituts der Max-Planck-
Gesellschaft anfertigen zu können. Dabei bedanke ich mich für die interessante wissenschaftliche
Fragestellung, die wertvollen Anregungen sowie für das sehr gute Arbeitsklima und die
Arbeitsbedingungen in der Abteilung.
Mein Dank gilt ebenfalls meinem Gruppenleiter Dr. Malte Behrens für sein Vertrauen, die ausgezeichnete
fachliche Betreuung, seinen diplomatischen und motivierenden Führungsstil und seine stete
Diskussionsbereitschaft zu allen Aspekten der Arbeit.
Ich danke allen, auch den nicht namentlich genannten Mitarbeitern, die auf ihre Art und Weise zum
Gelingen beigetragen haben, sei es durch praktische Hilfe oder fachliche Diskussionen. Der
“Nanostructure”-Gruppe danke ich für die wissenschaftliche Unterstützung. Meiner ehemaligen
Bürokollegin Dr. Antje Ota danke ich besonders für die gemeinsame Zeit mit vielen interessanten
Diskussionen, vor allem abseits der Forschung, Dr. Thomas Cotter, Dr. Lénárd-István Csepei und Patrick
Kast für das lockere Büroklima. Stefanie Kühl danke ich für die praktische Hilfe, vor allem in der ersten
Zeit, und die unzählig vielen guten Tipps. Julia Schumann sei gedankt für die Unterstützung bei der
Katalysatorpräparation, Dr. Edward Kunkes und Nygil Thomas für die katalytischen Messungen. Ich
danke Dr. Frank Girgsdies, der immer ausführlich und klar auf meine Fragen antwortete, Dr. Igor
Kasatkin für TEM-Messungen und die Diskussionen, Dr. Olaf Timpe für seine stete spontane
Bereitschaft, mir bei vielen kleineren Problemen zu helfen. Ich danke den Personen am FHI für die
Messung meiner Proben: Edith Kitzelmann (XRD und TG), Dr. Andrey Tarasov (TG), Gisela Lorenz und
Maike Hashagen (BET), Gisela Weinberg (SEM), Jutta Kröhnert (IR), Genka Tzolova-Müller (UV-Vis),
Doreen Steffen (Präparation), Achim Klein-Hoffmann (XRF) und Dr. Manfred E. Schuster (TEM).
Ich danke den Kooperationspartnern, Prof. Dr. Thorsten Ressler und Gregor Koch, TU Berlin, für
XANES-Messungen, Dr. Wolfgang Bensch, Beatrix Seidlhofer, Elena Antonova und Jing Wang,
Universität Kiel, für EDXRD-Messungen und dem HASYLAB (DESY), Hamburg, für die Bereitstellung
von Strahlzeit.
Mein Dank gilt dem BMBF und dem STMWFK für die finanzielle Unterstützung im Rahmen der
jeweiligen Projekte (01RI0529, NW-0810-0002), allen beteiligten Personen sowie den Mitarbeitern der
Clariant Produkte (Deutschland) GmbH, Dr. Patrick Kurr, Dr. Benjamin Kniep und Dr. Nikolas Jacobsen.
Meiner Frau Meike und meinem Sohn Mateo danke ich für die Liebe und Kraft, die sie mir während
dieser Zeit gegeben haben. Meinen Eltern, meinen Schwestern und meiner gesamten Familie danke ich
für die große Unterstützung und Geduld während der Arbeit.
Table of Contents
vii
Table of Contents
Abstract i
Zusammenfassung iii
Danksagung v
List of Figures x
List of Tables xvii
List of Abbreviations xviii
Chapter 1:
Introduction and Overview 1
1.1 Catalysis 1
1.2 Methanol synthesis over Cu/ZnO/Al
2
O
3
catalysts 2
1.2.1 Methanol 2
1.2.2 Methanol synthesis 3
1.2.3 Preparation and Characteristics of Cu,Zn based catalysts 4
1.3 Outline of the work 9
1.4 References 10
Chapter 2:
In-situ EDXRD Study of the Chemistry of Aging of Co-
precipitated Mixed Cu,Zn Hydroxycarbonates – Consequences for the
Preparation of Cu/ZnO Catalysts 13
2.1 Introduction 14
2.2 Experimental 17
2.2.1 Precursor Preparation 17
2.2.2 In-situ EDXRD and UV-Vis Spectroscopy during Simulated Aging 17
2.2.3 Ex-situ Characterization 18
2.3 Results and Discussion 20
2.3.1 General 20
2.3.2 Phase Evolution 20
2.3.3 The Effect of Temperature 27
2.3.4 The Effect of Acidity 30
2.3.5 The Effect of Potassium Counter Ions 31
Table of Contents
viii
2.4 Conclusion 33
2.5 References 35
Chapter 3:
Correlations between Preparation and Microstructure of
Cu/ZnO Catalysts for Methanol Synthesis – Influence of the pH value
during Synthesis of Cu,Zn Hydroxy Carbonates 41
3.1 Introduction 42
3.2 Experimental 43
3.2.1 Sample Preparation 43
3.2.2 Characterization 44
3.3 Results and Discussion 46
3.3.1 Precipitation and aging 46
3.3.2 Precursor and calcined materials 49
3.3.3 Reduction and reduced samples 56
3.4 Conclusions 58
3.5 References 60
Supplementary Information 61
Chapter 4:
Development of Cu-Catalysts for Methanol Synthesis
from CO
2
and CO 67
4.1 Introduction 68
4.2 Experimental 70
4.2.1 Catalyst Preparation 70
4.2.2 Characterization 71
4.2.3 Catalytic performance 71
4.3 Results and discussion 72
4.4 Conclusion 78
4.5 References 79
Supplementary Information 80
Chapter 5:
Promoting Methanol Synthesis Catalysts: Correlations
between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
81
Table of Contents
ix
5.1 Introduction 82
5.2 Experimental 83
5.2.1 Sample Preparation 83
5.2.2 Sample Labeling 84
5.2.3 Elemental Analysis 84
5.2.4 Characterization 84
5.2.5 Catalytic testing 86
5.3 Results and Discussion 87
5.3.1 The influence of gallia on the precursor chemistry 87
5.3.1.1 XRD analysis 87
5.3.1.2 Scanning electron microscopy 92
5.3.1.3 Thermal analysis 94
5.3.2 Calcined samples 97
5.3.2.1 XRD analysis 97
5.3.2.2 The influence of gallia on ZnO 98
5.3.2.3 Temperature programmed reduction 101
5.3.3 Activated samples 102
5.3.3.1 Transmission electron microscopy 102
5.3.3.2 Cu surface areas 104
5.3.4 Methanol synthesis activity 106
5.4 Conclusions 107
5.5 References 109
Chapter 6:
Final Summary and Conclusion 111
Appendix xxi
Curriculum vitae xxi
Publications xxii
Poster presentations xxiv
List of Figures
x
List of Figures
Figure 1-1: Distribution of global methanol consumption in 2010
[3]
2
Figure 1-2: Unit cell of zincian malachite according to Behrens
[37]
. For clarification, only the
Jahn–Teller elongated bonds of the CuO
6
units are shown. They are oriented either
perpendicular to (201
) (dark grey) or to (211
) (light grey) and are contracted upon
Cu/Zn substitution (see direction of arrows). 5
Figure 1-3: Graphic according to Behrens
[33]
showing d-spacing of the 201
netplanes of the
zincian malachite precursor phase (
), Cu surface area (
) and normalized
methanol production rates (
) of the final binary model catalyst as a function of
nominal Cu content. The presence of the phases aurichalcite (A), zincian malachite
(zM) and malachite (M) is indicated. 6
Figure 1-4: Proposed reaction scheme for precipitation, aging and subsequent stages in the
preparation of 2:1 Cu/Zn catalysts. Graphic according to
[9]
; redrawn from
[40]
. 7
Figure 2-1: (a) Cartoon of the preparation of Cu/ZnO catalysts comprising precipitation of
zincian georgeite, aging to form zincian malachite (meso-structuring),
decomposition in CuO/ZnO aggregates (nano-structuring) and activation by
reduction to Cu/ZnO. (b) pH evolution during precipitation and aging of a typical
binary sample with change in sample color and crystallinity during aging (insets)
The marked reflections in the XRD pattern refer to the aurichalcite by-phase,
(Cu,Zn)
5
(CO
3
)
2
(OH)
6
. All other reflections are due to zincian malachite
(Cu,Zn)
2
(CO
3
)(OH)
2
. 15
Figure 2-2: Rietveld refinement of the ex-situ XRD pattern of the sample aged in-situ at pH 6.5
and 333 K (ID 3) for quantitative analysis of the phase composition. Experimental
data (circles), background (dotted), background peak (dashed, due to the grease
used as sticking agent to keep the sample in place on the sample holder),
calculated pattern zincian malachite (green), calculated pattern aurichalcite
(orange), total calculated curve (line) and difference curve (grey, offset -100). This
plot is representative for the other aged samples. In this case the ratio of zincian
malachite to aurichalcite was calculated to be 92% to 8%. 22
Figure 2-3: EDXRD patterns (converted to °2θ values of Cu K
α
radiation) during aging of the
amorphous precursor at pH 7 and 323 K (ID 7) after two (a), 26 (b) and 98 min
(c). At the bottom PDF 72-75 (green bars) and PDF 1-457 (red bars) are shown as
references for zincian malachite (Cu,Zn)
2
(CO
3
)(OH)
2
and for sodium zinc
List of Figures
xi
carbonate Na
2
Zn
3
(CO
3
)
4
·3 H
2
O, respectively. The in-situ EDXRD spectra are
representative for all conducted experiments. The position of the 201
peak of
zincian malachite is shifted compared to the pure malachite reference because of
zinc incorporation (see text). 23
Figure 2-4: Integral intensity of selected EDXRD peaks vs. aging time (a); in Na
2
CO
3
; pH = 7;
T = 323 K (ID 7). Zincian malachite, (Cu,Zn)
2
(OH)
2
(CO
3
), is represented by the
201
peak (green), sodium zinc carbonate Na
2
Zn
3
(CO
3
)
4
·3 H
2
O by the 222 peak
(red). Corresponding d-spacing of the 201
peak of zincian malachite (b). 23
Figure 2-5: UV-Vis results for simulated aging in Na
2
CO
3
at pH 7; 323 K (ID 7). a) Measured
diffuse reflectance before and after aging. b) Difference of normalized spectra
relative to the initial spectrum at t = 2 min. c) Wavelength of the maximum
intensity in the UV-Vis spectra in the range of 425 to 900 nm as a function of
time. 25
Figure 2-6: Measured features vs. aging time for samples noted in the plot (cf. Table 2-1). Top
row a-d: Integral intensity of selected EDXRD peaks of zincian malachite 201
,
(green) and sodium zinc carbonate 222, (red). Bottom row e-h: Evolution of d201
values of zincian malachite. 26
Figure 2-7: Calibration of the d201
values versus the Zn content in zincian malachite. The
three solid data points stem from reference samples described in ref.
[18, 21]
and
were used for linear extrapolation. The open data points positioned onto the
extrapolated line refer to the values observed in this study (from ex-situ XRD).
The resulting Zn-contents are given in Table 2-1. 27
Figure 2-8: d201
values of zincian malachite (from ex-situ XRD) depending on the pH value of
simulated aging. Two different groups can be observed: The zincian malachite
samples crystallized without intermediate formation of sodium zinc carbonate (pH
5-6.5) show low d201
values indicating high Zn content. Crystallizations via the
intermediate (pH 7-8) led to high d201
values or low Zn content, respectively.
Lines are guides for the eye. 30
Figure 2-9: Radar plots illustrating the influence of the aging parameters pH value (a, at T =
333 K) and temperature (b, at pH = 7) on the aging reaction and the properties of
the resulting zincian malachite material. Variation of pH leads to different aging
mechanisms below pH 6.5 (a, broken lines) and above pH 7 (a, full lines) and
affects the Zn content of the catalyst precursor. Temperature leads to a gradual
change of the aging kinetics (b, full lines). Substitution of Na
+
by K
+
in the aging
solution at the same temperature (a, dotted lines) has a pronounced effect on the
List of Figures
xii
aging reaction and the phase composition, but not on the Zn content of the zincian
malachite precursor. As a function of temperature, a variation of the Zn content is
observed with Na
+
and K
+
. 33
Figure S2-1: Evolution of pH (red curve) with added Cu,Zn solution (green curve) and Na
2
CO
3
solution (blue curve) during co-precipitation of the Cu,Zn (70:30) precursor in the
continuous process. The slurry was continuously fed into a spray-dryer. 37
Figure S2-2: Schematic representation for the T- and pH-controlled precursor preparation from
aqueous solutions using an automated laboratory-reactor. The co-precipitation and
aging stages (right hand side) were decoupled by continuously removing the
“unaged” precursor (left hand side) and subsequent aging studies at varying
conditions with the same starting material (red arrow). The marked reflections in
the lower right hand corner XRD pattern refer to the aurichalcite by-phase,
(Cu,Zn)
5
(CO
3
)
2
(OH)
6
. All other reflection are due to zincian malachite
(Cu,Zn)
2
(OH)
2
CO
3
. 38
Figure S2-3: Detailed experimental setup of in-situ EDXRD reaction cell at the F3 beamline at
HASYLAB, Hamburg, Germany. 39
Figure S2-4: Ex-situ XRD patterns of all recovered sample (for labeling see Table 2-1). The
201
reflection of malachite and the characteristic peaks of aurichalcite are marked.
The labeling refers to the entry number of Table 2-1 in the main article. 39
Figure S2-5: Evolution of the FWHM of the 201
peak of zincian malachite from the EDXRD
spectra during simulated aging at pH 7, 323 K (ID 7). In all aging experiment
where reflections of the sodium zinc intermediate were detected, the FWHM of the
201
peak of zincian malachite did not decrease during disappearance of the
sodium zinc carbonate phase. Thus, the shift of the 201
peak seems not to be an
effect of overlapping peaks from both phases but rather of Zn incorporation into
the zincian malachite phase. 40
Figure S2-6: Integral intensity of selected EDXRD peaks of detected phases vs. aging time in
Na
2
CO
3
at T = 333 K at different pH-values. Zincian malachite
(Cu,Zn)
2
(OH)
2
(CO
3
) is represented by the 201
peak (green), sodium zinc
carbonate Na
2
Zn
3
(CO
3
)
4
·3H
2
O is represented by the 222 peak (red). 40
Figure 3-1: Evolution of pH (black curve), turbidity (grey curve) during pH-controlled (pH 8.5)
dosing of acidic Cu,Zn solution (dotted curve) and basic Na
2
CO
3
solution (dashed
curve) and subsequent free aging of a Cu,Zn (70:30) precursor. 47
Figure 3-2: Results from the recorded data during precipitation and aging in dependence on the
precipitation pH value: time of the occurrence of the pH drop (a), depth of the pH
List of Figures
xiii
drop (b), ratio of dosed moles of Na
2
CO
3
and metals (Cu+Zn) (c). Shown values
are representing average values, endings of the error bars represent the real values
obtained for the two samples for each precipitation pH value. 47
Figure 3-3: XRD patterns of one selected precursor sample for each preparation pH value. The
well-resolved peaks of aurichalcite are marked with a star. All other reflections
can be assigned to zincian malachite or to an overlap of peaks from both phases. 49
Figure 3-4: Rietveld refinement for the XRD pattern of a precursor sample prepared at pH 7.5:
With one zincian malachite phase (a); with two zincian malachite phases (b);
experimental data (black), total calculated curve (red), background (light grey),
difference curve (grey), calculated pattern zincian malachite 1 (orange curve),
calculated pattern zincian malachite 2 (green curve), calculated pattern aurichalcite
(blue curve). The thick marks indicate the positions of the Bragg reflections. 51
Figure 3-5: Results of XRD full pattern refinement and BET surface areas of Cu,Zn hydroxy
carbonate precursors and calcined samples in dependence on the precipitation pH
value. The zincian malachite phase “zM1” and “zM2” refer to the refinement with
two different zincian malachite phases: weight fraction of the aurichalcite phase
(black) and BET surface areas (grey) of the precursors (a); d-spacing of the 201
reflection of zM1 (red) and zM2 (green) (b); difference between the d201
values
of both zincian malachite phases zM1 and zM2 (c); Zn-content (x
Cu
+ x
Zn
= 100%)
in zM1 (red), zM2 (green) and overall weighted average Zn content in both
zincian malachite phases (black) (d); domain size of CuO (black) and BET surface
areas (grey) of the calcined samples (e). Shown values are representing average
values, endings of the error bars represent the real values obtained for the two
samples for each precipitation pH value. Lines are guides for the eyes. 52
Figure 3-6: Transmission electron microscopy images of reduced Cu,Zn sample prepared at
pH 6.5 (a,b) and pH 9.0 (c,d) showing the typical arrangement of Cu particles (a),
the particle size distribution (b), areas before electron beam sintering (c) and fused
Cu particles after electron beam sintering (d). 56
Figure 3-7: Cu surface areas of reduced Cu,Zn samples with respect to the calcined sample
mass. Shown values are representing average values, endings of the error bars
represent the real values obtained for the two samples for each precipitation pH
value. The dotted line is a guide for the eyes. 58
Figure S3-1: Scanning electron microscopy images (2.5 keV) of a Cu,Zn hydroxy carbonate
precursor sample prepared at pH 9.0 showing an overview image with roundish
List of Figures
xiv
primary particles (a), a primary particle (b), zincian malachite needles, platelets
and particles (c), plane areas (d). 62
Figure S3-2: TG-MS results of hydroxy carbonate precursor sample prepared at pH 7.5: mass
loss (black), MS traces of H
2
O (blue) and CO
2
(green). 64
Figure S3-3: TG-MS results of Cu,Zn hydroxy carbonate precursors: mass loss after heating to
973 K (a), temperature of highest CO
2
emission rate (b) and CO
2
emission above
673 K relative to overall CO
2
emission according to MS trace (c). Shown values
are representing average values, endings of the error bars represent the real values
obtained for the two samples for each precipitation pH value. Lines are guides for
the eyes. 64
Figure S3-4: TPR profile of a calcined Cu,Zn sample prepared at pH 6.5. 65
Figure 4-1: a) Schematic representations of the necessary ingredients for a high performance
methanol synthesis catalyst. b) Scheme of the role of precursor composition for the
Cu dispersion in the final catalyst. 69
Figure 4-2: a) XRD patterns of the precursor materials of CZ (green) and CM (red). The pattern
of malachite (black; ICSD: 72-75) is shown as bar graph. b) XRD patterns of the
calcined samples CZ (green), CM (red) and CMZ (blue). CuO (black; ICSD: 80-
76) is included as reference. 72
Figure 4-3: Scanning electron microscopy images (2.5 keV) of CZ (a) and CM (b) 72
Figure 4-4: (HR-)TEM images of the reduced CZ catalyst. The insets show power spectra of
the neighbouring particles and are used for phase identification. 73
Figure 4-5: (HR-)TEM images of the reduced CM catalyst. 74
Figure 4-6: Catalytic results of methanol synthesis of the CZ, CM and CMZ catalysts in
different feed gas compositions at 30 bar and 503 K. 75
Figure 4-7: (HR-)TEM images of the reduced CMZ catalyst. 76
Figure 5-1: Selected XRD patterns of Cu,Zn,Ga hydroxy carbonate precursors with different
Ga contents. Bars of the references malachite (green), aurichalcite (red) and
hydrotalcite-like phase (blue) are included. 88
Figure 5-2: Rietveld refinement for the XRD pattern of a Cu,Zn,Ga hydroxy carbonate
precursor sample containing 1.0% Ga, experimental data (black), total calculated
curve (red), background (light grey), difference curve (grey), calculated pattern
zincian malachite (orange curve), calculated pattern aurichalcite (blue curve). The
thick marks indicate the positions of the Bragg reflections. The fit quality is
comparable for the fits of all other samples. 89
List of Figures
xv
Figure 5-3: Results of XRD full pattern refinement of Cu,Zn,Ga hydroxy carbonate precursors.
Top: phase fraction of zincian malachite (green), aurichalcite (red) and
hydrotalcite-like phase (blue); Top: domain size of zincian malachite (black) and
d201
value (grey); errors of domain sizes are smaller than used symbols. 90
Figure 5-4: Zincian malachite domain sizes (black) and BET surface areas (red) of Cu,Zn,Ga
hydroxy carbonate precursors in dependence from Zn (Ga) incorporation into
zincian malachite and d201
value, respectively. 91
Figure 5-5: Electron microscopy images (2.5 keV) and EDX-results of the Ga13.0 hydroxy
carbonate precursor showing an overview (a), the same overview in backscattered
electron mode (b), zincian malachite needles and some platelets (c), Ga-rich area
(d) and local elemental distribution from SEM-EDX (e). 93
Figure 5-6: TG-MS results of Ga2.5 hydroxy carbonate precursor: mass loss (black), MS traces
of H
2
O (blue) and CO
2
(green). 95
Figure 5-7: TG-MS results of Cu,Zn,Ga hydroxy carbonate precursors: mass loss after heating
to 973 K (a), temperature of highest CO
2
emission rate (b) and CO
2
emission
above 673 K relative to overall CO
2
emission according to MS trace (c). 96
Figure 5-8: XRD pattern of calcined Ga2.5 sample. 97
Figure 5-9: Determination of absorption edge energies from UV-Vis measurements by the
intercept of a linear fit. 99
Figure 5-10: Ga K-edge XANES of the calcined Ga2.5 sample (black) and results of the linear
combination fit in the range of -20 to 50 eV (related to the Ga-K-edge) using
experimental spectra of Ga oxide reference materials. 100
Figure 5-11: TPR profiles of ZnO (black) and ZnO/3%Ga (blue). 101
Figure 5-12: TPR profiles of Cu,Zn,Ga calcined samples. 102
Figure 5-13: TEM image of the reduced Ga2.5 sample showing the typical arrangement. 103
Figure 5-14: Results from transmission electron microscopy of reduced Ga,Zn,Ga samples.
Ga13.0: elemental composition from TEM-EDX showing a strong tendency of
ZnGa
2
O
4
spinel formation (a) and TEM image of Cu particles covered with
ZnGa
2
O
4
spinel and ZnO (b). Ga2.5: elemental composition from TEM-EDX
showing homogeneous distribution (c) and TEM image of crystalline α-gallia (d).
104
Figure 5-15: Cu surface areas of reduced Cu,Zn,Ga samples with respect to the calcined sample
mass (a) and contained CuO mass (b). The error was estimated to be ± 1 m
2
g
−1
in
the top graph. CuO domain sizes of the calcined samples are given for comparison
(c). Errors for domain sizes were smaller than the used symbols. 105
List of Figures
xvi
Figure 5-16: Methanol productivity of selected Cu/ZnO/Ga
2
O
3
catalysts in methanol synthesis
relative to the unpromoted sample Ga0.0 (blue bars) and content of non-Jahn-
Teller ions (Zn
2+
and Ga
3+
) in the zincian malachite structure (red points)
calculated from d201
values of the binary Cu,Zn system according to
[35]
. 106
Figure 5-17: Intrinsic activities (related to the Cu surface area) of Cu/ZnO/Ga
2
O
3
catalysts in
methanol synthesis relative to the unpromoted sample Ga0.0. 107
List of Tables
xvii
List of Tables
Table 2-1: Summary of aging parameters and aging results. The sample ID 0 refers to the
unaged precursor. 21
Table S2-1: Aging parameters and internal sample numbers. The sample ID 0 refers to the
unaged precursor. 37
Table S3-1: Denotation of samples according to their precipitation pH value. Averaged Zn
content (x
Zn
+ x
Cu
= 100%) of the calcined samples from XRF. Listed values are
representing average values, errors given in brackets represent the real values
obtained for the two samples for each precipitation pH value. 61
Table 4-1: Properties of the CZ, CM and CMZ catalysts (prec. = precursor material, calc. =
calcined material). 73
Table S4-1: Internal sample numbers 80
Table 5-1: Results of precursor sample characterization 94
Table 5-2: Results of calcined sample characterization and Cu surface areas 98
Table 5-3: Results of linear combination fit of the Ga-K-edge XANES of the calcined Ga2.5
sample by Ga-oxide-reference spectra (the lowest R-value corresponds to the best
fit). 100
Table 5-4: Results of reduced sample characterization and activity measurements 103
List of Abbreviations
xviii
List of Abbreviations
BET Adsorption isotherm model of Brunauer, Emmet and Teller
EDX Energy dispersive X-ray spectroscopy
EDXRD Energy dispersive X-ray diffraction
HRTEM High resolution transmission electron microscopy
MS Mass spectrometry
PSD Particle size distribution
RFC Reactive frontal chromatography
SA Surface area
SEM Scanning electron microscopy
TCD Thermal conductivity detector
TEM Transmission electron microscopy
TG Thermo gravimetry
TPR Temperature programmed reduction
UV-Vis Ultraviolet-visible spectroscopy
XANES X-ray absorption near edge spectroscopy
XRD X-ray diffraction
XRF X-ray fluorescence
Chapter 1: Introduction and Overview
1
Chapter 1: Introduction and Overview
1.1 Catalysis
Catalysis is one of the central concepts in chemistry for organic as well as inorganic processes.
Catalysts accelerate certain chemical reactions by decreasing the activation energy of single
elementary reaction steps. As a consequence, a reaction can be steered into a desired direction
with the constraint that only the kinetics can be changed but not the thermodynamics.
Nowadays, around 90% of the chemical processes use heterogeneous catalysts in chemical,
food, pharmaceutical, automobile and petrochemical industries. More modern fields are fuel
cells, green chemistry, nanotechnology and biotechnology. The main advantage of
heterogeneous catalysis is that the catalyst and the reaction products can be easily separated
from each other. Biocatalysts (enzymes) are mainly applied for the production of fine chemicals
when high (chemo-, regio- and stereo-) selectivity is required whereas inorganic catalysts
(metals, metal oxides) are employed for large scales processes. In the latter case, high selectivity
is desirable to save energy and natural resources
[1]
.
The understanding of the reaction mechanism during a heterogeneously catalyzed reaction is
still limited and recently, there is only one process (ammonia synthesis over iron catalysts)
which can be claimed to be almost fully understood. However, the direct observation of the
single elementary steps is not possible to date and the mechanism was derived from surface
studies with iron single crystals
[2]
. Since the catalyzed reaction requires adsorption of the educts
and intermediates and takes place on the surface of the catalyst, (in-situ) surface studies can be
seen as one of the keys to a wider insight into the mode of operation of a catalyst. However, due
to physical limitations, surface sensitive methods can often only be carried out under low
pressures which are far away from in-operando conditions which usually require high pressures.
The difficulty of transferring the obtained results is denoted as pressure gap. Thus,
complementary methods of material science are needed to correlate atomic structure and
macroscopic properties. With the acquired knowledge about these so-called structure function
relationships, the conventional way of trial and error catalyst development can be shifted to a
more rational catalyst design.
Chapter 1: Introduction
and Overview
2
1.2
Methanol synthesis over Cu/ZnO/Al
1.2.1 Methanol
Methanol is one of the most important basic components in the
worldwide production volume was about 45 million tons in 2010, with a rising tendency over
the past years
[3]
. The major amount is used to synthesize formaldehyde (precursor for organic
synthesis), further products are methyl
(precursor for monomers,
conservation) (
Figure 1-1:
Distribution of global methanol consumption in 2010
Furthermore,
a steadily increasing fraction and absolute amount of methanol is applied as
alternative fuel acting as additive or for the production of biodiesel and dimethylether. The
reasons are that methanol has a high energy content of 726.3
combustion properties and is a potential chemical H
than pure hydrogen its
elf and only methane has an equally high H/C ratio. Advantages follow
from the fact, that methanol is liquid at room temperature and therefore offers better transport
and storing properties compared to elemental hydrogen. The freezing point is at around 17
so an application in cold regions can be performed unproblematically. On the other hand
methanol requires resistant materials due to its polar character unlike
combustion of methanol
in engines proceeds almost without harmful byproducts like SO
NO
x
, such as being produced from impurities in gasoline. The application in fuel cells is not
and Overview
Methanol synthesis over Cu/ZnO/Al
2
O
3
catalysts
Methanol is one of the most important basic components in the
chemical industry. The
worldwide production volume was about 45 million tons in 2010, with a rising tendency over
. The major amount is used to synthesize formaldehyde (precursor for organic
synthesis), further products are methyl
-tert-butyl-
ether (antiknock agent) and acetic acid
conservation) (
Figure 1-1).
Distribution of global methanol consumption in 2010
[3]
a steadily increasing fraction and absolute amount of methanol is applied as
alternative fuel acting as additive or for the production of biodiesel and dimethylether. The
reasons are that methanol has a high energy content of 726.3
kJ mol
-1
[
4
combustion properties and is a potential chemical H
2
-
carrier. One liter contains more hydrogen
elf and only methane has an equally high H/C ratio. Advantages follow
from the fact, that methanol is liquid at room temperature and therefore offers better transport
and storing properties compared to elemental hydrogen. The freezing point is at around 17
so an application in cold regions can be performed unproblematically. On the other hand
methanol requires resistant materials due to its polar character unlike
non
polar gasoline
in engines proceeds almost without harmful byproducts like SO
, such as being produced from impurities in gasoline. The application in fuel cells is not
chemical industry. The
worldwide production volume was about 45 million tons in 2010, with a rising tendency over
. The major amount is used to synthesize formaldehyde (precursor for organic
ether (antiknock agent) and acetic acid
a steadily increasing fraction and absolute amount of methanol is applied as
alternative fuel acting as additive or for the production of biodiesel and dimethylether. The
4
]
. It shows good
carrier. One liter contains more hydrogen
elf and only methane has an equally high H/C ratio. Advantages follow
from the fact, that methanol is liquid at room temperature and therefore offers better transport
and storing properties compared to elemental hydrogen. The freezing point is at around 17
6 K,
so an application in cold regions can be performed unproblematically. On the other hand
polar gasoline
[5]
. The
in engines proceeds almost without harmful byproducts like SO
x
or
, such as being produced from impurities in gasoline. The application in fuel cells is not
Chapter 1: Introduction and Overview
3
limited to the conventional hydrogen based type with previous steam reforming but also direct
methanol fuel cells are possible
[6]
. Impurities like sulfur and produced CO from methanol steam
reforming have to be removed to prevent poisoning the electrodes of the fuel cell
[7]
.
In the chemical industry, methanol acts as a starting material to produce gasoline, olefins (for
polymers) und aromates. Hence, it competes with the normally applied educts from coal,
petroleum and natural gas, otherwise it promises a bigger independence from these fossil
sources
[8]
.
1.2.2 Methanol synthesis
Methanol synthesis
[9]
from synthesis gas (H
2
, CO and CO
2
) over solid catalysts was first
reported in 1921 by Patart
[10-11]
. BASF launched the first large industrial methanol plant using
ZnO/Cr
2
O
3
catalysts, temperatures of 573-633 K and pressures of 150-250 bars
[12]
. In the
1960s, changing feedstock from coal to naphtha or natural gas led to less impurities (especially
sulfur) in the synthesis gas and the known Cu/ZnO/Al
2
O
3
catalysts were favored thenceforth
applying somewhat lower temperatures and pressures up to 100 bars
[13]
. Nowadays, modern
plants produce more than 5000 tons every day. Methanol synthesis from synthesis gas is
exothermic (∆H
0
< 0) accompanied by decreasing entropy and therefore favored to proceed at
low reaction temperatures and high reaction pressures. Thermodynamically, methanol is one of
the least likely products compared to methane or higher alcohols. According to ref.
[14]
, the
following reactions are involved:
2
∆
H
0
= -91 kJ mol
-
1
(Eq. 1.1)
3
∆
H
0
= -49 kJ mol
-
1
(Eq. 1.2)
∆
H
0
= -41 kJ mol
-
1
(Eq. 1.3)
The hydrogenation-step can occur on both, CO (Eq. 1.1) or CO
2
(Eq. 1.2). The water-gas-shift
reaction (Eq. 1.3) can be regarded as the difference of (Eq. 1.1) and (Eq. 1.2) and thus, is not an
independent reaction. The mechanism and the carbon source of industrial methanol synthesis
are still under debate today. In the early 1950's, a mechanism for methanol formation from CO
over ZnO/Cr
2
O
3
catalysts was reported
[15]
. Later, CO was assumed to be the predominant
carbon source over Cu/ZnO catalysts
[16]
. CO
2
was only regarded to reoxidize Cu
0
to the active
Cu
+
state. But at the same time, a Russian group reported methanol to be formed almost
Chapter 1: Introduction and Overview
4
exclusively from CO
2
. These results were obtained over Cu/ZnO/Al
2
O
3
and were based on
kinetic experiments
[17-18]
and radioactively labeled carbon dioxide isotopes
[19-21]
. No methanol
was formed when using a pure CO/H
2
mixture over Cu/ZnO
[22]
or Cu/ZnO/Al
2
O
3
[17]
. Today,
mechanism starting from CO
2
is mostly accepted when regarding industrial conditions.
Other synthesis routes for methanol formation start with CH
4
or pure CO
2
. Both substances are
known as "greenhouse gases", hence their economic conversion is desired but not possible to
date
[23]
. Supported copper nanoparticles were not only widely applied as active catalysts in
methanol synthesis but, depending on feed composition, temperature and pressure, also in
methanol steam reforming and water-gas shift reaction because of the similar elemental
reactions.
With a fundamental understanding of the reaction mechanism and correlations between catalytic
activity and surface / bulk structure improved catalysts can be designed
[24]
. However, only for
Cu single crystals the elemental steps of the methanol synthesis are known
[25]
. The questions
concerning the microkinetics and the active centers remain still unanswered in the case of the
more complex Cu/ZnO/Al
2
O
3
catalytic systems. Reasons are the varying structure of Cu/ZnO
catalysts accompanied by the change of the catalysts surface depending on the ambient
conditions (oxidizing / reducing atmosphere)
[26-28]
.
1.2.3 Preparation and Characteristics of Cu,Zn based catalysts
Preparation of Cu/ZnO/Al
2
O
3
catalytic systems has been optimized in the last 40 years of
industrial application. Different methods like co-precipitation, kneading, impregnation and
leaching have been tested. Currently, most syntheses are carried out as a multi-step synthesis as
follows
[29]
: Mixed metal hydroxy carbonate precursors are formed by controlled co-
precipitation (pH 6.5, T = 338 K) from aqueous Cu,Zn,Al (6:3:1) nitrate solutions and Na
2
CO
3
solution as precipitating agent. Chlorides and sulfates cannot be used because chloride and
sulfur poison the final catalyst. Subsequently, the precipitate is aged in the mother liquor,
filtrated, washed, dried and calcined to give the metal oxides. To obtain the desired activity in
methanol synthesis, CuO has to be reduced to metallic copper. Normally, this takes place
directly in the synthesis gas feed by hydrogen or carbon monoxide.
In order to better understand the catalytic systems, relationships between preparation
parameters, microstructure and activity of Cu/ZnO/Al
2
O
3
are investigated
[30]
. All parameters of
the catalyst preparation influence the bulk and surface structure and therewith the characteristics
Chapter 1: Introduction and Overview
5
and activity of the resulting catalyst. This phenomenon is also called the "chemical memory"
[31]
.
Depending on the nominal metal composition, many different mixed metal hydroxy carbonate
precursor phases can emerge in the course of catalyst preparation. With Cu as the major
fraction, these are notably Cu
2
(OH)
2
CO
3
(malachite) for pure Cu samples, (Cu
1-x
Zn
x
)
2
(OH)
2
CO
3
(zincian malachite) with x < 0.3, (Cu
1-y
Zn
y
)
5
(OH)
6
(CO
3
)
2
(aurichalcite) with y > 0.5, and
(Cu,Zn)
6
Al
2
(OH)
16
CO
3
·4H
2
O (hydrotalcite-like phase), only when a significant amount of Al
3+
is present. The last phase should also be formed with other trivalent ions instead of Al
3+
, such as
Ga
3+
or Cr
3+
. Controversial discussions are present in the literature about the relevant precursor
phase. Increased Cu dispersion, intrinsic activity (activity per Cu surface area) and overall
activity were reported to be a consequence of the predominant presence of the precursors
aurichalcite
[32]
or zincian malachite
[33]
in the Cu,Zn system and rosasite
[34]
in Cu,Zn,Al
systems.
In further XRD studies, a shift of the 201
(and 211
) reflection of zincian malachite has been
observed when varying the Cu:Zn ratio
[35-36]
. This is related to the decrease of the d-spacing
which is the result of incorporation of Zn
2+
into the malachite structure (Figure 1-2).
Figure 1-2: Unit cell of zincian malachite according to Behrens
[37]
. For clarification, only the Jahn–Teller elongated
bonds of the CuO
6
units are shown. They are oriented either perpendicular to (201
) (dark grey) or to (211
) (light
grey) and are contracted upon Cu/Zn substitution (see direction of arrows).
Chapter 1: Introduction and Overview
6
Cu
2+
(3d
9
) is a Jahn-Teller ion, Zn
2+
(3d
10
) is not. With increased Zn
2+
incorporation, the
elongated axial O-Cu-O units (perpendicular to the 201
netplanes) in the CuO
6
octahedra of
pure Cu malachite are substituted by not elongated O-Zn-O units leading to a decrease of the
average Jahn-Teller distortion. The shift is a direct measure for the incorporation of Zn
2+
in the
malachite structure and might predetermine the Cu dispersion and the activity of the resulting
catalyst. The substitution of Cu
2+
by Zn
2+
in zincian malachite was found to be limited to
approximately 28%. Higher Zn
2+
contents would lead to the undesired Zn enriched phase
aurichalcite
[29, 31]
(Figure 1-3).
Figure 1-3: Graphic according to Behrens
[33]
showing d-spacing of the 201
netplanes of the zincian malachite
precursor phase (), Cu surface area () and normalized methanol production rates () of the final binary model
catalyst as a function of nominal Cu content. The presence of the phases aurichalcite (A), zincian malachite (zM) and
malachite (M) is indicated.
During precursor preparation, aging of binary Cu,Zn precipitates was reported to be crucial
[29,
38-39]
and to lead to a loss of the by-phase aurichalcite phase yielding more and higher Zn
substituted zincian malachite (Figure 1-4). As a consequence, small and well distributed Cu and
ZnO crystallites in the active catalyst lead to better performance in methanol synthesis. During
aging the meta-stable product (amorphous precipitate) is transformed in the thermodynamic
product (crystalline precipitate) by stirring in the mother liquor. The aging process has not been
well understood yet and the proceeding reaction steps can hardly be optimized independently,
because they are coupled to the synthesis parameters during co-precipitation.
Chapter 1: Introduction and Overview
7
Figure 1-4: Proposed reaction scheme for precipitation, aging and subsequent stages in the
preparation of 2:1 Cu/Zn catalysts. Graphic according to
[9]
; redrawn from
[40]
.
A precondition for superior performance of Cu/ZnO/(Al
2
O
3
) catalysts in methanol synthesis is
definitely a high Cu surface area due to the possibly increasing number of active sites. Whereas
the activity of supported catalysts with low metal content is often referenced to the metal
loading of the active phase, this does not work for the Cu/ZnO/(Al
2
O
3
) system because of the
high Cu content where only a small fraction of the Cu atoms is accessible to the reaction gas.
Linear correlations between Cu surface area and activity have been reported in literature for
Cu,Zn
[41]
and Cu,Zn,Al system
[30, 42]
. However, deviations from this behavior have been
observed depending on structural defects, i.e. microstrain
[24, 43-46]
. In general, a non-ideal form
of copper
[44, 47]
is required to obtain active sites.
ZnO acts not only as dispersant and stabilizer but was also reported to cause synergetic effects
and to provide the active centers due to the interface contact with the copper phase
[39, 48]
. That
results in a beneficial electronic structure for adsorption of reactants and products. As a
consequence, the activity of Cu/ZnO is several magnitudes higher than that of individual Cu or
ZnO. Different active species have been proposed, e.g. Cu-Zn alloy formed during reduction
[49]
,
dissolved Cu
+
in ZnO
[50]
or electron rich Cu at Schottky-junctions
[51]
. In a recently published
model of the active site of industrial methanol synthesis over Cu/ZnO/Al
2
O
3
, that was partially
Chapter 1: Introduction and Overview
8
based on the work presented in this thesis, the synergetic effect was accounted for by strong
metal support interaction (SMSI)
[52]
, which has been observed in high-performance catalysts by
HRTEM and XPS. Therein, the intrinsic activity of the exposed Cu surface area scaled with the
abundance of stacking faults in Cu nanoparticles. This correlation was rationalized by the
generation of high energy sites at the surface at the positions, where the planar defect
terminates. Also residual oxygen in Cu as a result of incomplete reduction might play a role for
the defect structure of active Cu. SMSI between Cu and ZnO has previously been reported in
literature and studied by Cu surface area determination
[42]
, EXAFS
[27]
and IR spectroscopy of
CO adsorption
[53]
.
For industrial applications of Cu/ZnO during methanol synthesis, mostly Al
2
O
3
is used as a
promoter. Al
2
O
3
inhibits thermal sintering of the particles, prevents poisoning of the active
metal surface and ensures additional chemical and thermal performance stability, which is very
important for industrial catalysts
[54-56]
. Furthermore, addition of Al
2
O
3
leads to higher intrinsic
activities
[42]
. Other promoters like zirconia
[57]
, silica
[58]
, gallia and chromia
[59]
are also able to
beneficially affect Cu dispersion, stability, activity and selectivity of the catalyst. Saito et al.
[59]
reported that addition of metal oxide promoters can have different effects, first the increase of
the Cu dispersion in the case of alumina or zirconia, secondly the improvement of the specific
activity in the case of gallia and chromia. The authors claim that the latter feature is due to the
optimization of the Cu
+
/Cu
0
ratio on the Cu surface under reaction conditions
[60]
. Recently, we
found that an Al content of around 3 mol% in the ternary Cu/ZnO/Al
2
O
3
system leads to an
optimized beneficial promoting effect
[61]
. The obtained precursor during preparation was pure
zinc (aluminum) containing malachite without any aurichalcite or Cu,Zn,Al hydrotalcite. After
calcination, Al was introduced into the ZnO phase at Zn
2+
sites in tetrahedral coordination.
Alike ZnO, the function of the Al
2
O
3
promoter was divided into a geometrical and a synergetic
contribution. The former affects the Cu dispersion and leads to an increase of the Cu surface
area. The latter promotes the intrinsic activity of Cu and was related to the incorporation of Al
into the ZnO lattice and an influence onto the Cu/ZnO synergy.
The lifetime of industrial Cu/ZnO/Al
2
O
3
catalysts lies in the range of years. Deactivation
processes of Cu based catalysts mainly comprise sintering and poisoning. The first is diminished
by the presence of ZnO and Al
2
O
3
and can be regarded as a mechanical spacing effect
[54, 62-64]
.
The latter can be alleviated by ZnO which absorbs sulfur (present as H
2
S) from the feed gas
[54,
64]
.
Chapter 1: Introduction and Overview
9
1.3 Outline of the work
This work contains systematic studies concerning the preparation and characterization of Cu,Zn-
based catalytic systems for methanol synthesis. The binary Cu,Zn system presents a functional
model system of the industrially applied Cu,Zn,Al system, but with less complexity.
Correlations between preparation parameters, microstructure and catalytic activity in methanol
synthesis shall be identified and investigated. Assuming zincian malachite as the relevant
precursor phase
[33]
, the preparation of phase pure samples with homogeneous metal distribution
is targeted to enable unambiguous structure-function-relationships. Especially correlations of
precursor chemistry (e.g.
d
value of zincian malachite) with properties of the final catalyst
(e. g. activity) are of great interest. Due to the "chemical memory", reproducibility of all
preparation steps is an important factor.
One strategy is to perform systematic variation of preparation parameters for a system of fixed
composition (here: Cu:Zn = 70:30). In order to better understand the aging process of the initial
Cu,Zn precipitate during precursor preparation, this step is decoupled from co-precipitation and
investigated independently with the help of in-situ energy dispersive X-ray diffraction.
Application of different aging parameters like pH value and temperature shall reveal the
influence on the process of precursor crystallization. Additionally, the pH value during co-
precipitation is varied. Although studies for the ternary Cu,Zn,Al system are available
[30, 65]
,
data of comprehensive investigation of the less complex binary Cu,Zn system is still lacking.
The second strategy is to work with constant preparation conditions and apply modifications on
the system. Promoting the binary Cu,Zn (70:30) system with small amounts of Al
3+
was recently
reported to lead to phase pure zincian malachite precursors with high Zn incorporation and
subsequent higher activities
[61]
. The incorporation of Al
3+
itself into the zincian malachite
structure is possible but limited due to the charge mismatch. Similar results are expected when
using Ga
3+
as a promoter because of the not too different ionic radii of Al
3+
and Ga
3+
and
analogous modifications of the oxides. The advantage of using Ga
3+
instead of Al
3+
is a better
access to spectroscopic methods (XAS) which can account for elucidation of the promoter
effect. Therefore, Cu,Zn,Ga samples with different Ga
3+
concentrations up to 10 mol% are
prepared.
Although the phenomenon of Cu-ZnO-synergy is controversially discussed in literature, the
beneficial effect on the activity in industrial methanol synthesis is not questioned. We recently
Chapter 1: Introduction and Overview
10
reported that the generation of “methanol copper” is induced by strong metal support
interactions (SMSI) in the presence of ZnO
[52]
. If this image is right, then MgO cannot
adequately replace ZnO, despite a possible enhancement of Cu dispersion. A combination of the
effects of ZnO and MgO on Cu might lead to catalysts exhibiting a large and highly active Cu
surface area.
Characterization data of the samples is presented from different stages of their preparation
(precursors, calcined and reduced samples) with respect to crystalline phases (XRD), surface
area (BET), thermal properties (TG-MS), reduction behavior (TPR), morphology and real
composition (SEM, TEM, XRF) and Cu surface area (N
2
O-RFC). Selected samples are
subjected to further characterization (XAS) and testing in methanol synthesis. The process of
aging is studied by in-situ energy dispersive XRD (University of Kiel, HASYLAB).
1.4 References
[1] J. A. Dumesic, G. W. Huber, M. Boudart, in Handbook of Heterogeneous Catalysis,
Vol. 6 (Eds.: G. Ertl, G. Knözinger, F. Schüth, J. Weitkamp), Wiley-VCH, Weinheim
2nd ed., 2008, pp. 2501-2575.
[2] R. Schlögl, in Handbook of Heterogeneous Catalysis (Eds.: G. Ertl, G. Knözinger, F.
Schüth, J. Weitkamp), Wiley-VCH, Weinheim 2nd ed., 2008.
[3] www.methanolmsa.com, April 2012.
[4] F. D. Rossini, P. Natl. Acad. Sci. USA 1931, 17, 343-347.
[5] G. A. Olah, A. Goeppert, G. K. Surya Prakash, Beyond Oil and Gas: The Methanol
Economy, Wiley-VCH, Weinheim, 2006.
[6] S. Ahmed, M. Krumpelt, Int. J. Hydrogen Energ. 2001, 26, 291-301.
[7] S. Velu, K. Suzuki, T. Osaki, Catal. Lett. 1999, 62, 159-167.
[8] U. Onken, A. Behr, Chemische Prozesskunde, Lehrbuch der Technischen Chemie, Vol.
3, Thieme, Stuttgart, 1996.
[9] J. B. Hansen, P. E. H. Nielsen, in Handbook of Heterogeneous Catalysis, Vol. 6 (Eds.:
G. Ertl, G. Knözinger, F. Schüth, J. Weitkamp), Wiley-VCH, Weinheim 2nd ed., 2008,
pp. 2920-2949.
[10] G. Patart, in French Patent 540, 1921, p. 343.
[11] C. Lormand, Ind. Eng. Chem. 1925, 17, 430-432.
[12] BASF, in German Patents 415 686, 441 433, 462 837, 1923.
[13] ICI, in GB Patents 1 010 871 (1961), 1 159 212 (1969), 1 296 212 (1972).
[14] M. R. Rahimpour, Chem. Eng. Comm. 2007, 194, 1638-1653.
[15] G. Natta, P. Pino, G. Mazzanti, I. Pasquon, Chim. Ind. Milan 1953, 35, 705-724.
[16] K. Klier, V. Chatikavanij, R. G. Herman, G. W. Simmons, J. Catal. 1982, 74, 343-360.
[17] A. Y. Rozovskii, Y. B. Kagan, G. I. Lin, E. V. Slivinskii, S. M. Loktev, L. G. Liberov,
A. N. Bashkirov, Kinet. Catal. 1975, 16, 706-707.
[18] A. Y. Rozovskii, Y. B. Kagan, G. I. Lin, E. V. Slivinskii, S. M. Loktev, L. G. Liberov,
A. N. Bashkirov, Kinet. Catal. 1976, 17, 1132-1138.
[19] Y. B. Kagan, G. I. Lin, A. Y. Rozovskii, S. M. Loktev, E. V. Slivinskii, A. N.
Bashkirov, I. P. Naumov, I. K. Khludenev, S. A. Kudinov, Y. I. Golovkin, Kinet. Catal.
1976, 17, 380-384.
Chapter 1: Introduction and Overview
11
[20] A. Y. Rozovskii, G. I. Lin, L. G. Liberov, E. V. Slivinskii, S. M. Loktev, Y. B. Kagan,
A. N. Bashkirov, Kinet. Catal. 1977, 18, 578-585.
[21] A. Y. Rozovskii, Kinet. Catal. 1980, 21, 78-87.
[22] B. Denise, R. P. A. Sneeden, Appl. Catal. 1986, 28, 235-239.
[23] K.-O. Hinrichsen, J. Strunk, Nachr. Chem. 2006, 54, 1080-1084.
[24] B. L. Kniep, T. Ressler, A. Rabis, F. Girgsdies, M. Baenitz, F. Steglich, R. Schlögl,
Angew. Chem. Int. Edit. 2004, 43, 112-115.
[25] T. S. Askgaard, J. K. Norskov, C. V. Ovesen, P. Stoltze, J. Catal. 1995, 156, 229-242.
[26] N. Y. Topsoe, H. Topsoe, Top. Catal. 1999, 8, 267-270.
[27] J. D. Grunwaldt, A. M. Molenbroek, N. Y. Topsoe, H. Topsoe, B. S. Clausen, J. Catal.
2000, 194, 452-460.
[28] P. L. Hansen, J. B. Wagner, S. Helveg, J. R. Rostrup-Nielsen, B. S. Clausen, H. Topsoe,
Science 2002, 295, 2053-2055.
[29] D. Waller, D. Stirling, F. S. Stone, M. S. Spencer, Faraday Discuss. 1989, 87, 107-120.
[30] C. Baltes, S. Vukojevic, F. Schüth, J. Catal. 2008, 258, 334-344.
[31] B. Bems, M. Schur, A. Dassenoy, H. Junkes, D. Herein, R. Schlögl, Chem. Eur. J.
2003, 9, 2039-2052.
[32] T. Fujitani, J. Nakamura, Catal. Lett. 1998, 56, 119-124.
[33] M. Behrens, J. Catal. 2009, 267, 24-29.
[34] R. H. Höppener, E. B. M. Doesburg, J. J. F. Scholten, Appl. Catal. 1986, 25, 109-119.
[35] M. Behrens, F. Girgsdies, A. Trunschke, R. Schlögl, Eur. J. Inorg. Chem. 2009, 1347-
1357.
[36] P. Porta, S. Derossi, G. Ferraris, M. Lojacono, G. Minelli, G. Moretti, J. Catal. 1988,
109, 367-377.
[37] M. Behrens, F. Girgsdies, Z. Anorg. Allg. Chem. 2010, 636, 919-927.
[38] M. S. Spencer, Catal. Lett. 2000, 66, 255-257.
[39] J. C. J. Bart, R. P. A. Sneeden, Catal. Today 1987, 2, 1-124.
[40] A. M. Pollard, M. S. Spencer, R. G. Thomas, P. A. Williams, J. Holt, J. R. Jennings,
Appl. Catal. A 1992, 85, 1-11.
[41] G. C. Chinchen, K. C. Waugh, D. A. Whan, Appl. Catal. 1986, 25, 101-107.
[42] M. Kurtz, N. Bauer, C. Buscher, H. Wilmer, O. Hinrichsen, R. Becker, S. Rabe, K.
Merz, M. Driess, R. A. Fischer, M. Muhler, Catal. Lett. 2004, 92, 49-52.
[43] E. N. Muhamad, R. Irmawati, Y. H. Tautiq-Yap, A. H. Abdullah, B. L. Kniep, F.
Girgsdies, T. Ressler, Catal. Today 2008, 131, 118-124.
[44] P. Kurr, Technical University Berlin 2008.
[45] M. M. Günter, T. Ressler, B. Bems, C. Buscher, T. Genger, O. Hinrichsen, M. Muhler,
R. Schlögl, Catal. Lett. 2001, 71, 37-44.
[46] T. Ressler, B. L. Kniep, I. Kasatkin, R. Schlögl, Angew. Chem. Int. Ed. 2005, 44, 4704-
4707.
[47] I. Kasatkin, P. Kurr, B. Kniep, A. Trunschke, R. Schlögl, Angew. Chem. Int. Edit. 2007,
46, 7324-7327.
[48] R. Burch, R. J. Chappell, S. E. Golunski, Catal. Lett. 1988, 1, 439-443.
[49] Y. Kanai, T. Watanabe, T. Fujitani, M. Saito, J. Nakamura, T. Uchijima, Catal. Lett.
1994, 27, 67-78.
[50] R. G. Herman, K. Klier, G. W. Simmons, B. P. Finn, J. B. Bulko, T. P. Kobylinski, J.
Catal. 1979, 56, 407-429.
[51] J. C. Frost, Nature 1988, 334, 577-580.
[52] M. Behrens, F. Studt, I. Kasatkin, S. Kühl, M. Hävecker, F. Abild-Pedersen, S. Zander,
F. Girgsdies, P. Kurr, B.-L. Kniep, M. Tovar, R. W. Fischer, J. K. Nørskov, R. Schlögl,
Science 2012, 336, 893-897.
[53] R. N. d'Alnoncourt, X. Xia, J. Strunk, E. Löffler, O. Hinrichsen, M. Muhler, Phys.
Chem. Chem. Phys. 2006, 8, 1525-1538.
[54] M. V. Twigg, M. S. Spencer, Top. Catal. 2003, 22, 191-203.
Chapter 1: Introduction and Overview
12
[55] M. Kurtz, H. Wilmer, T. Genger, O. Hinrichsen, M. Muhler, Catal. Lett. 2003, 86, 77-
80.
[56] M. V. Twigg, M. S. Spencer, Appl. Catal. A 2001, 212, 161-174.
[57] C. Yang, Z. Y. Ma, N. Zhao, W. Wei, T. D. Hu, Y. H. Sun, Catal. Today 2006, 115,
222-227.
[58] E. K. Poels, D. S. Brands, Appl. Catal. A 2000, 191, 83-96.
[59] M. Saito, T. Fujitani, M. Takeuchi, T. Watanabe, Appl. Catal. A 1996, 138, 311-318.
[60] T. Fujitani, M. Saito, Y. Kanai, T. Kakumoto, T. Watanabe, J. Nakamura, T. Uchijima,
Catal. Lett. 1994, 25, 271-276.
[61] M. Behrens, S. Zander, P. Kurr, N. Jacobsen, J. Senker, G. Koch, T. Ressler, R. W.
Fischer, R. Schlögl, in Performance Improvement of Nano-Catalysts by Promoter-
Induced Defects in the Support Material: Methanol Synthesis over Cu/ZnO:Al,
submitted to J. Am. Chem. Soc.
[62] J. Sloczynski, Chem. Eng. Sci. 1994, 49, 115-121.
[63] I. Løvik, PhD Thesis, Norwegian University of Science and Technology 2001.
[64] M. S. Spencer, M. V. Twigg, in Ann. Rev. Mater. Res., Vol. 35, 2005, pp. 427-464.
[65] J. L. Li, T. Inui, Appl. Catal. A 1996, 137, 105-117.
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
13
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-
precipitated Mixed Cu,Zn Hydroxycarbonates – Consequences for the
Preparation of Cu/ZnO Catalysts
Stefan Zander, Beatrix Seidlhofer, Malte Behrens
Abstract
In order to better understand the critical influence of the synthesis parameters during preparation
of Cu/ZnO catalysts at the early stages of preparation, the aging process of mixed Cu,Zn
hydroxide carbonate precursors was decoupled from the precipitation and studied independently
under different conditions, i.e. variations in pH, temperature and additives, using in-situ energy-
dispersive XRD and in-situ UV-Vis spectroscopy. Crystalline zincian malachite, the relevant
precursor phase for industrial catalysts, was formed from the amorphous starting material in all
experiments under controlled conditions by aging in solutions of similar composition to the
mother liquor. The efficient incorporation of Zn into zincian malachite can be seen as the key of
Cu/ZnO catalyst synthesis. Two pathways were observed: Direct co-condensation of Cu
2+
and
Zn
2+
into Zn-rich malachite at 5 ≥ pH ≥ 6.5, or simultaneous initial crystallization of Cu-rich
malachite and a transient Zn-storage phase. This intermediate re-dissolved and allowed for
enrichment of Zn into malachite at pH ≥ 7 at later stages of solid formation. The former
mechanism generally yielded a higher Zn-incorporation. On the basis of these results, the effect
of synthesis parameters like temperature and acidity are discussed and their effect on the final
Cu/ZnO catalyst can be rationalized.
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
14
2.1 Introduction
Due to the enormous economical relevance of solid catalysts in chemical industry,
[1]
their
skillful synthesis and phenomenological optimization often is far more advanced than the
understanding of the rationale behind the resulting individual values of synthesis parameters.
Modern analytical methods can help to develop phenomenological catalyst synthesis towards
knowledge-based design. The Cu-based methanol synthesis catalyst is a prominent example for
this evolution.
Binary Cu/ZnO samples (Cu:Zn ca. 70:30) serve as a model system for the industrially applied
Cu/ZnO/Al
2
O
3
catalyst, which contains ca. 5-10 mol% Al
2
O
3
as a structural promoter.
Performance of the catalysts scales linearly with the accessible Cu surface area, but only within
certain families of Cu/ZnO catalyst, which were prepared by a similar method, e.g. by co-
precipitation or from citric acid melts.
[2]
This observation highlights the crucial influence of the
synthesis route on the catalytic properties of Cu/ZnO,
[3]
which is also termed the “chemical
memory” of the system.
[4]
The difference among the material families are attributed to intrinsic
promoting effects. Cu dispersion and intrinsic activity are beneficially influenced by the
presence and homogeneous distribution of ZnO in the catalyst. Firstly, it stabilizes small Cu
nanoparticles acting as a geometrical spacer between them.
[5]
Secondly, strong metal-oxide
interactions between ZnO and Cu are assumed to contribute to the in-situ formation of
catalytically active sites. Different models for this latter synergetic effect are discussed in
literature.
[6-13]
The most successful and industrially applied synthesis route of Cu/ZnO catalysts follows a
multi-step procedure in which mixed metal hydroxide carbonate precursors are formed by
controlled co-precipitation from aqueous Cu/Zn/(Al) nitrate solutions using soda solution as
precipitating agent.
[14]
Subsequently, the precipitate is aged in the mother liquor, filtrated,
washed, dried, calcined and finally reduced to yield the active catalyst. It is described in
literature, that aging is a crucial step during synthesis of the precursor and that it is essential for
preparation of a successful catalyst.
[4, 14-17]
In the following, we will discuss the influence of aging conditions on the properties of the
catalyst on the basis of the recently published model of hierarchical meso- and nano-structuring
of industrial methanol synthesis catalysts, which explains the benefit of the hydroxide carbonate
precursor method for preparation of Cu/ZnO catalyst (Figure 2-1a).
[18]
In brief, the co-
precipitate undergoes two micro-structure directing steps during preparation. Firstly, a mixture
Chapter 2: In-
situ EDXRD Study of the Chemistry of Aging of Co
Hydroxycarbona
of zincian malachite crystallizes from the initially amorphous co
during aging, both with the elemental formula (Cu,Zn)
with a minimum in pH and a color change from blue to bluish
amount
s of aurichalcite, (Cu,Zn)
was reported to lower the fraction of the aurichalcite phase in favor of zinc enriched
malachite.
[14, 16]
Crystallization of zincian malachite occurs preferably in form of very
interwoven needles, which leads to th
Figure 2-1: (
a) Cartoon of the preparation of Cu/ZnO catalysts comprising precipitation of zincian georgeite, aging to
form zincian malachite (meso-
structuring), decomposition in CuO/ZnO aggregates (nano
by reduction to Cu/ZnO. (
b) pH evolution during precipitation and aging of a typical binary sample with change in
sample color and crystallinity durin
aurichalcite by-phase, (Cu,Zn)
5
(CO
3
Secondly, the nano-
structuring of the individual precursor needles upon thermal decomposition
yields aggregates of CuO and ZnO nanoparticles. Thus, the hierarchical pore structure of the
final catalyst is already predetermined at the stage of the precursor
in zincian malachite needles is the crucial parameter, since significant amounts of atomically
distributed Zn in the joint cationic lattice of zincian malachite lead to an effective stabilization
situ EDXRD Study of the Chemistry of Aging of Co
-
precipitated Mixed Cu,Zn
Hydroxycarbona
tes –
Consequences for the Preparation of Cu/ZnO Catalysts
of zincian malachite crystallizes from the initially amorphous co
-
precipitate zincian georgeite
during aging, both with the elemental formula (Cu,Zn)
2
(OH)
2
(CO
3
).
[18]
This step is associated
with a minimum in pH and a color change from blue to bluish
green (
Figure
s of aurichalcite, (Cu,Zn)
5
(CO
3
)
2
(OH)
6
are often observed as a side-
phase. Further aging
was reported to lower the fraction of the aurichalcite phase in favor of zinc enriched
Crystallization of zincian malachite occurs preferably in form of very
interwoven needles, which leads to th
e proper porous meso-structure.
a) Cartoon of the preparation of Cu/ZnO catalysts comprising precipitation of zincian georgeite, aging to
structuring), decomposition in CuO/ZnO aggregates (nano
-
structuring) and activation
b) pH evolution during precipitation and aging of a typical binary sample with change in
sample color and crystallinity durin
g aging (insets) The marked reflections in the XRD pattern refer to the
3
)
2
(OH)
6
. All other reflections are due to zincian malachite (Cu,Zn)
structuring of the individual precursor needles upon thermal decomposition
yields aggregates of CuO and ZnO nanoparticles. Thus, the hierarchical pore structure of the
final catalyst is already predetermined at the stage of the precursor
. Here, the Zn concentration
in zincian malachite needles is the crucial parameter, since significant amounts of atomically
distributed Zn in the joint cationic lattice of zincian malachite lead to an effective stabilization
precipitated Mixed Cu,Zn
Consequences for the Preparation of Cu/ZnO Catalysts
15
precipitate zincian georgeite
[19]
This step is associated
Figure
2-1b). Small
phase. Further aging
was reported to lower the fraction of the aurichalcite phase in favor of zinc enriched
Crystallization of zincian malachite occurs preferably in form of very
thin and
a) Cartoon of the preparation of Cu/ZnO catalysts comprising precipitation of zincian georgeite, aging to
structuring) and activation
b) pH evolution during precipitation and aging of a typical binary sample with change in
g aging (insets) The marked reflections in the XRD pattern refer to the
. All other reflections are due to zincian malachite (Cu,Zn)
2
(CO
3
)(OH)
2
.
structuring of the individual precursor needles upon thermal decomposition
yields aggregates of CuO and ZnO nanoparticles. Thus, the hierarchical pore structure of the
. Here, the Zn concentration
in zincian malachite needles is the crucial parameter, since significant amounts of atomically
distributed Zn in the joint cationic lattice of zincian malachite lead to an effective stabilization
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
16
of the Cu phase in high dispersion in the decomposition product (Figure 2-1a).
[18]
Due to solid
state chemical constraints, the minimal Cu:Zn ratio in the zincian malachite phase is near
70:30.
[20]
For an efficient nano-structuring, the highest possible fraction of the available Zn
amount should be incorporated in the zincian malachite precursor during aging.
The Zn fraction in this phase can be determined from the peak position of the 201
reflection in
the XRD pattern of zincian malachite. A low corresponding d-spacing is indicative of a high Zn
content, which can be explained by a gradual contraction of this net plane distance caused by
the average lowering of Jahn-Teller distortions of the octahedral MO
6
building blocks in
malachite as Cu
2+
is gradually replaced by Zn
2+
.
[18, 21]
The shift of the 201
reflection in the XRD
pattern is, thus, a direct measure of the desired incorporation of Zn
2+
into the malachite structure
and serves as an estimate of the Cu dispersion in the final catalyst.
Hence, aging, i.e. the period of crystalline phase formation of the precursor, plays a key role for
catalyst preparation and for the so-called chemical memory of Cu/ZnO catalysts. However, the
effects of synthesis parameters like pH, temperature or mother liquor composition on the
precipitate are not well understood and are so far related to the catalytic performance of the
resulting Cu/ZnO catalyst only in a merely phenomenological manner. This lack of
understanding can be seen as a major hindrance for further rational optimization of the
Cu/ZnO/(Al
2
O
3
) system and requires a systematic and fundamental study of the chemistry of
precipitate aging. Such a study is complicated by the fact that variation of a given parameter
affects upstream precipitation as well as aging. The ambiguity if an observed change in the
properties of the precipitate is a result of modified chemistry of aging or of changes in the
precipitation process (resulting in a different starting material for downstream aging), requires
experimental decoupling of both events. Furthermore, application of in-situ methods is desirable
to ensure complete monitoring of all transformations happening during aging of the co-
precipitate. In-situ energy-dispersive X-ray diffraction (EDXRD) has been shown to be a
powerful method to study the mechanism
[22]
and kinetics
[23]
of solid state reactions,
[24-25]
e.g.
under hydrothermal conditions
[26]
or in intercalation/de-intercalation reactions.
[27]
In this paper we report a novel approach for the investigation of the aging process during
preparation of Cu/ZnO catalysts using decoupled precipitation and aging steps and in-situ
EDXRD and UV-Vis spectroscopy to monitor the latter.
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
17
2.2 Experimental
2.2.1 Precursor Preparation
Decoupling of precipitation and aging was realized by continuously feeding the initial
amorphous co-precipitate slurry directly into a spray-dryer in order to suppress aging by fast
drying. This “quenching technique” was necessary due to the fact that zincian georgeite is quite
unstable in the mother liquor against crystallization. In course of the preparation, constant pH
co-precipitation was performed in an automated laboratory reactor (Mettler-Toledo LabMax, 2
L, prefilled with 400 mL water) at T = 338 K and pH 7 from aqueous 1.6 M Na
2
CO
3
solution
and 1 M aqueous metal nitrate solution (Cu:Zn = 70:30). It is noted that the conditions of co-
precipitation correspond to the conventional preparation process described in literature and were
similar to the conditions of industrial catalyst preparation. A graphical representation of the
precipitation log file can be found as supporting information (Figure S2-1). The resulting slurry
was continuously removed from the co-precipitation reactor at the rate of addition of solutions
(23 mL/min) and directly spray-dried (Niro Minor Mobile, T
inlet
= 473 K, T
outlet
= 373 K) after
an estimated residence time of less than 20 min in the reactor and the connecting tubes. This is
well below the aging period necessary for crystallization of the precursor material considering
that pH minimum and color change are not expected before ca. 30 minutes of stirring in the
mother liquor under these conditions.
[18]
Thus, the dried, solid product was X-ray amorphous
except for some NaNO
3
resulting from crystallization of the counter ions during spray-drying
(not shown). To remove NaNO
3
the precursor was thoroughly washed with cold water and
spray-dried again leading to completely X-ray amorphous zincian georgeite. The Cu:Zn ratio of
the solid was confirmed to be 73:27 (± 2%) by XRF. The resulting precursor is referred to as
“unaged” despite its residence time of 20 min in the mother liquor because of the fact that it still
was amorphous. Using this procedure, which is schematically summarized in Figure S2-2, we
were able to employ a batch of unaged zincian georgeite as identical starting material for aging
experiments under different conditions (T, pH, additives) in mother liquor-analogous media
without affecting the co-precipitation process.
2.2.2 In-situ EDXRD and UV-Vis Spectroscopy during Simulated Aging
In order to simulate the aging process, 200 mg of the precursor were suspended in 2 mL of
aging solution in a glass tube (internal diameter: 10 mm; volume: 7 mL). To keep the
concentrations of the relevant ions in the aging solution near to the concentrations in the real
mother liquor, it was freshly prepared by mixing appropriate amounts of the basic precipitating
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
18
agent (1.6 M alkaline carbonate solution A
2
CO
3
, A = Na, K) and HNO
3
of a concentration
corresponding to that of the mixed metal nitrate solution (1 M) until the desired pH was reached
and stable. The in-situ measurements were started directly after preparation of the suspension.
The uncovered glass tube was placed into a metal block, whose temperature was controlled by
an oil bath. The suspension inside the glass tube was stirred during the aging experiment using a
magnetic stir bar.
All in-situ aging investigations were carried out at the beamline F3 at HASYLAB/DESY,
Hamburg, Germany. The beamline station receives white synchrotron radiation from a bending
magnet with a critical energy of 16 keV and gives a positron beam energy of 4.5 GeV allowing
detection of an energy range from 10 to 60 keV with a maximum in intensity at about 20 keV.
An energy dispersive germanium detector was used to monitor the diffracted beam after
transmission through the sample at a fixed angle, which was chosen as approximately 3.6°
covering a d-spacing range of 2.6 to 12.2 Å. The beam was collimated to 100 × 100 µm. An
acquisition time of 120 s yielded time-resolved X-ray powder patterns with sufficient counting
statistics. The time span from placing the sample in the sample holder and start of recording the
first diffraction patterns was less than 60 sec. The resulting spectra were evaluated using the
EDXPowd
[28]
program package. More details on the experimental setup used can be found in the
supporting information (Figure S2-3) and literature.
[29]
Phase evolution was followed by plotting
the integral intensity of selected well-resolved reflection as a function of time. Additionally, the
change of the color of the samples was tracked by UV-Vis spectroscopy in order to monitor the
conversion of blue amorphous zincian georgeite to green crystalline zincian malachite. Diffuse
reflectance measurements were performed with an OceanOptics optical fiber probe placed in the
suspension well above the synchrotron beam. The probe was connected with a
TopSensorSystems halogen lamp and an OceanOptics high resolution spectrometer
HR2000CG-UV-NIR. The acquisition time was set to 120 s per spectrum
2.2.3 Ex-situ Characterization
All samples subjected to EDXRD measurements were cooled to room temperature after the in-
situ experiments within 5 minutes, filtrated and washed with water. Conventional X-ray
diffraction (XRD) measurements were performed with a STOE STADI P transmission
diffractometer equipped with a primary focusing Ge monochromator (Cu-K
α1
radiation) and
position-sensitive detector to determine the peak positions more accurately than was possible
with EDXRD. All XRD patterns are presented as supporting information (Figure S2-4). The
samples were mounted in the form of a clamped sandwich of small amounts of powder fixed
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
19
with a small amount of grease between two layers of thin polyacetate film. Refinements were
done in the 2θ range 4-80° using the software package TOPAS.
[30]
Domain sizes were
determined from the XRD peak widths and are given as volume weighted mean column heights.
Surface area determination was performed in a Quantachrome Autosorb-6 machine by N
2
-
adsorption-desorption using the BET method. Cu:Zn ratios of the samples were obtained from
X-ray fluorescence (XRF) measurements using a Bruker S4 Pioneer X-ray spectrometer.
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
20
2.3 Results and Discussion
2.3.1 General
Using the method of sample preparation described in the experimental section allowed
simulating the aging process of an amorphous binary zincian georgeite co-precipitate under
controlled conditions similar to those used in course of preparation of industrial methanol
synthesis catalysts (Figure 2-1). In-situ EDXRD and UV-Vis measurements allowed insight into
the chemistry of aging, which is of crucial importance for the phase formation of the catalyst
precursor and, thus, for the preparation of highly active catalysts.
The catalyst precursor was aged at different temperatures (323-343 K), starting acidities (pH
5.0–8.0) and using different counter cations (Na
+
and K
+
). Crystalline zincian malachite
(Cu,Zn)
2
(OH)
2
(CO
3
) was finally detected by XRD after aging for all conditions applied
(supporting information, Figure S2-4). It is interesting to note that minor amounts of
aurichalcite are typically observed for the Cu:Zn ratio of 70:30, e.g. after aging in a
conventional 2-L batch reactor at 338 K and pH of 7.0 (see also marked reflections in Figure
2-1b).
[18]
The presence of aurichalcite was barely detectable in the samples after simulated aging
by ex-situ XRD (supporting information, Figure S2-4). Inclusion of the aurichalcite phase in the
Rietveld fits led to improved R-values for some samples and resulted in varying amounts of
aurichalcite between 0 and 13 wt.% (Table 2-1). However, due to the poor crystallinity of the
sample and the low amount of this phase the error is estimated to be at least ± 5 wt.%. A typical
graphical representation of a typical Rietveld fit is given in Figure 2-2. In addition to the small
differences in phase composition among the aged samples, variations in crystallinity, Zn content
of the zincian malachite phase and specific surface area as a result of different aging conditions
are reflected in the XRD domain sizes scattering between 9.2 and 10.9 nm, the d
value
ranging between 2.757 and 2.775 Å and the BET surface areas being between 69 and 85 m
2
g
-1
(Table 2-1). These observations confirm the sensitivity of relevant properties of the Cu,Zn
precursor to the exact conditions of crystallization for the same starting material and a
systematic discussion will be given in the following.
2.3.2 Phase Evolution
The phase evolution during precursor aging will be discussed for the experiment conducted at T
= 323 K and pH 7.0 using a Na
+
containing aging solution (ID 7 in Table 2-1). Despite the X-
ray amorphous dry starting material, some very weak XRD peaks are already observed in the
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
21
first in-situ pattern of the slurry recorded after less than 180 sec after starting the experiment
(Figure 2-3a). After an aging time of some minutes, a steep increase in Bragg peak intensity is
observed and two phases can be clearly distinguished (Figure 2-3b): Sodium zinc carbonate,
Na
2
Zn
3
(CO
3
)
4
•3H
2
O,
[31]
and the target material zincian malachite, (Cu,Zn)
2
(OH)
2
(CO
3
).
[32]
At
the end of the experiment, only zincian malachite was observed as the final product (Figure
2-3c). Sodium zinc carbonate Na
2
Zn
3
(CO
3
)
4
•3H
2
O has been reported before in literature in the
context of Cu/ZnO catalyst preparation. It was identified as the initial precipitate in Zn-rich or
pure Zn systems, which upon aging transformed into aurichalcite or hydrozincite.
[15, 33]
In one
study, it has probably also been detected in a Cu-rich system as a transient phase during
aging,
[34]
but was assigned as “crystalline zincian georgeite”.
Table 2-1: Summary of aging parameters and aging results. The sample ID 0 refers to the unaged precursor.
Aging
Conditions In-situ results
(EDXRD data) Ex-situ results
(recovered samples after EDXRD measurement)
ID
pH
T
[K]
A
+
in
A
2
CO
3
Na,Zn
Inter-
mediate Onset
[min]
Reaction
Time
[min]
[a]
Aurichalcite
[wt%]
(± 5 wt%)
[Å]
Zn in
zM
[b]
[%] BET
[m²/g]
FWHM-
LVol
[c]
[nm]
Cu:Zn XRF
[mol%]
(± 2 mol%)
0
- - - - - - - - - 15 - 73.3:26.7
1
5 333
Na
+
- 20 - 6 2.757
29.2 85 9.9 73.2:26.8
2
6 333
Na
+
- 34 - 8 2.759
28.5 81 9.7 72.3:27.7
3
6.5
333
Na
+
- 36 - 8 2.760
28.4 83 10.0 72.8:27.2
4
7 333
Na
+
x 12 30 0 2.767
26.2 82 9.6 73.8:26.2
5
7.5
333
Na
+
x 12 34 0 2.767
26.3 80 9.8 72.1:27.9
6
8 333
Na
+
x 14 34 0 2.768
26.0 83 9.6 72.5:27.5
7
7 323
Na
+
x 24 98 0 2.775
23.8 81 9.2 73.9:26.1
8
7 343
Na
+
x 6 16 0 2.765
26.7 72 10.9 73.9:26.1
9
7 333
K
+
- 56 - 13 2.767
26.2 77 9.8 72.3:27.7
10
7 343
K
+
- 18 - 11 2.775
23.8 69 10.0 72.0:28.0
[a] Time interval between appearance and complete consumption of the intermediate.
[b] Zn content in zincian malachite (zM) calculated
from d
values
; see also Figure 2-7.
[c] Crystallite sizes of zincian malachite determined from the half width of the XRD peaks using the TOPAS refinement software.
One suitable, well resolved peak of both phases was chosen for further EDXRD data evaluation.
The phase fraction of zincian malachite was represented by the integral intensity of the 201
peak, sodium zinc carbonate by the 222 peak. The phase evolution with time is shown in Figure
2-4a. The sodium zinc carbonate phase crystallizes in parallel to the zincian malachite phase and
re-dissolves upon prolonged aging. The onset of crystallization occurs after 24 min and the re-
dissolution of the sodium zinc carbonate occurs over 98 min without a significant increase in the
zincian malachite phase.
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
22
The amorphous starting material (“zincian georgeite”) is hard to comprehensively characterize.
IR-spectroscopic studies of the unaged material have shown the presence of both hydroxide as
well as carbonate anions.
[4, 18]
Here, we describe the starting precipitate as an amorphous double-
salt with unknown anionic composition: a-Cu
0.7
(OH)
x
(CO
3
)
0.7-x/2
•Zn
0.3
(OH)
y
(CO
3
)
0.3-y/2
. In the
presence of varying amount of H
2
O, OH
-aq
and CO
32-aq
we can write for the two steps of the
aging reaction:
a-Cu
0.7
(OH)
x
(CO
3
)
0.7-x/2
•Zn
0.3
(OH)
y
(CO
3
)
0.3-y/2
+ z Na
+
aq
→ (Cu
>0.7
Zn
<0.3
)
2
(OH)
2
CO
3
+ z/2 Na
2
Zn
3
(CO
3
)
4
•3H
2
O
→ (Cu
0.7
Zn
0.3
)
2
(OH)
2
CO
3
+ z Na
+aq
(Eq. 2-1)
Figure 2-2: Rietveld refinement of the ex-situ XRD pattern of the sample aged in-situ at pH 6.5 and 333 K (ID 3) for
quantitative analysis of the phase composition. Experimental data (circles), background (dotted), background peak
(dashed, due to the grease used as sticking agent to keep the sample in place on the sample holder), calculated pattern
zincian malachite (green), calculated pattern aurichalcite (orange), total calculated curve (line) and difference curve
(grey, offset -100). This plot is representative for the other aged samples. In this case the ratio of zincian malachite to
aurichalcite was calculated to be 92% to 8%.
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
23
Figure 2-3: EDXRD patterns (converted to °2θ values of Cu K
α
radiation) during aging of the amorphous precursor
at pH 7 and 323 K (ID 7) after two (a), 26 (b) and 98 min (c). At the bottom PDF 72-75 (green bars) and PDF 1-457
(red bars) are shown as references for zincian malachite (Cu,Zn)
2
(CO
3
)(OH)
2
and for sodium zinc carbonate
Na
2
Zn
3
(CO
3
)
4
·3 H
2
O, respectively. The in-situ EDXRD spectra are representative for all conducted experiments. The
position of the 201
peak of zincian malachite is shifted compared to the pure malachite reference because of zinc
incorporation (see text).
Figure 2-4: Integral intensity of selected EDXRD peaks vs. aging time (a); in Na
2
CO
3
; pH = 7; T = 323 K (ID 7).
Zincian malachite, (Cu,Zn)
2
(OH)
2
(CO
3
), is represented by the 201
peak (green), sodium zinc carbonate
Na
2
Zn
3
(CO
3
)
4
·3 H
2
O by the 222 peak (red). Corresponding d-spacing of the 201
peak of zincian malachite (b).
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
24
If we assume Na
2
Zn
3
(CO
3
)
4
•3H
2
O to be a pure Zn-phase with no Cu incorporation, the initially
formed zincian malachite should be poor in Zn. In the following aging step, the sodium zinc salt
acts as a Zn storage phase slowly deliberating its Zn content by dissolution. This fraction of Zn
can either re-precipitate in form of a Zn-phase not detectable by XRD, or – as proposed in
equation (1) – it can be incorporated into zincian malachite by re-crystallization increasing the
Zn-content of this phase. The latter possibility is supported by the evolution of the 201
peak
position (Figure 2-4b). As the Zn storage phase re-dissolves, the peak is shifted to a lower d-
spacing, indicating further incorporation of Zn into zincian malachite. It is noted, however, that
a final proof of this mechanism is still lacking as there is a peak overlap of the 201
of zincian
malachite around 32.8 °2θ (for Cu K
α
radiation) and the 422 of the sodium zinc salt located at
32.2 °2θ according to PDF 1-457. The intensity ratio of these two peaks is around 5:1 in this
stadium. Diminishing of the latter peak due to dissolution may alone result in an artificial profile
shift to higher angles in the EDXRD patterns. Unfortunately, the quality of the in-situ EDXRD
patterns is not sufficient for a whole pattern refinement. After all, the above made assumption of
an intrinsic peak shift of the 201
of zincian malachite seems reasonable, because no narrowing
of the peak profile with time was observed, which would be associated with decrease of a
shoulder (see supporting information, Figure S2-5). Furthermore, the intensity of the 422 of
sodium zinc carbonate is only 20% of the most intensive reflection of that phase, while the 201
of zincian malachite is the strongest reflection of this phase. In Figure 2-3a, where only the
sodium zinc carbonate phase is present, no significant intensity due to the 422 can be seen at a
position corresponding to ca. 32 °2θ for Cu K
α
, suggesting that the contribution of the
overlapping to the peak position of the phase mixture has only a minor influence.
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
25
Figure 2-5: UV-Vis results for simulated aging in Na
2
CO
3
at pH 7; 323 K (ID 7). a) Measured diffuse reflectance
before and after aging. b) Difference of normalized spectra relative to the initial spectrum at t = 2 min. c) Wavelength
of the maximum intensity in the UV-Vis spectra in the range of 425 to 900 nm as a function of time.
The UV-Vis diffuse reflectance spectra of the suspension corresponding to the starting material
and the final product (aging conditions pH 7, 323 K, ID 7) are shown in Figure 2-5a. The
change of the position of the broad signal from 505 to 515 nm reflects a change in crystal field
splitting around the Cu
2+
ions and the transition from blue to bluish green.
[35]
Difference plots of
the normalized in-situ recorded spectra are shown in Figure 2-5b. It can be seen that several
smaller bands contribute to the spectra. The presence of an isosbestic point near 510 nm was
observed for all experiments and suggests that the starting material directly transforms into a
single optically active product. This does not contradict the transient presence of the sodium
zinc carbonate phase, but rather confirms the assumption that this phase does not contain Cu
2+
ions and does not contribute to the reflectance in the Vis-range of the optical spectrum. The
green part of the spectrum does hardly change and it can be seen that the change of color from
blue to green is mostly due to an increase in reflectance in the yellow regime of the spectrum,
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
26
which is much stronger compared to the increase in the blue part. The temporal evolution of the
UV-Vis spectrum shows only minor changes in the beginning of the reaction up to aging times
of 22 min. During this period a slight decrease of reflectance in the yellow and blue part is
observed, which started directly as the starting material was in contact with the aging medium.
In accordance with the EDXRD results, abrupt changes occur at an aging time of 24 min and
reflectance in these parts of the spectrum sharply increases. Interestingly, the color change is
finished almost immediately (after less than 4 min) and again only little changes are observed at
26 < t < 120 min, while the process of phase formation observed by EDXRD persists for 98
min. This clearly shows that the change of the color of the precursor slurry is not a suitable
indicator for the end of the chemical changes happening during aging. UV-Vis spectroscopy
probes changes on the molecular level, which naturally precede the detection with (ED)XRD
technique as crystallization requires “oversaturation” of the newly formed complexes, which
happens over a longer time scale under the conditions applied. In Figure 2-5c, the maxima of the
broad reflectance signal are shown as a function of aging time, showing again the step-like
change at the time of crystallization of malachite.
Figure 2-6: Measured features vs. aging time for samples noted in the plot (cf. Table 2-1). Top row a-d: Integral
intensity of selected EDXRD peaks of zincian malachite 201
, (green) and sodium zinc carbonate 222, (red). Bottom
row e-h: Evolution of d
values of zincian malachite.
No residual sodium zinc carbonate or other by-phases were detected after 120 min of aging. In
particular, no aurichalcite was detected in the in-situ EDXRD patterns. Aurichalcite might play
a similar role as a Zn-uptake phase during aging.
Waller et al.
[14]
investigated the aging mechanism of a Cu:Zn = 67:33 system and observed that
at first a mixture of crystalline zincian malachite (Cu:Zn ≈ 85:15) and aurichalcite (Cu:Zn ≈
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
27
60:40) were formed and subsequently transformed into zinc richer malachite (Cu:Zn ≈ 67:33) at
the expanse of aurichalcite. We previously found that for conventional batch aging of a binary
precursors (Cu:Zn = 70:30) for 2 hours at 338 K low amounts of the zinc richer phase
aurichalcite co-exist with zincian malachite showing a Zn content of 27%.
[18, 20]
As mentioned
above, low amounts of aurichalcite were detected by ex-situ XRD indicating that this phase
indeed may act as a stable sink for Zn, but its amount probably is too little to be detected by in-
situ XRD or that it has crystallized only upon drying of the samples.
2.3.3 The Effect of Temperature
At an aging temperature of 323 K, the sodium zinc carbonate storage phase was re-dissolved
within 16 - 98 min upon aging at pH 7.0, depending on the temperature (Figure 2-4a and Figure
2-6a,b, Table 2-1, ID 7, 4, 8). The change of the color, assigned to the beginning formation of
crystalline zincian malachite, always occurred within a few minutes and was tracked by UV-
Vis-spectroscopy (Figure 2-5c).
Figure 2-7: Calibration of the d
values versus the Zn content in zincian malachite. The three solid data points
stem from reference samples described in ref.
[18, 21]
and were used for linear extrapolation. The open data points
positioned onto the extrapolated line refer to the values observed in this study (from ex-situ XRD). The resulting Zn-
contents are given in Table 2-1.
Increasing the temperature from 323 K to 333 or 343 K changed the kinetics of aging and led to
earlier onset of crystallization and a shorter time period of existence of the sodium zinc phase
(Figure 2-6a,b and Table 2-1, ID 4, 7 and 8), but no changes in the mechanism of aging were
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
28
observed. The monotonous shift to lower d
values (Figure 2-6e,f) indicates the incorporation
of Zn into zincian malachite with time after crystallization. Ex-situ XRD was applied on the
recovered samples, which allows for a more accurate determination of the absolute peak
positions than in-situ EDXRD due to higher instrumental resolution, longer acquisition time,
more precise calibration and lower contribution of the background. The effect of temperature on
the final degree of Zn incorporation in zincian malachite is reflected in the shift of the d
spacing (Table 2-1, ID 4, 7 and 8). While d
was similar for the higher temperatures, it was
found to be significantly larger for the zincian malachite sample prepared at 323 K indicating a
lower final degree of Zn incorporation at this temperature. This observation shows that, despite
stemming from the same amorphous starting material, the degree of Zn incorporation can be
affected by the aging conditions. The detrimental effect of low preparation and aging
temperatures has been reported in the literature.
[3, 36]
A lack of Zn has been also observed for
ternary Cu/ZnO/Al
2
O
3
catalysts prepared at low pH or low temperature.
[3]
Our results suggest
that this effect can be explained with a lack of Zn in the zincian malachite precursor phase. This
is consistent with observations recently made during titration experiments
[37]
showing that the
precipitation pH of Zn
2+
is shifted to higher pH values as temperature decreases. Thus, the
applied aging pH value may not be sufficiently basic to keep all Zn in the solid state at low
temperatures and Zn
2+
may be leached out of the precipitate at low temperatures and pH values.
Interestingly, XRF measurements of the sample recovered after aging at different temperatures
all showed the same average Cu:Zn ratio near 70:30 (Table 2-1). We thus conclude that the final
pH is high enough to completely precipitate Zn
2+
also at 323 K, but suggest that during the
crystallization process, which is associated with an intermediate minimum in pH peaking
roughly a full pH unit below the initial aging pH under these conditions
[18]
(Figure 2-1b), a
transient leaching of Zn
2+
from the starting material may occur at low temperatures during the
pH minimum. This can explain a lack of Zn in the zincian malachite phase due to (partial) Zn
dissolution at the time of its crystallization. Later re-precipitation leads to formation of low
amounts of undetected Zn-rich phases resulting in the same average composition, but in an
inhomogeneous and thus unfavorable Zn distribution in the product.
Thus, equation 1 has to be revised as (Cu
0.7
Zn
0.3
)
2
(OH)
2
CO
3
is not an appropriate representation
of the final product, which exhibits variations in its Cu:Zn ratio. We add an unknown Zn phase
denoted Zn↓ as sink for “extra-lattice” Zn. The nature of this phase may be residual but
undetected sodium zinc carbonate, amorphous or due to low abundance undetectable crystalline
aurichalcite or another form of Zn-containing hydroxide or basic carbonate. As a function of the
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
29
aging conditions this phase is present in various amounts and affects the Cu:Zn ratio in zincian
malachite by limiting the available amount of Zn:
a-Cu
0.7
(OH)
x
(CO
3
)
0.7-x/2
•Zn
0.3
(OH)
y
(CO
3
)
0.3-y/2
+ z Na
+
aq
→ (Cu
>>0.7
Zn
<<0.3
)
2
(OH)
2
CO
3
+ z/2 Na
2
Zn
3
(CO
3
)
4
•3H
2
O
→ (Cu
>0.7
Zn
<0.3
)
2
(OH)
2
CO
3
+ z Na
+aq
+ w Zn↓
(Eq. 2-2)
A quantification of the Zn-content in the final zincian malachite phase is possible by assuming a
Vegard-type behavior of this lattice spacing and calibrating the obtained d
values with
reference values from (zincian) malachite samples far from its limit of Zn incorporation. This is
shown in Figure 2-7. If the data points are arranged on the extrapolated line according to their
d
values, they cover variations in the Zn-content of zincian malachite between 23.8 and
29.2% Zn as a function of different aging conditions (Table 2-1). The largest Zn-contents
determined with this method are very close to the nominal Cu:Zn-ratio applied during synthesis
of 70:30, but exceed the Zn-content of the starting material determined by XRF. This
discrepancy is attributed to the difference of the two methods and their calibration errors. In
case of the sample obtained at 323 K the Zn content in zincian malachite is only 23.8%, while
the rest of Zn is trapped in a relatively large amount of the Zn-sink phase Zn↓. At 333 or 343 K,
the amount of Zn↓ is lower and the Zn-content in zincian malachite is 26.2 and 26.7%,
respectively.
It is noted that a beneficial effect of lowering the temperature on the crystallite size is observed,
which decreases with temperature from 10.9 nm at 343 K to 9.2 nm at 323 K. Accordingly, the
lowest BET surface areas were found for the samples prepared at 343 K (Table 2-1).
In summary, the effect of increasing temperature accelerate the crystallization kinetics, leads to
larger crystallites and, thus, is detrimental for the meso-structure of the catalyst precursor.
Lowering the temperature to 323 K, however, leads to intermediate leaching of Zn
2+
from the
co-precipitate and to an unfavorable Zn distribution resulting in a lower degree of Cu,Zn-
substitution of the zincian malachite phase. This hinders an effective nano-structuring of the
catalyst. Thus, the empirically optimized aging temperature around 338 K can be understood
from the chemistry of aging of the precursor and envisaged as the optimum of two antagonistic
trends.
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
30
Figure 2-8: d
values of zincian malachite (from ex-situ XRD) depending on the pH value of simulated aging. Two
different groups can be observed: The zincian malachite samples crystallized without intermediate formation of
sodium zinc carbonate (pH 5-6.5) show low d
values indicating high Zn content. Crystallizations via the
intermediate (pH 7-8) led to high d
values or low Zn content, respectively. Lines are guides for the eye.
2.3.4 The Effect of Acidity
The effect of different pH values during aging was investigated at 333 K with Na
+
containing
aging solutions. The phase evolution data for aging at pH 7.0 is shown in Figure 2-6a and only
minor differences were observed if the experiment was conducted at a higher pH of 7.5 or 8.0
(Table 2-1, ID 4-6; supporting information: Figure S2-6). Also the crystallite sizes, specific
surface areas and Zn contents of the resulting zincian malachite precursors were similar.
Reducing the pH, however, strongly affected the mechanism of the aging process. Most striking
is the absence of the sodium zinc carbonate phase at low pH associated with a delay of
crystallization (Table 2-1, ID 1-3). Only for pH ≥ 7 sodium zinc carbonate was detected, while
zincian malachite crystallized from the amorphous starting material at pH ≤ 6.5 without
participation of any other EDXRD-detectable phase. The corresponding evolution of crystalline
phases is presented in Figure 2-6c for aging at pH 5 and as supporting information for the other
pH values (Figure S2-6). The decrease of intensity in Figure 2-6c is probably due to partial
dissolution of zincian malachite at low pH value. Figure 2-6g shows that d
is constant or
even slightly increasing directly after the crystallization period suggesting that there is hardly
any change of the Zn-content in zincian malachite with aging time for these samples. Thus, a
second, simpler mechanism of aging is present with only one detectable step:
a-Cu
0.7
(OH)
x
(CO
3
)
0.7-x/2
•Zn
0.3
(OH)
y
(CO
3
)
0.3-y/2
(Eq. 2-3)
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
31
→
(Cu
>0.7
Zn
<0.3
)
2
(OH)
2
CO
3
+ w Zn↓ (Eq. 3)
It can be seen from Table 2-1 that this mechanism of crystallization seems kinetically hindered
as longer isothermal induction periods are required compared to reactions following the
mechanism of equation 2 at the same temperature. The preferred Zn-sink phase Zn↓ for this
mechanism can be identified as aurichalcite (Table 2-1).
Interestingly, despite the presence of significant amounts of aurichalcite, there was a
significantly higher degree of Zn-incorporation in the resulting zincian malachite if crystallized
in absence of the sodium zinc carbonate according to equation 3. Accordingly, ex-situ XRD
evaluation suggested the presence of two groups of precursors (Figure 2-8): The precursors
crystallized without sodium zinc carbonate obtained at low pH with large amounts of Zn
incorporated in the cationic lattice of zincian malachite (small d
, 28.4-29.2% Zn in zincian
malachite) and the ones obtained at higher pH with significantly lower amounts of Zn on Cu-
sites (large d
, 26.0-26.3% Zn in zincian malachite). The overall Cu:Zn ratio of the recovered
solid detected by XRF was always near the starting composition for all samples (Table 2-1)
suggesting again that other non-detectable Zn-rich by-phases Zn↓ are present and act as a sink
for Zn, in particular if crystallization occurred at high pH in presence of sodium zinc carbonate.
The highest Zn incorporation of this study was detected for the samples obtained at T = 333 K
and pH 5. According to Figure 2-7, the Zn content of zincian malachite is 29.2%. The successful
minimization of any form of Zn segregation – like the transient crystallization of sodium zinc
carbonate or the formation of the Zn-rich aurichalcite phase – helps to prepare a homogeneous
precipitate capable of efficient nanostructuring during thermal decomposition according to the
scheme presented in Figure 2-1a.
2.3.5 The Effect of Potassium Counter Ions
In case of the mechanism described by equation 2, it is tempting to relate the lack of Zn in the
zincian malachite precursor to the amount of Zn, which has intermediately formed the sodium
zinc salt. In order to test this hypothesis, analogous experiments were performed using an aging
solution based on neutralized K
2
CO
3
solution. The absence of Na
+aq
should suppress the
formation of the transient sodium salt also at higher pH and therefore may have a promoting
effect on the desired incorporation of Zn in zincian malachite at these conditions.
The experiments at pH 7 (T = 333, 343 K) were repeated using K
2
CO
3
instead of Na
2
CO
3
in the
aging medium (Table 2-1, ID 9,10). As expected, no sodium zinc carbonate and no other by-
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
32
phases were found in contrast to the aging experiment with Na
2
CO
3
under the same conditions.
This observation highlights the unexpected influence of the alkali metal counter ion on the
chemistry of aging under these conditions. The crystallization of zincian malachite in the
absence of Na
+
was strongly delayed (Figure 2-6a,d cf. ID 4 and ID 9 F; Table 2-1, cf. ID 4, 8
and 9, 10). With exception of the first two patterns, there was no significant down-shift in d
indicating the absence of any other transient Zn-storage phase (Figure 2-6h). Surprisingly,
although the transient formation of sodium zinc carbonate was suppressed, the final d
values
of the resulting zincian malachite phase were comparable to those obtained with Na
2
CO
3
. The
crystallization of (Cu-rich) malachite can happen faster if a transient storage phase for Zn can
form, but the final degree of Zn incorporation in zincian malachite is similar at a given pH
value. This indicates that Zn incorporation is rather determined by the acidity of the aging
medium, i.e. by thermodynamics, than affected by the transient segregation chemistry, i.e. by
kinetics. It is the availability of H
+
, which seems to promote or limit the zinc incorporation into
zincian malachite and balances the ratio of Zn deposited into the Zn-sink phase Zn↓ (Figure 2-8)
via the one or the other pathway. The lower limit of the pH value is, however, given by the
partial dissolution of Zn
2+
in acidic environment leading to incomplete solidification
[36]
or
leaching and unfavorable Zn distribution (cf. section 2.3.4).
Chapter 2: In-
situ EDXRD Study of the Chemistry of Aging of Co
Hydroxycarbona
Figure 2-9:
Radar plots illustrating the influence of the aging parameters pH value (a, at T = 333
(b, at pH = 7) on the aging reaction and the properties of the resulting zincian malachite material. Variation of pH
leads to different aging mechani
sms below pH 6.5 (a, broken lines) and above pH 7 (a, full lines) and affects the Zn
content of the catalyst precursor. Temperature leads to a gradual change of the aging kinetics (b, full lines).
Substitution of Na
+
by K
+
in the aging solution at the same
the aging reaction and the phase composition, but not on the Zn content of the zincian malachite precursor. As a
function of temperature, a variation of the Zn content is observed with Na
2.4 Conclusion
The aging process of mixed
precipitation and studied independently using
Crystalline zincian malachite, the desired precursor phase for Cu/Zn
successfully formed from the amorphous starting material in all experiments by aging in
solutions with a composition near to the mother liquor under controlled co
situ EDXRD Study of the Chemistry of Aging of Co
-
precipitated Mixed Cu,Zn
Hydroxycarbona
tes –
Consequences for the Preparation of Cu/ZnO Catalysts
Radar plots illustrating the influence of the aging parameters pH value (a, at T = 333
(b, at pH = 7) on the aging reaction and the properties of the resulting zincian malachite material. Variation of pH
sms below pH 6.5 (a, broken lines) and above pH 7 (a, full lines) and affects the Zn
content of the catalyst precursor. Temperature leads to a gradual change of the aging kinetics (b, full lines).
in the aging solution at the same
temperature (a, dotted lines) has a pronounced effect on
the aging reaction and the phase composition, but not on the Zn content of the zincian malachite precursor. As a
function of temperature, a variation of the Zn content is observed with Na
+
and K
+
.
The aging process of mixed
Cu,Zn
hydroxycarbonate precursors was decoupled from the
precipitation and studied independently using
in-situ EDXRD and in-situ UV
-
Crystalline zincian malachite, the desired precursor phase for Cu/Zn
O catalysts, was
successfully formed from the amorphous starting material in all experiments by aging in
solutions with a composition near to the mother liquor under controlled co
precipitated Mixed Cu,Zn
Consequences for the Preparation of Cu/ZnO Catalysts
33
Radar plots illustrating the influence of the aging parameters pH value (a, at T = 333
K) and temperature
(b, at pH = 7) on the aging reaction and the properties of the resulting zincian malachite material. Variation of pH
sms below pH 6.5 (a, broken lines) and above pH 7 (a, full lines) and affects the Zn
content of the catalyst precursor. Temperature leads to a gradual change of the aging kinetics (b, full lines).
temperature (a, dotted lines) has a pronounced effect on
the aging reaction and the phase composition, but not on the Zn content of the zincian malachite precursor. As a
hydroxycarbonate precursors was decoupled from the
-
Vis spectroscopy.
O catalysts, was
successfully formed from the amorphous starting material in all experiments by aging in
solutions with a composition near to the mother liquor under controlled co
-precipitation
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
34
conditions. As a function of different aging conditions, a variation of the Zn content in zincian
malachite between ca. 24 and 29% was observed despite the same nominal Zn-content in the
starting material of 30% indicating that a varying fraction of Zn was present in an undetected
phase “Zn↓” acting as a sink for Zn. Two mechanisms to approach the maximal Zn
incorporation into the zincian malachite catalyst precursor were observed: by direct co-
condensation of Cu
2+
and Zn
2+
into Zn-rich malachite, or by first simultaneous crystallization of
Cu-rich malachite and a transient Zn-storage phase, which in course of aging re-dissolved and
allows for later Zn-enrichment of malachite. The latter mechanism is favored at pH ≥ 7 in the
presence of Na
+
leading to crystallization of sodium zinc carbonate as Zn-storage phase. The
former mechanism was observed at 5 ≥ pH ≥ 6.5 and yields a higher Zn-incorporation into
zincian malachite. The radar plots shown in Figure 2-9 summarize the effects of pH,
temperature and alkali cation on the aging process. It can be seen that variation in pH changes
the aging mechanism, while variation of temperature (at pH 7) leads to gradual changes. Thus,
the acidity of the aging medium was identified as the most critical synthesis parameter to
determine the final Zn-content in zincian malachite. Interestingly, Zn incorporation is
independent of the crystallization mechanism. Even in the absence of Na
+
, suppressing the
transient crystallization of the sodium zinc carbonate storage phase, a lower degree of Zn
incorporation was observed in the final sample at pH 7, although the reaction was following the
direct co-condensation mechanism. The effect of individual synthesis parameters like
temperature or acidity during catalyst preparation can be better rationalized on basis of the
complex chemistry of precursor aging. They should be optimized to give a low amount of Zn↓
and a maximal Zn-substitution in malachite approaching the nominal Cu:Zn ratio of the
synthesis.
Acknowledgement
This paper has emerged from a joint research project “Next generation methanol synthesis
catalysts”, which was funded by the German Federal Ministry of Education and Research
(BMBF, FKZ 01RI0529). We acknowledge Edith Kitzelmann, Achim Klein-Hofmann, Gisela
Lorenz, and Doreen Steffen for their substantial support in the lab, Elena Antonova, Jing Wang
and Wolfgang Bensch for support with the EDXRD measurements, and HASYLAB (Hamburg,
Germany) for allocation of beamtime. Martin Muhler and Stefan Kaluza are acknowledged for
fruitful discussions. Robert Schlögl is greatly acknowledged for his continuous support.
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
35
2.5 References
[1] K. de Jong, Synthesis of Solid Catalysts, Wiley VCH., Weinheim, 2009.
[2] M. Kurtz, N. Bauer, C. Buscher, H. Wilmer, O. Hinrichsen, R. Becker, S. Rabe, K.
Merz, M. Driess, R. A. Fischer, M. Muhler, Catal. Lett. 2004, 92, 49-52.
[3] C. Baltes, S. Vukojevic, F. Schüth, J. Catal. 2008, 258, 334-344.
[4] B. Bems, M. Schur, A. Dassenoy, H. Junkes, D. Herein, R. Schlögl, Chem. Eur. J.
2003, 9, 2039-2052.
[5] I. Kasatkin, P. Kurr, B. Kniep, A. Trunschke, R. Schlögl, Angew. Chem. Int. Edit. 2007,
46, 7324-7327.
[6] R. G. Herman, K. Klier, G. W. Simmons, B. P. Finn, J. B. Bulko, T. P. Kobylinski, J.
Catal. 1979, 56, 407-429.
[7] J. C. Frost, Nature 1988, 334, 577-580.
[8] V. Ponec, Surf. Sci. 1992, 272, 111-117.
[9] Y. Kanai, T. Watanabe, T. Fujitani, M. Saito, J. Nakamura, T. Uchijima, Catal. Lett.
1994, 27, 67-78.
[10] N. Y. Topsoe, H. Topsoe, Top. Catal. 1999, 8, 267-270.
[11] J. D. Grunwaldt, A. M. Molenbroek, N. Y. Topsoe, H. Topsoe, B. S. Clausen, J. Catal.
2000, 194, 452-460.
[12] T. Ressler, B. L. Kniep, I. Kasatkin, R. Schlögl, Angew. Chem. Int. Ed. 2005, 44, 4704-
4707.
[13] M. Behrens, F. Studt, I. Kasatkin, S. Kühl, M. Hävecker, F. Abild-Pedersen, S. Zander,
F. Girgsdies, P. Kurr, B.-L. Kniep, M. Tovar, R. W. Fischer, J. K. Nørskov, R. Schlögl,
Science 2012, 336, 893-897.
[14] D. Waller, D. Stirling, F. S. Stone, M. S. Spencer, Faraday Discuss. 1989, 87, 107-120.
[15] S. Fujita, A. M. Satriyo, G. C. Shen, N. Takezawa, Catal. Lett. 1995, 34, 85-92.
[16] D. M. Whittle, A. A. Mirzaei, J. S. J. Hargreaves, R. W. Joyner, C. J. Kiely, S. H.
Taylor, G. J. Hutchings, Phys. Chem. Chem. Phys. 2002, 4, 5915-5920.
[17] B. L. Kniep, T. Ressler, A. Rabis, F. Girgsdies, M. Baenitz, F. Steglich, R. Schlögl,
Angew. Chem. Int. Edit. 2004, 43, 112-115.
[18] M. Behrens, J. Catal. 2009, 267, 24-29.
[19] A. M. Pollard, M. S. Spencer, R. G. Thomas, P. A. Williams, J. Holt, J. R. Jennings,
Appl. Catal. A 1992, 85, 1-11.
[20] M. Behrens, F. Girgsdies, Z. Anorg. Allg. Chem. 2010, 636, 919-927.
[21] M. Behrens, F. Girgsdies, A. Trunschke, R. Schlögl, Eur. J. Inorg. Chem. 2009, 1347-
1357.
[22] R. Kiebach, N. Pienack, M. E. Ordolff, F. Studt, W. Bensch, Chem. Mater. 2006, 18,
1196-1205.
[23] L. Engelke, M. Schaefer, F. Porsch, W. Bensch, Eur. J. Inorg. Chem. 2003, 506-513.
[24] N. Pienack, W. Bensch, Angew. Chem. Int. Edit. 2011, 50, 2014-2034.
[25] R. I. Walton, D. O'Hare, Chem. Comm. 2000, 2283-2291.
[26] A. Michailovski, J. D. Grunwaldt, A. Baiker, R. Kiebach, W. Bensch, G. R. Patzke,
Angew. Chem. Int. Edit. 2005, 44, 5643-5647.
[27] M. Behrens, R. Kiebach, J. Ophey, O. Riemenschneider, W. Bensch, Chem. Eur. J.
2006, 12, 6348-6355.
[28] F. Porsch, 3.155 ed., RTI GmbH, Paderborn, Germany, 2004.
[29] L. Engelke, M. Schaefer, M. Schur, W. Bensch, Chem. Mater. 2001, 13, 1383-1390.
[30] A. Coelho, 4.2 ed., Bruker AXS GmbH, Karlsruhe, Germany, 2003-2009.
[31] T. E. Gier, X. H. Bu, S. L. Wang, G. D. Stucky, J. Am. Chem. Soc. 1996, 118, 3039-
3040.
[32] F. Zigan, W. Joswig, H. D. Schuster, S. A. Mason, Z. Kristallogr. 1977, 145, 412-426.
Chapter 2: In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts
36
[33] S. Kaluza, M. Schroter, R. d'Alnoncourt, T. Reinecke, M. Muhler, Advanced Functional
Materials 2008, 18, 3670-3677.
[34] H. Jung, D.-R. Yang, O.-S. Joo, K.-D. Jung, B. Kor. Chem. Soc. 2010, 31, 1241-1246.
[35] S. Klokishner, M. Behrens, O. Reu, G. Tzolova-Müller, F. Girgsdies, A. Trunschke, R.
Schlögl, J. Phys. Chem. A 2011, 115, 9954-9968.
[36] J. L. Li, T. Inui, Appl. Catal. A 1996, 137, 105-117.
[37] M. Behrens, D. Brennecke, F. Girgsdies, S. Kißner, A. Trunschke, N. Nasrudin, S.
Zakaria, N. F. Idris, S. B. Abd Hamid, B. Kniep, R. Fischer, W. Busser, M. Muhler, R.
Schlögl, Appl. Catal. A 2011, 392, 93-102.
Chapter 2: Supplementary Information
37
Supplementary Information
Table S2-1: Aging parameters and internal sample numbers. The sample ID 0 refers to the unaged precursor.
ID pH T [K] A
+
in A
2
CO
3
Internal sample number
0 - - - 7005
1 5 333 Na
+
7037
2 6 333 Na
+
7057
3 6.5 333 Na
+
7064
4 7 333 Na
+
7045
5 7.5 333 Na
+
7036
6 8 333 Na
+
7038
7 7 323 Na
+
7044
8 7 343 Na
+
7043
9 7 333 K
+
7063
10 7 343 K
+
7058
Figure S2-1: Evolution of pH (red curve) with added Cu,Zn solution (green curve) and Na
2
CO
3
solution (blue curve)
during co-precipitation of the Cu,Zn (70:30) precursor in the continuous process. The slurry was continuously fed
into a spray-dryer.
Chapter 2:
Supplementary Information
38
Figure S2-2: Sc
hematic representation for the T
using an automated laboratory-
reactor. The co
continuously removing the “unaged” precursor (
with the same starting material (red arrow). The marked reflections in the lower right hand corner XRD pattern refer
to the aurichalcite by-
phase, (Cu,Zn)
(Cu,Zn)
2
(OH)
2
CO
3
.
Supplementary Information
hematic representation for the T
- and pH-
controlled precursor preparation from aqueous solutions
reactor. The co
-
precipitation and aging stages (right hand side) were decoupled by
continuously removing the “unaged” precursor (
left hand side) and subsequent aging studies at varying conditions
with the same starting material (red arrow). The marked reflections in the lower right hand corner XRD pattern refer
phase, (Cu,Zn)
5
(CO
3
)
2
(OH)
6
. All other reflection
are due to zincian malachite
controlled precursor preparation from aqueous solutions
precipitation and aging stages (right hand side) were decoupled by
left hand side) and subsequent aging studies at varying conditions
with the same starting material (red arrow). The marked reflections in the lower right hand corner XRD pattern refer
are due to zincian malachite
Figure S2-3:
Detailed experimental setup of
Hamburg, Germany.
Figure S2-4: Ex-situ
XRD patterns of all recovered sample (for
malachite and the characteristic peaks of aurichalcite are marked. The labeling refers to the entry
in the main article.
Chapter 2:
Supplementary Information
Detailed experimental setup of
in-situ
EDXRD reaction cell at the F3 beamline at HASYLAB,
XRD patterns of all recovered sample (for
labeling see Table 2-1).
T
malachite and the characteristic peaks of aurichalcite are marked. The labeling refers to the entry
Supplementary Information
39
EDXRD reaction cell at the F3 beamline at HASYLAB,
T
he 20-1 reflection of
malachite and the characteristic peaks of aurichalcite are marked. The labeling refers to the entry
number of Table 2-1
Chapter 2:
Supplementary Information
40
Figure S2-5:
Evolution of the FWHM of
simulated aging at pH 7, 323
K (ID
were detected, the FWHM of
the 20
sodium zinc carbonate phase. Thus, the shift
from both phases but rather of Zn incorporation into the zincian malachite phase.
Figure S2-6:
Integral intensity of selected EDXRD peaks of detected phases vs. aging time in Na
at different pH-
values. Zincian malachite (Cu,Zn)
zinc carbonate Na
2
Zn
3
(CO
3
)
4
·3H
2
O is represented by the 222 peak (red).
Supplementary Information
Evolution of the FWHM of
the 20-1 reflection
of zincian malachite from the EDXRD spectra during
K (ID
7). In all aging experiment where reflections of the sodium zinc intermediate
the 20
-1 reflection
of zincian malachite did not decrease during
sodium zinc carbonate phase. Thus, the shift
of the 20-1 reflection seems
not to be an effect of overlapping peaks
from both phases but rather of Zn incorporation into the zincian malachite phase.
Integral intensity of selected EDXRD peaks of detected phases vs. aging time in Na
values. Zincian malachite (Cu,Zn)
2
(OH)
2
(CO
3
) is represented by the 20-1
reflection
O is represented by the 222 peak (red).
of zincian malachite from the EDXRD spectra during
7). In all aging experiment where reflections of the sodium zinc intermediate
of zincian malachite did not decrease during
disappearance of the
not to be an effect of overlapping peaks
Integral intensity of selected EDXRD peaks of detected phases vs. aging time in Na
2
CO
3
at T = 333 K
reflection
(green), sodium
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
41
Chapter 3: Correlations between Preparation and Microstructure of
Cu/ZnO Catalysts for Methanol Synthesis – Influence of the pH value
during Synthesis of Cu,Zn Hydroxy Carbonates
Stefan Zander, Igor Kasatkin, Gisela Weinberg, Patrick Kurr, Benjamin Kniep, Malte Behrens.
Abstract
The co-precipitation of mixed Cu,Zn hydroxide carbonate precursors is the first step during
preparation of Cu/ZnO catalysts for methanol synthesis. The pH value influences the precursor
chemistry because it controls the precipitation behavior and also the subsequent aging process.
Application of different pH values in the range of pH 6.0-9.0 were chosen based on previous
results. To check for the reproducibility of the kinetically controlled co-precipitation process,
two sample series were prepared under identical conditions. Use of pH values ≥ 6.5 led to
higher phase fraction of zincian malachite at the expense of the undesired Zn-rich by-phase
aurichalcite. As a consequence, more Zn was inserted into zincian malachite which was verified
from the position of the 201
reflection by Rietveld refinement of the XRD patterns of the
precursors. Samples prepared at pH 7.5 and higher showed a split up signal of the 201
reflection
indicating two different zincian malachite phases with different Zn substitution. Samples
prepared at 6.0 ≤ pH ≤ 7.0 exhibited a better homogeneity of the Zn distribution within the
zincian malachite. After calcination, samples prepared at pH 6.0 showed the largest CuO
domain size. Reduced Cu/ZnO samples exhibited Cu surface areas in the range of 18 to
20 m
2
g
-1
. Only samples prepared at pH 8.5 showed a larger Cu surface area of around 25 m
2
g
-1
.
No simple correlation was found between the Cu surface area and any microstructural
characteristics of the samples. It is expected that the activity of the catalysts scales linearly with
the Cu surface area.
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
42
3.1 Introduction
Cu/ZnO/(Al
2
O
3
) catalysts are of major industrial interest as they have been successfully applied
in methanol synthesis for over 40 years. These catalysts are also used in methanol steam
reforming and water-gas-shift reaction. The synthesis route for preparing Cu/ZnO/(Al
2
O
3
)
catalysts follows a multi-step procedure including temperature- and pH-controlled co-
precipitation of aqueous Cu,Zn,Al nitrate solution with sodium carbonate solution as
precipitating agent. Subsequently, the precipitate is aged in the mother liquor, filtrated, washed
and dried to obtain the mixed metal hydroxy carbonate precursor phase. After forming the metal
oxides by calcination, reduction of CuO to metallic copper yields the active catalyst
[1]
.
All parameters applied in every single step of the catalyst preparation can influence the bulk and
surface structure and therewith the characteristics and activity of the resulting catalyst. This
phenomenon is also called the "chemical memory" and relates to the influence of early stage
parameters on the characteristics of the precursor phase and, finally, on the microstructure and
activity of the resulting catalyst
[2-3]
. Accordingly, the precipitation and aging conditions were
shown to have a significant impact on the final catalyst
[1, 4-13]
. Regarding the activity of the
resulting catalyst, co-precipitation of Cu,Zn,Al systems turned out to be most successful
applying pH 7 and a temperature of 343 K
[10-11]
. Baltes et al.
[11]
reported that variation of pH
values and temperature during the first step of the catalyst preparation for the industrially
relevant ternary system (Cu:Zn:Al = 60:30:10) yielded differences in Cu surface areas which
were found to be linearly correlated with the catalyst’s activity in methanol synthesis. While
their study was rather focused on the physico-chemical properties of the calcined and reduced
samples, we herein report a systematic study of pH variation on the properties of the hydroxy
carbonate precursors.
To reduce the complexity of ternary Cu,Zn,Al systems, often binary Cu,Zn functional model
systems are applied. Bems et al. investigated phase formation and thermal decomposition of
precursors with different Cu:Zn ratios and found Cu
2
(OH)
2
CO
3
(malachite) for pure Cu
samples, (Cu
1-x
Zn
x
)
2
(OH)
2
CO
3
(zincian malachite) with x < 0.3 and (Cu
1-y
Zn
y
)
5
(OH)
6
(CO
3
)
2
(aurichalcite) with y > 0.5 as the predominant precursor phases. In the XRD patterns, they
observed a shift of the 20
reflection of zincian malachite with increasing Zn content in this
phase
[3]
, which also was reported earlier by Porta et al.
[14]
. This shift can be correlated to the
accessible copper surface area of the final catalyst and to a first approximation also to the
catalytic activity
[15]
. Thus, the critical role of the precursor chemistry has been emphasized as
follows: To obtain highly active catalysts, zincian malachite is the relevant precursor phase
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
43
because it combines effects of particle morphology and Zn content better than aurichalcite. As a
result, a higher Cu surface area is obtained after meso- and nanostructuring during catalyst
synthesis. Nevertheless, the presence of aurichalcite indicates that the maximal Zn incorporation
into the zincian malachite phase is reached.
The critical step for precursor phase formation is the aging of the precipitate upon which it
crystallizes. In a recent in-situ study of the aging process, we identified pH of 6.5 and lower and
temperatures of 333-343 K to result in the highest Zn incorporation in the zincian malachite
phase for decoupled precipitation and aging in a down-scaled synthesis of a few milligrams
[13]
.
The aim of the present study was to elucidate the impact of the pH value during precipitation for
larger batch synthesis using a 2 L-volume reactor with yields in the 0.1 kg scale. Furthermore,
analysis of calcined and reduced samples should deliver correlations between synthesis pH and
microstructure, in particular test the proposed correlation of the position of the 20
reflection of
zincian malachite with the accessible copper surface area which is closely related to the catalytic
activity in methanol synthesis.
3.2 Experimental
3.2.1 Sample Preparation
Metal hydroxy carbonate precursors with fixed Cu:Zn ratio (70:30) were synthesized by co-
precipitation from acidic Cu,Zn nitrate solution and Na
2
CO
3
solution as basic precipitating agent
in an automated lab reactor (LabMax, Mettler Toledo) under controlled conditions like dosing,
stirring, temperature (338 K) and constant pH value. While all other parameters were fixed, the
pH value was varied in different experiments between 6.0 and 9.0 in steps of 0.5 pH units. The
range of pH was chosen because pH values lower than 6.0 lead to incomplete Zn precipitation,
pH values higher than 9.0 to oxolation of the precipitates to form CuO, which is undesired in
this stage of preparation
[9]
. The co-precipitate slurry was aged until a transient pH drop
[3, 15]
was observed and stirred for additional 30 min. Then, the slurry was filtered, the aged
precipitate was washed several times with water and spraydried (Niro minor mobile,
T
inlet
= 473 K, T
outlet
= 373 K). Calcination was carried out in static air at 603 K (2 K min
-1
) for
3 h. The high Cu:Zn ratio of 70:30 is typically applied in industrial catalyst preparation and
aims at a maximal incorporation of Zn into the zincian malachite precursor phase
[15]
. The
designations of the samples are given in Table S3-1.
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
44
3.2.2 Characterization
X-ray fluorescence spectroscopy (XRF) of the calcined samples was performed with powders or
after glassing with Li
2
B
4
O
7
in a Bruker S4 Pioneer X-ray spectrometer.
X-ray diffraction (XRD) was applied to the catalyst precursors and calcined samples. The
samples were measured on a STOE STADI P transmission diffractometer equipped with a
primary focusing Ge monochromator (Cu K
α1
radiation) and a linear position sensitive detector
(moving mode, step size 0.1 °, counting time 10 s/step, resolution 0.01 °, total accumulation
time 634 s). The samples were mounted in the form of a clamped sandwich of small amounts of
powder fixed with a small amount of grease between two layers of thin polyacetate film. The
phase composition was determined by full pattern refinement in the 2θ range 4-80 ° according
to the Rietveld method using the TOPAS software
[16]
and crystal structure data from the ICSD
database.
Specific surface areas were determined by N
2
physisorption in a Quantachrome Autosorb-6
machine after degassing the samples at 353 K for 2 h. Isotherms were recorded at liquid
nitrogen temperature and evaluated according to the BET method. The recorded isotherms of all
precursors and calcined samples featured the characteristics of mesoporous substances (not
shown).
Thermogravimetric experiments (TGMS) were done on a NETZSCH Jupiter thermobalance in
flowing air at a heating rate of 2 K min
-1
. The gas evolution was measured with a quadrupole
mass spectrometer (Pfeiffer Vacuum, Omnistar).
Scanning electron microscopy (SEM) images were taken in a Hitachi S-4800 field emission gun
(FEG) system. Transmission electron microscopy (TEM) was performed with a Philips
CM200FEG microscope operated at 200 kV and equipped with an EDX spectrometer. For TEM
investigation, the samples were reduced up to a temperature of 523 K and transferred to the
microscope in inert atmosphere. The coefficient of spherical aberration was Cs = 1.35 mm and
the information limit was better than 0.18 nm. High-resolution images with a pixel size of
0.016 nm were acquired at the magnification of 1083000x with a CCD camera, and selected
areas were processed to obtain power spectra (square of the Fourier transform of the image),
which were used for measuring interplanar distances and angles (accuracy ± 1% and ± 0.5 deg,
correspondingly) for phase identification. Projected areas have been measured and equivalent
diameters calculated for 1500-3000 Cu particles in each sample. In all cases the values of
standard error of the mean diameter were ≤ 0.1 nm. Frequency distributions of the particle sizes
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
45
fitted well to Lognormal functions. EDX analyses were performed for 5-15 larger aggregates
containing at least several hundred particles in each sample.
Temperature programmed reduction (TPR) was performed with around 40 mg of each sample in
a glass reactor, fixed by means of quartz wool plugs. The reduction was carried out in a CE
instruments TPDRO 1100 machine with 80 mL min
−1
5% H
2
in Ar up to a temperature of 623 K
(6 K min
−1
). The K-values according to Monti and Baiker
[17]
were 100-120 s, the P-values
according to Caballero
[18]
10-12 K. The reduction progress was followed with an internal
thermal conductivity detector. Analysis was performed with regard to the temperature with the
highest H
2
consumption (T
max
) and the total H
2
consumption with respect to the CuO content in
the sample (compared with a pure CuO reference). In the following, the term "reducibility" is
used for the latter feature. The CuO content of the samples was derived from XRF data with the
assumption that only CuO and ZnO were present and under neglect of residual carbonate
species (see below).
The copper surface area was determined by applying N
2
O reactive frontal chromatography
(N
2
O-RFC) based on the method proposed by Chinchen et al.
[19]
. Around 100 mg of a sieve
fraction (100-200 µm) of each sample were placed in a stainless steel U-tube reactor and fixed
by means of quartz wool plugs. The prior reduction was carried out in the same device and
conditions as for TPR, but only up to a temperature of 523 K and with a holding time of 30 min.
The reduction progress was additionally followed with a quadrupole mass spectrometer (Pfeiffer
Vacuum, Omnistar). After cooling down to 303 K, the catalyst has been flushed for 30 min in
pure Ar and 15 min in pure He in order to achieve an adsorbate-free reduced catalyst surface.
N
2
O-RFC was performed with 10 mL min
−1
1% N
2
O in He, at which the N
2
O reacts
quantitatively with the Cu surface atoms forming gas-phase N
2
. The specific Cu metal surface
area has been calculated from the formed amount of N
2
using a value of 1.47*10
19
atoms per m
2
for the mean Cu surface atom density. The error of the specific Cu surface area is about
± 1 m
2
g
−1
.
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
46
3.3 Results and Discussion
Two identical sets of samples were prepared to check for the reproducibility of our preparation
and estimate the role of possible “batch effects” on the results. All characterizations were
performed for both sets of samples unless otherwise noted. The designation of the binary Cu,Zn
samples was chosen with regard to pH value of the precipitation, respectively. CZ6.0 means that
the sample(s) were prepared at pH 6.0. For reasons of visualization, the shown values in the
plots and tables include values of the original and the reproduction experiment as well as the
average values. The endings of the given error bars represent the real values obtained for the
two samples for each precipitation pH value and give information about the reproducibility of
the sample preparation.
3.3.1 Precipitation and aging
A typical evolution of pH value and turbidity during precursor preparation is shown in Figure
3-1 and is similar to the data described in ref.
[15]
in detail. The pH value is kept at a constant
value (here: 8.5) during simultaneous dosing of metal solution and precipitation agent. After the
end of dosing (t
aging
= 0), the free aging begins and is characterized by an uncontrolled evolution
of the pH value. The pH drop (t
aging
= 14 min) is accompanied by an increase in the turbidity and
a change of the color of the slurry from blue to green. The initially amorphous precipitate is
transformed into a crystalline product
[15]
.
Having a closer look to the recorded pH traces of all samples during the aging of the freshly
precipitated solids (after dosing) some trends can be observed within the whole pH series. First,
we inspect the duration until the local minimum of the pH drop (Figure 3-2a). For CZ6.0 the
averaged time amounts 142 minutes and decreases down to 12 minutes for CZ8.0. After this
minimum, the value increases again up to 33 minutes (CZ9.0). An earlier pH drop with
increasing preparation pH values was also observed in our former study where we decoupled
precipitation and aging and investigated the aging process at different pH values from 5.0 to 8.0
[13]
. The error bars, i.e. the reproducibility within two experiments, of the obtained values in
Figure 3-2a are remarkably low. Only CZ7.0 showed large error bars indicating differences in
the crystallization kinetics between the two runs.
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
47
Figure 3-1: Evolution of pH (black curve), turbidity (grey curve) during pH-controlled (pH 8.5) dosing of acidic
Cu,Zn solution (dotted curve) and basic Na
2
CO
3
solution (dashed curve) and subsequent free aging of a Cu,Zn
(70:30) precursor.
Figure 3-2: Results from the recorded data during precipitation and aging in dependence on the precipitation pH
value: time of the occurrence of the pH drop (a), depth of the pH drop (b), ratio of dosed moles of Na
2
CO
3
and metals
(Cu+Zn) (c). Shown values are representing average values, endings of the error bars represent the real values
obtained for the two samples for each precipitation pH value.
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
48
As a second feature, the depth of the pH drop was evaluated as a hint, how pronounced the
chemical transformations during the pH drop are (Figure 3-2b). Again, the largest deviation
between the two experiments was observed for the synthesis conducted at pH 7.0. With
exception of pH 6 and the one experiment at pH 7, similar values without a general trend were
observed.
Another characteristic is the ratio of added CO
32-
and metal (Cu
2+
+ Zn
2+
) ions. While the
reproducibility is good for all experiments, the amount of CO
32-
ions in the reaction mixture is
generally higher for higher precipitation pH values. This is expected, because Na
2
CO
3
is used to
regulate the pH value during precipitation whereas the amount of added Cu
2+
and Zn
2+
ions
stays the same for every preparation (Figure 3-2c). Nevertheless, two different regions can be
identified. In the precipitation range of pH 6.0 to pH 7.0, the consumption of Na
2
CO
3
is
relatively constant, whereas a linear increase can be observed starting at pH 7.5.
The kinetics of crystallization of the hydroxide carbonate precursors will depend on the CO
32-
and OH
–
concentrations and this dependency is reflected in changes in the time period until
occurrence of the pH drop and its depth. The details and elementary steps of this event are not
easily probed experimentally and no obvious correlation can be established between the
observed changes in pH and specific chemical reactions. It is known that the pH drop can be
assigned to modifications in the phase composition by dissolution and re-precipitation leading to
crystallization of the initially amorphous precipitate zincian georgeite to crystalline zincian
malachite and aurichalcite
[3-4, 8]
. Effectively, such a reaction will be related to a transient
exchange of anions (and cations) between the solid precursor and the solution leading to the
observed characteristic pH traces.
To check for a complete precipitation of all Cu,Zn species, the Cu,Zn oxides (calcined samples)
were subjected to elemental analysis by XRF. The experimental values (Table S3-1) were in
good agreement with the nominal Zn content of 30% (x
Cu
+ x
Zn
= 100%) from the initial Cu,Zn
solution, with two exceptions: For CZ7.5, there was a low, but significant increase of the Zn
content. It should be noted, that this behavior was reproducible but was not found in the results
of Baltes et al.
[11]
for the Cu,Zn,Al system. CZ9.0 showed a loss of Zn. However, regarding the
large error in this case, reproducibility seems to be limited at this pH value. These minor
variations are tentatively related to formation of soluble species at the respective pH or by loss
of Cu- or Zn-enriched materials during recovering the samples, e.g. due to preferred sticking to
the walls of the glass reactor.
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
49
3.3.2 Precursor and calcined materials
Figure 3-3 shows the XRD patterns of one set of precursors which is also qualitatively
representing the set of reproduced samples. As expected, aurichalcite and zincian malachite
were identified as the only crystalline phases for all precursor samples.
Figure 3-3: XRD patterns of one selected precursor sample for each preparation pH value. The well-resolved peaks
of aurichalcite are marked with a star. All other reflections can be assigned to zincian malachite or to an overlap of
peaks from both phases.
XRD patterns of all precursor samples and calcined samples were subjected to Rietveld
refinement for phase identification and determination of composition and crystallite domain
sizes. As a representative example, the refinement result of a sample prepared at pH 7.5 is
depicted in Figure 3-4a. It is noted that the accurate determination of the exact weight fractions
in the phase mixture is difficult due to the low amounts of aurichalcite, the generally low
crystallinity and the high noise of the XRD patterns. The absolute values of individual samples
depend on the fitting constraints and have to be compared with care. However, the general trend
seen within the series of samples are regarded as reliable, as the used fitting constraints have
been equal for all samples. Depending on the preparation pH, some fits in the region between 31
and 33 ° 2θ were dissatisfying and the corresponding 201
reflection signal of zincian malachite
appeared to be split up into two peaks (inset, Figure 3-4a). Since the position of the 201
reflection is a measure for the degree of Zn incorporation in the zincian malachite phase,
apparently two differently Zn-substituted zincian malachite phases were present. This double
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
50
peak can already be seen in Figure 3-3 for CZ7.5 and CZ8.0. A Rietveld refinement using two
independent zincian malachite phases (and aurichalcite) led to a better agreement of the
experimental data with the simulated data in the mentioned region (Figure 3-4b). As reported in
ref.
[13]
, where the aging process was investigated in detail, two different mechanisms can lead to
the formation of zincian malachite: 1) direct co-condensation of Cu
2+
and Zn
2+
into a Zn-rich
malachite; 2) first simultaneous crystallization of Cu-rich malachite and a transient Zn-storage
phase, which in course of aging re-dissolves and allows for later Zn-enrichment of malachite.
The former mechanism was observed at 5 ≤ pH ≤ 6.5, resulting in a higher Zn-incorporation
into zincian malachite. The latter mechanism was found to be favored at 7 ≤ pH ≤ 8 in the
presence of Na
+
leading to crystallization of sodium zinc carbonate as Zn-storage phase. The
observation of two different zincian malachite phases in the precursor material recovered at pH
7.5 and pH 8.0 indicated that in our experiments both mechanisms may operate simultaneously.
The Zn-richer component was presumably formed via mechanism 1), while the Zn-poorer
fraction of zincian malachite has formed via mechanism 2).
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
51
Figure 3-4: Rietveld refinement for the XRD pattern of a precursor sample prepared at pH 7.5: With one zincian
malachite phase (a); with two zincian malachite phases (b); experimental data (black), total calculated curve (red),
background (light grey), difference curve (grey), calculated pattern zincian malachite 1 (orange curve), calculated
pattern zincian malachite 2 (green curve), calculated pattern aurichalcite (blue curve). The thick marks indicate the
positions of the Bragg reflections.
In Figure 3-5a (black symbols), the fractions of aurichalcite in the precursor samples are shown.
CZ6.0 contained most aurichalcite (30 wt%) at the expense of zincian malachite. Precipitation
pH values of 6.0 and higher led to a fraction of 5 to 10 wt% aurichalcite but no distinct
minimum was detected. The averaged BET surface area values (Figure 3-5a, grey symbols) for
precursors lay in the range of 49 m
2
g
-1
(CZ7.5) to 127 m
2
g
-1
(CZ6.0) again without an obvious
trend with variation in synthesis pH value.
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
52
Figure 3-5: Results of XRD full pattern refinement and BET surface areas of Cu,Zn hydroxy carbonate precursors
and calcined samples in dependence on the precipitation pH value. The zincian malachite phase “zM1” and “zM2”
refer to the refinement with two different zincian malachite phases: weight fraction of the aurichalcite phase (black)
and BET surface areas (grey) of the precursors (a); d-spacing of the 201
reflection of zM1 (red) and zM2 (green) (b);
difference between the d
values of both zincian malachite phases zM1 and zM2 (c); Zn-content (x
Cu
+ x
Zn
=
100%) in zM1 (red), zM2 (green) and overall weighted average Zn content in both zincian malachite phases (black)
(d); domain size of CuO (black) and BET surface areas (grey) of the calcined samples (e). Shown values are
representing average values, endings of the error bars represent the real values obtained for the two samples for each
precipitation pH value. Lines are guides for the eyes.
In Figure 3-5b, the d
values are shown. These were obtained resulting from full pattern
refinement using two different zincian malachite phases for all samples, even those which can
be satisfactorily fitted with only one zincian malachite phase, to allow comparability. The d
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
53
value can be used as a quantitative measure of the incorporation of non Jahn-Teller-distorted
Zn
2+
into this phase and was first reported by Porta et al.
[14]
. Low d
values correspond to
high incorporation. The d
value of pure malachite, Cu
2
(OH)
2
CO
3
, is about 2.863 Å. The Zn-
richer phase is labeled Zincian malachite 1 (zM1, red symbols) and exhibits a lower d
value
than that of zincian malachite 2 (zM2, green symbols). For CZ6.0 and CZ6.5 the d
values of
zM1 and zM2 are almost identical confirming that indeed only one zincian malachite phase with
a homogeneous Zn content is present. For pH values of 7.0 and higher, the two curves separate
indicating inhomogeneous Zn distribution within the zincian malachite. The lowest d
values
for zM1 were obtained for CZ7.5 and CZ8.0 (2.75 Å).
The difference of the both obtained d
values (zM1 and zM2) gives an insight into the
homogeneity of the Zn distribution within the zincian malachite phases (Figure 3-5c).
Homogeneous distribution, i.e. a single zincian malachite phase zM1 = zM2, was found for
CZ6.0, where the difference is quite small, and for CZ6.5, where it is exactly zero. For CZ7.0,
the difference is still quite small, but higher pH values lead to larger differences caused by
inhomogeneous Zn distribution. Remarkably, the reproducibility of this feature is very good.
The error bars are small and in some cases, reproduction delivers again the same values. Low
reproducibility was again detected for CZ9.0. Inhomogeneity of the particle morphology and
fluctuation of the local Cu:Zn ratio was also observed by SEM-EDX investigation on selected
samples (for details see supporting information).
The d
value can be directly converted into the Zn content (x
Zn
+ x
Cu
= 100%) in the zincian
malachite phase according to the correlation of the Cu,Zn system shown in
[13, 20-21]
. The results
are plotted in Figure 3-5d and extreme values are obtained for zM2 (13%, CZ7.5, green
symbols) and zM1 (32%, CZ8.0, red symbols). It is noted that the latter value exceeds the
proposed maximum of 28% synthetic zincian malachite samples prepared at pH 7.0
[21]
.
Including the phase fractions of zM1 and zM2, an overall weighted average Zn content in the
malachite phases was calculated (Figure 3-5d; black symbols). These averaged Zn contents are
below the nominal Zn content of 30% used during synthesis and the XRF results. The remainder
of Zn is assumed to be present in the aurichalcite by-phase. The lowest average value of 16%
was found for CZ6.0, the sample that also shows the highest aurichalcite fraction. All other
samples exhibited average Zn contents in the zincian malachite component between 23%
(CZ6.5) and 26% (CZ8.0).
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
54
To interpret the data obtained from XRD pattern refinement, it is important to recapitulate that
zincian malachite is the relevant precursor phase to obtain highly active catalysts
[15]
. These
results show that not only the average Zn incorporation into this catalyst precursor phase plays a
decisive role, but that at higher precipitation pH also the homogeneity of the Zn distribution
within the zincian malachite phase fraction has to be considered. Both features can influence the
dispersion of CuO particles, separated by ZnO particles, after calcination. Higher Zn
incorporation into the zincian malachite should increase the inter-dispersion of ZnO and CuO
particles. A more homogeneous Zn distribution in the zincian malachite should cause a narrower
statistical distribution of the CuO particle size.
Under these assumptions, the samples prepared at pH 6.0 can be estimated to result in poor
catalysts, because they show large fractions of aurichalcite and therefore contain little Zn in the
zincian malachite phase, although it is homogeneously distributed. The preparation at pH 6.5
and higher results in an increased Zn incorporation. Samples prepared at pH 6.5 and 7.0 seem to
be most suitable because they additionally show a good homogeneity, in contrast to pH 7.5 and
higher. In the latter case, the nanostructuring of the CuO after calcination will proceed probably
not equally efficiently for both the zM1 and the zM2 phases, while there might be a
compensating effect of the Zn-richer domains for the Zn-poorer fraction of the catalyst. With
these considerations in mind, the calcined samples have been investigated to shed more light on
this issue.
The process of calcination has been monitored by TGA-EGA and the results of this study are
reported as supporting information. Due to the similar phase contrast of ZnO and CuO,
statistical evaluation of TEM data of the calcined samples is not easy and here we use the
domain size of the CuO crystallites as determined from XRD as a proxy for the particle size.
“Domain (crystallite) size” values from XRD should not be mistaken for the real “particle size”
as observed (in projection) in TEM images, which is usually higher because particles may
consist of several domains. This has been shown in a detailed XRD and TEM study for reduced
Cu/ZnO/Al
2
O
3
catalysts by Kasatkin et al.
[22]
. However, in nanostructured samples, where the
particle size is in the low nm-range and each particle contains several hundreds or thousands of
unit cells, there usually is a clear relation between XRD and TEM size, because the number of
domain boundaries in one particle is limited by geometrical reasons
[23]
. Typically a reliable
correlation of XRD and TEM size is observed in such materials.
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
55
Rietveld refinement of the XRD patterns of the calcined samples (not shown) revealed CuO as
the main phase and only small amounts (up to 2 wt%) of crystalline ZnO indicating that most of
the ZnO is amorphous or finely dispersed. The domain size of CuO is shown in Figure 3-5e
(black symbols). CZ6.0 exhibited the largest domain size (5.5 nm) as expected from the poor
inter-dispersion of CuO and ZnO as a result of low Zn-incorporation into the zincian malachite
phase. All other values are very similar around ≈ 4 nm and the differences are probably not
significant with regard to the error bars. The difference between CZ6.0 and the other samples
5.5 and 4 nm seems to be small but a simple geometric calculation reveals that the surface area
of spherical particles increases from 100% (5.5 nm) to 137.5% (4 nm) at a constant total mass.
The BET surface areas of the calcined samples (Figure 3-5e, grey symbols) exhibited somewhat
higher values compared to the precursors (Figure 3-5b, grey line) with the exception of CZ6.0,
which cannot undergo efficient nanostructuring during calcinations due to a lack of Zn in the
zincian malachite precursor phase. All other samples increased their surface area after thermal
decomposition. This effect was most prominent for CZ6.5 and CZ7.5, where the surface area
was increased by around 75%. The BET values are in a narrower range of 79 m
2
g
-1
(CZ6.0) to
105 m
2
g
-1
(CZ6.5, CZ8.0 and CZ8.5) compared to the precursor materials. They were in line
with the double maximum around pH 6.5 and 8.5 found by Baltes et al. for ternary
CuO/ZnO/Al
2
O
3
samples (Fig.3 for T = 338 K = 65 °C in ref.
[11]
). An inverse correlation is
observed between BET surface areas and CuO domain size. This is expected because small CuO
domains are a result of better nanostructuring which leads to higher BET surface areas.
To conclude this part, it was shown that the pH value of synthesis has only a low impact on the
phase composition regarding zincian malachite and aurichalcite for pH ≥ 6.5 and on the Zn
incorporation into the former phase. However, a pronounced effect on the Zn distribution in the
zincian malachite precursor was observed. Inhomogeneous Zn distributions with a Zn-poorer
and a Zn-richer zincian malachite component were observed for pH ≥ 7.5. The pH range
typically applied for catalyst preparation of 6.5 – 7.0 yields homogeneous zincian malachite
precursors with a high and uniform degree of Zn incorporation. An effect of the Zn
incorporation on the ZnO-CuO inter-dispersion after calcinations and on the microstructural
homogeneity of the resulting catalysts is expected and will be in the focus of the next section of
this work.
Chapter 3:
Correlations between
Synthesis –
Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
56
3.3.3
Reduction and reduced samples
The reduction behavior of the calcined Cu,Zn oxide samples was investigated by temperature
programmed reduction (TPR) in hydrogen and the results are reported as supporting
information.
TEM investigatio
n was done on one reduced sample on CZ6.5 and CZ9.0, representing the two
precursor families with homogeneous and inhomogeneous Zn distribution in the zincian
malachite precursor. CZ6.5 showed
particle size of around 11.1
nm (
from TEM-
EDX at different spots, the Cu:Zn ratio was 69.8:30.2. The low standard deviation
(2.0) shows, that the sample was quite homogeneous and the composition is in reasonable
agreement with the findings from XRF o
Figure 3-6:
Transmission electron microscopy images of reduced Cu,Zn sample prepared at pH
(c,d) showing the typical arrangement of Cu particles (a), the
sintering (c) and fused Cu particles after electron beam sintering (d).
Correlations between
Preparation and
Microstructure of Cu/ZnO Catalysts for Methanol
Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
Reduction and reduced samples
The reduction behavior of the calcined Cu,Zn oxide samples was investigated by temperature
programmed reduction (TPR) in hydrogen and the results are reported as supporting
n was done on one reduced sample on CZ6.5 and CZ9.0, representing the two
precursor families with homogeneous and inhomogeneous Zn distribution in the zincian
malachite precursor. CZ6.5 showed
round shaped Cu particles (Figure 3-6
a) and a mean Cu
nm (
Figure 3-6b). Accord
ing to the local elemental
EDX at different spots, the Cu:Zn ratio was 69.8:30.2. The low standard deviation
(2.0) shows, that the sample was quite homogeneous and the composition is in reasonable
agreement with the findings from XRF o
f the calcined sample.
Transmission electron microscopy images of reduced Cu,Zn sample prepared at pH
(c,d) showing the typical arrangement of Cu particles (a), the
particle size distribution (b), areas before electron beam
sintering (c) and fused Cu particles after electron beam sintering (d).
Microstructure of Cu/ZnO Catalysts for Methanol
The reduction behavior of the calcined Cu,Zn oxide samples was investigated by temperature
programmed reduction (TPR) in hydrogen and the results are reported as supporting
n was done on one reduced sample on CZ6.5 and CZ9.0, representing the two
precursor families with homogeneous and inhomogeneous Zn distribution in the zincian
a) and a mean Cu
ing to the local elemental
composition
EDX at different spots, the Cu:Zn ratio was 69.8:30.2. The low standard deviation
(2.0) shows, that the sample was quite homogeneous and the composition is in reasonable
Transmission electron microscopy images of reduced Cu,Zn sample prepared at pH
6.5 (a,b) and pH 9.0
particle size distribution (b), areas before electron beam
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
57
CZ9.0 turned out to be very unstable during the measurement. Although some images at lowest
magnification were taken (Figure 3-6c), rapid partial or complete fusing of the Cu particles was
observed for many areas. This effect was worse, when the electron beam was concentrated on
them at higher magnification (Figure 3-6d). ZnO had a stabilizing effect against sintering
because areas with high Cu:Zn ratio (76:24) were less stable than areas with low ratio (58:42).
The former can be assigned to ex zincian malachite domains, the latter to ex aurichalcite. In the
Zn rich stable area, the Cu particle size was 10.1 nm, but probably this value is not
representative for the whole sample. Figure 3-6c shows that in agreement with the observation
of Baltes et al.
[11]
, that the Cu/ZnO aggregates still exhibit the pseudo-morphology of the
zincian malachite precursor needles.
The observed differences in microstructural homogeneity are in agreement with the result of the
precursor material analysis. Despite the presence of a similar amount of aurichalcite in this
sample, at pH 6.5 a homogeneous microstructure is observed. The major fraction of the material
stems from the zincian malachite precursor and the local compositions are similar to those
determined for this precursor phase by XRD. These domains exhibit a fine distribution of Cu
and ZnO particles with a unimodal Cu particle size distribution. In case of the sample prepared
at pH 9, the ex-zincian malachite domains are unstable in the electron beam, which is tentatively
related to the very Cu-rich zM2 phase, which cannot offer much stabilization of the Cu
nanoparticles against sintering by inter-dispersion with ZnO.
Cu surface area determination of the reduced catalysts with N
2
O-RFC revealed a maximum
averaged value of around 25 m
2
g
calc-1
for CZ8.5 (Figure 3-7). All other pH values led to Cu
surface areas in a small region between 18 and 20 m
2
g
calc-1
. CZ6.0 had the largest CuO domain
size and showed the lowest Cu surface area. This can be explained with the large amount of the
aurichalcite by-phase in this sample, which acts as a sink to Zn, which is in turn not available as
a stabilizer against sintering in the ex-zincian malachite domains. It was surprising that CZ8.5
has a Cu surface area completely different from CZ8.0 or CZ7.0. The observed clear maximum
in copper surface area for the sample CZ8.5 came as a surprise as it was not predictable from
any characteristics of the precursor or the calcined material. However, because of the
inhomogeneity of this sample and the complexity of possible compensating effects on the
macroscopic observables of the different areas in the materials possessing a different
microstructure, it will be very difficult to find the reason for the maximal copper surface area of
this sample on basis of the available data.
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
58
Figure 3-7: Cu surface areas of reduced Cu,Zn samples with respect to the calcined sample mass. Shown values are
representing average values, endings of the error bars represent the real values obtained for the two samples for each
precipitation pH value. The dotted line is a guide for the eyes.
Because the Cu surface area is known to scale linearly with the activity in methanol synthesis
within the same Cu,Zn catalytic system
[24]
, the highest activity is expected for samples prepared
at pH 8.5. Accordingly, the intrinsic activities, which are a measure for the density of
catalytically active sites on the surface of the Cu particles, are expected to be similar for all
samples.
Literature data from ref.
[10-11]
revealed precipitation pH values around 7 to yield the largest Cu
surface areas and activity in methanol synthesis. However, this data was obtained for Cu,Zn,Al
systems which might behave different from the binary Cu,Zn system applied in this study.
3.4 Conclusions
It was shown that the variation of the pH value during the precipitation of Cu,Zn hydroxy
carbonate precursors (Cu:Zn = 70:30) had an influence on the subsequent aging process and also
on the ratio of the obtained precursor phases. Rietveld analysis was performed on the XRD
pattern of the precursors. For pH 6.0, large fractions of the by-phase aurichalcite were found
beyond zincian malachite as the main phase. Higher pH values decreased the aurichalcite
fraction, with the consequence, that more Zn was introduced into the zincian malachite phase,
visible from the shifted position of the 201
reflection. Samples prepared at pH 7.5 and higher
showed a split up signal of the 201
reflection indicating two clearly different Zn substituted
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
59
zincian malachite phases. Samples prepared at 6.0 ≤ pH ≤ 7.0 showed a better homogeneity of
the Zn distribution within the zincian malachite. As a consequence of the Zn incorporation, the
calcined samples prepared at pH 6.0 showed the largest CuO domain size. Cu surface areas of
the reduced Cu/ZnO catalysts revealed similar values in the region of 18 to 20 m
2
g
-1
. Only
samples prepared at pH 8.5 showed a larger Cu surface area of around 25 m
2
g
-1
. Activity in
methanol synthesis was not measured but is expected to scale linearly with the Cu surface area.
This study has shown the complexity of a catalyst synthesis by co-precipitation. The properties
of the precursor materials obtained by aging of the co-precipitate have decisive influence on the
structural properties and performance of the final catalysts. Unfortunately, directly tracking
back the catalytic performance to the synthesis pH in a simple synthesis parameter–structure–
performance relationship was found to be very complex as variation of the parameter pH
induced numerous changes in the precursor material that lead to different and partially
compensating effects for the resulting catalyst.
Acknowledgement
This paper has emerged from a joint research project “Next generation methanol Frank
Girgsdies (help with XRD pattern analysis), Edith Kitzelmann (XRD measurements), Achim
Klein-Hoffmann and Olaf Timpe (XRF) and Gisela Lorenz (BET measurements) are
acknowledged. Financial support was given by the German Federal Ministry of Education and
Research (BMBF, FKZ 01RI0529, 2005-2008) and the STMWFK (NW-0810-0002, since
2010). Robert Schlögl is greatly acknowledged for valuable discussions and his continuous
support.
Chapter 3: Correlations between Preparation and Microstructure of Cu/ZnO Catalysts for Methanol
Synthesis – Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates
60
3.5 References
[1] D. Waller, D. Stirling, F. S. Stone, M. S. Spencer, Faraday Discuss. 1989, 87, 107-120.
[2] M. S. Spencer, Catal. Lett. 2000, 66, 255-257.
[3] B. Bems, M. Schur, A. Dassenoy, H. Junkes, D. Herein, R. Schlögl, Chem. Eur. J.
2003, 9, 2039-2052.
[4] A. M. Pollard, M. S. Spencer, R. G. Thomas, P. A. Williams, J. Holt, J. R. Jennings,
Appl. Catal. A 1992, 85, 1-11.
[5] E. N. Muhamad, R. Irmawati, Y. H. Tautiq-Yap, A. H. Abdullah, B. L. Kniep, F.
Girgsdies, T. Ressler, Catal. Today 2008, 131, 118-124.
[6] H. Jung, D.-R. Yang, O.-S. Joo, K.-D. Jung, B. Kor. Chem. Soc. 2010, 31, 1241-1246.
[7] S. Kaluza, M. Behrens, N. Schiefenhoevel, B. Kniep, R. Fischer, R. Schlögl, M.
Muhler, Chem. Cat. Chem. 2011, 3, 189-199.
[8] S. Fujita, A. M. Satriyo, G. C. Shen, N. Takezawa, Catal. Lett. 1995, 34, 85-92.
[9] M. Behrens, D. Brennecke, F. Girgsdies, S. Kißner, A. Trunschke, N. Nasrudin, S.
Zakaria, N. F. Idris, S. B. Abd Hamid, B. Kniep, R. Fischer, W. Busser, M. Muhler, R.
Schlögl, Appl. Catal. A 2011, 392, 93-102.
[10] J. L. Li, T. Inui, Appl. Catal. A 1996, 137, 105-117.
[11] C. Baltes, S. Vukojevic, F. Schüth, J. Catal. 2008, 258, 334-344.
[12] J. S. Campbell, Ind. Eng. Chem. Process Des. Dev. 1970, 9, 588-&.
[13] S. Zander, B. Seidlhofer, M. Behrens, Dalton Trans. 2012, 41, 13413-13422.
[14] P. Porta, S. Derossi, G. Ferraris, M. Lojacono, G. Minelli, G. Moretti, J. Catal. 1988,
109, 367-377.
[15] M. Behrens, J. Catal. 2009, 267, 24-29.
[16] A. Coelho, 4.2 ed., Bruker AXS GmbH, Karlsruhe, Germany, 2003-2009.
[17] D. A. M. Monti, A. Baiker, J. Catal. 1983, 83, 323-335.
[18] P. Malet, A. Caballero, J. Chem. Soc. Faraday T. 1 1988, 84, 2369-2375.
[19] G. C. Chinchen, K. C. Waugh, D. A. Whan, Appl. Catal. 1986, 25, 101-107.
[20] M. Behrens, F. Girgsdies, Z. Anorg. Allg. Chem. 2010, 636, 919-927.
[21] M. Behrens, F. Girgsdies, A. Trunschke, R. Schlögl, Eur. J. Inorg. Chem. 2009, 1347-
1357.
[22] I. Kasatkin, P. Kurr, B. Kniep, A. Trunschke, R. Schlögl, Angew. Chem. Int. Edit. 2007,
46, 7324-7327.
[23] T. Ungar, J. Gubicza, G. Ribarik, A. Borbely, J. Appl. Cryst. 2001, 34, 298-310.
[24] M. Kurtz, H. Wilmer, T. Genger, O. Hinrichsen, M. Muhler, Catal. Lett. 2003, 86, 77-
80.
Chapter 3: Supplementary information
61
Supplementary Information
Table S3-1: Denotation of samples according to their precipitation pH value. Averaged Zn content
(x
Zn
+ x
Cu
= 100%) of the calcined samples from XRF. Listed values are representing average values, errors given in
brackets represent the real values obtained for the two samples for each precipitation pH value.
Label Precipitation
pH value Precursor sample
number (internal) Calcined sample
number (internal) Zn content
[%] T
max[a]
[K]
Reducibility
[b]
[%]
CZ6.0 6.0 10785 10786 29.5 (± 0.3) 473 (± 1) 102 (± 1)
10787 10788
CZ6.5 6.5 7399 7400 29.2 (± 0.3) 477 (± 4) 99 (± 2)
10084 10085
CZ7.0 7.0 7192 7193 29.2 (± 0.2) 476 (± 2) 95 (± 3)
10783 10784
CZ7.5 7.5 7852 7853 33.8 (± 0.3) 475 (± 0) 104 (± 2)
9385 9386
CZ8.0 8.0 7282 7283 29.3 (± 0.3) 479 (± 3) 97 (± 1)
7632 7633
CZ8.5 8.5 7938 7939 29.7 (± 0.2) 475 (± 0) 103 (± 1)
10147 10148
CZ9.0 9.0 6501 6521 27.5 (± 2.1) 475 (± 1) 97 (± 1)
10086 10087
[a] Temperature of maximum H
2
consumption rate during TPR
[b] Hydrogen consumption relative to the CuO content after TPR
Scanning electron microscopy
SEM investigation was done on one precursor sample in each case of CZ7.0, CZ7.5 and CZ9.0
with respect to the question whether evidence for the mentioned inhomogeneity was found.
Exemplarily, images of CZ9.0 showing round shaped large aggregates are depicted in Figure
S3-a,b. This droplet-like shape is a result of the spray-drying of the washed precursor slurry. No
obvious differences in the morphology were found within the three investigated samples.
Indeed, no homogeneous morphology of the primary particles was observed, but needles,
platelets, particles and some areas with smooth surfaces were found in the samples (Figure
S3-c,d). The needle like morphology was predominant and was already observed in the
Cu,Zn,Al-system where this was assigned to zincian malachite
[1]
. All samples additionally
contain different phases of zincian malachite and aurichalcite. The range of observed Cu:Zn
ratios measured by EDX at different spots (at least 10 per sample) was similar for CZ7.0 and
CZ7.5 (≈ from 55:45 to 75:25) and somewhat narrower for CZ9.0 (≈ from 63:37 to 72:28). Zn
richer spots should be due to the presence of aurichalcite and Cu rich spots due to zincian
malachite. Because in CZ7.5 the spot with the lowest Zn content measured by EDX was
Chapter 3: Supplementary i
nformation
62
25 mol%
, no spot of pure zM2 with an expected Zn content of 15
XRD, Figure 3-5e)
was detected. This is due to the fact that EDX cannot be applied to infinitely
small regions. The elemental composition of the measured spots is probably a superposition of
many small crystalli
tes from different phases (aurichalcite, zM1, zM2) and gives rather an
averaged picture than a detailed result.
Figure S3-1:
Scanning electron microscopy
prepared at pH 9.0 showing an overview image with roundish primary particles (a), a primary particle (b), zincian
malachite needles, platelets and particles (c), plane areas (d).
Thermal analysis
Thermal deco
mposition of the precursors was performed to obtain nano
particles. This calcination step was monitored by thermogravimetric measurements (TG)
combined with evolved gas analysis (EGA). The TG
973 K, in
contrast to the calcinations for catalyst preparation, which is performed up to 603
Anions in the Cu,Zn hydroxy carbonates are decomposed under emission of water and carbon
dioxide. XRD patterns of all samples after thermal treatment up to 973
crystalline CuO and ZnO (not shown).
nformation
, no spot of pure zM2 with an expected Zn content of 15
%
(Zn poor zM2, according to
was detected. This is due to the fact that EDX cannot be applied to infinitely
small regions. The elemental composition of the measured spots is probably a superposition of
tes from different phases (aurichalcite, zM1, zM2) and gives rather an
averaged picture than a detailed result.
Scanning electron microscopy
images (2.5 keV)
of a Cu,Zn hydroxy carbonate precursor sample
prepared at pH 9.0 showing an overview image with roundish primary particles (a), a primary particle (b), zincian
malachite needles, platelets and particles (c), plane areas (d).
mposition of the precursors was performed to obtain nano
-
sized CuO and ZnO
particles. This calcination step was monitored by thermogravimetric measurements (TG)
combined with evolved gas analysis (EGA). The TG
-
EGA experiments were executed up to
contrast to the calcinations for catalyst preparation, which is performed up to 603
Anions in the Cu,Zn hydroxy carbonates are decomposed under emission of water and carbon
dioxide. XRD patterns of all samples after thermal treatment up to 973
crystalline CuO and ZnO (not shown).
(Zn poor zM2, according to
was detected. This is due to the fact that EDX cannot be applied to infinitely
small regions. The elemental composition of the measured spots is probably a superposition of
tes from different phases (aurichalcite, zM1, zM2) and gives rather an
of a Cu,Zn hydroxy carbonate precursor sample
prepared at pH 9.0 showing an overview image with roundish primary particles (a), a primary particle (b), zincian
sized CuO and ZnO
particles. This calcination step was monitored by thermogravimetric measurements (TG)
EGA experiments were executed up to
contrast to the calcinations for catalyst preparation, which is performed up to 603
K.
Anions in the Cu,Zn hydroxy carbonates are decomposed under emission of water and carbon
K revealed well-
Chapter 3: Supplementary information
63
Exemplarily, the evolutions of the mass loss and the normalized H
2
O and CO
2
traces for a
Cu,Zn hydroxy carbonate precursor sample prepared at pH 7.5 are depicted in Figure S3-. EGA
shows that the decomposition mainly proceeds in three steps. After the release of physisorbed or
incorporated water (range I, up to 403 K), the second step is characterized by simultaneous
emission of H
2
O and CO
2
(range II, ca. 403-673 K). In the third step, only CO
2
is emitted at
high temperatures (range III, ca. 673-873 K). The reason of this last decomposition step is the
presence of temperature stable carbonate species (HT-CO
32-
) which are probably located at the
interface between the formed CuO and ZnO
[2-3]
. The role of this residual carbonate, which is
still present after calcination at not too high temperatures, is debated and it has been proposed
that it can stabilize oxidized copper species in the reduced catalyst and increase the activity
[4]
.
The abundance and the thermal stability of these species are suggested to be a measure for the
amount and the strength of the interactions across interfaces and grain boundaries of CuO/ZnO
aggregates. Therefore, pure malachite Cu
2
(OH)
2
CO
3
and hydrozincite Zn
5
(OH)
6
(CO
3
)
2
do not
show the emission of these species
[2-3, 5]
.
The theoretical mass losses for zincian malachite (Cu
1-x
Zn
x
)
2
(OH)
2
CO
3
and aurichalcite
(Cu
1-y
Zn
y
)
5
(OH)
6
(CO
3
)
2
can be calculated to 28% and 26%, respectively, and depend only
slightly on the Cu:Zn ratio because of the similar molar masses of Cu and Zn. Thus, samples
containing low zincian malachite fractions according to XRD should show low mass losses.
Indeed, CZ6.0 exhibits the lowest value of 26.9 wt% (Figure S3-a) and all other precipitation
pH values show higher mass losses (28.2 to 29.2 wt%) due to lower aurichalcite fractions.
The HT-CO
32-
decomposition temperatures indicate the strength of interaction between CuO and
ZnO particles. The averaged values are in the range from around 720 (CZ7.5 and CZ8.0) to
750 K (CZ8.5) but no simple trend can be observed (Figure S3-b) and the error bars are
relatively large. The HT-CO
32-
amount can be calculated from the fraction of HT-CO
2
related to
the overall CO
2
emission in a semi-quantitative manner based on the integrals of the MS traces.
This fraction is relatively constant around 48-50% for precipitation pH values up to 7.0 and
ranges between 38 and 45% for higher pH values (Figure S3-c). The smallest proportions are
detected for pH 7.5 and 8.0 with about 40%. Theses samples also showed the lowest
temperature for HT-CO
2
emission. Again, no clear trend with preparation pH can be observed.
Because the calcination of the precursors is performed only at 603 K, the high temperature
carbonate species stays present in the sample. It was suggested that it can contribute to increased
activity by the formation of copper suboxide species
[4]
.
Chapter 3: Supplementary information
64
Figure S3-2: TG-MS results of hydroxy carbonate precursor sample prepared at pH 7.5: mass loss (black), MS traces
of H
2
O (blue) and CO
2
(green).
Figure S3-3: TG-MS results of Cu,Zn hydroxy carbonate precursors: mass loss after heating to 973 K (a),
temperature of highest CO
2
emission rate (b) and CO
2
emission above 673 K relative to overall CO
2
emission
according to MS trace (c). Shown values are representing average values, endings of the error bars represent the real
values obtained for the two samples for each precipitation pH value. Lines are guides for the eyes.
Chapter 3: Supplementary information
65
Temperature programmed reduction
Description of the reduction profiles with a single peak was not possible because at least one
shoulder was observed (Figure S3-). This can be attributed to the reduction of CuO in multiple
steps
[6]
, reduction of multiple CuO species, e.g. from different precursor phases
[7]
, or reduction
of other components than CuO.
All reduction profiles were analyzed with respect to the temperature of the highest H
2
consumption (473-479 K) and the reducibility of CuO (95-104%). The results were summarized
in Table S3-1. Within the series, neither distinct changes nor clear trends were observed.
Figure S3-4: TPR profile of a calcined Cu,Zn sample prepared at pH 6.5.
[1] M. Behrens, J. Catal. 2009, 267, 24-29.
[2] B. Bems, M. Schur, A. Dassenoy, H. Junkes, D. Herein, R. Schlögl, Chem. Eur. J.
2003, 9, 2039-2052.
[3] G. J. Millar, I. H. Holm, P. J. R. Uwins, J. Drennan, J. Chem. Soc. Faraday Trans.
1998, 94, 593-600.
[4] L. M. Plyasova, T. M. Yureva, T. A. Kriger, O. V. Makarova, V. I. Zaikovskii, L. P.
Soloveva, A. N. Shmakov, Kinet. Catal. 1995, 36, 425-433.
[5] M. Behrens, F. Girgsdies, A. Trunschke, R. Schlögl, Eur. J. Inorg. Chem. 2009, 1347-
1357.
[6] M. M. Günter, B. Bems, R. Schlögl, T. Ressler, J. Synchrotron Rad. 2001, 8, 619-621.
[7] G. Fierro, M. LoJacono, M. Inversi, P. Porta, F. Cioci, R. Lavecchia, Appl. Catal. A
1996, 137, 327-348.
Chapter 4: Development of Cu-Catalysts for Methanol Synthesis from CO
2
and CO
67
Chapter 4: Development of Cu-Catalysts for Methanol Synthesis
from CO
2
and CO
Stefan Zander, Edward Kunkes, Manfred E. Schuster, Julia Schumann, Gisela Weinberg, Robert
Schlögl, Malte Behrens.
Abstract
In this work we compare the classical zincian malachite-derived Cu/ZnO with a new Cu/MgO
catalyst at a fixed molar ratio of Cu to Zn and Mg, respectively, of 80:20. The geometric
influence of MgO turned out to be better compared to ZnO but the synergetic effect of Cu and
ZnO during methanol synthesis from CO
2
/CO/H
2
was unequaled. Both geometric and synergetic
effects were combined by preparation of Cu/MgO/ZnO sample which exhibited a higher activity
than Cu/ZnO and Cu/MgO. Changing the feed gas to CO/H
2
, Cu/MgO was most active.
Chapter 4: Development of Cu-Catalysts for Methanol Synthesis from CO
2
and CO
68
4.1 Introduction
The development and optimization of industrially applied high performance catalysts is usually
a continuous process that to a large extent is based on the experience of the manufacturer. The
accumulated knowledge from the combination of empirical trial-and-error experimentation and
an afterward structure-function-relationship-guided optimization approach within the boundary
conditions of a feasible and scalable synthesis often leads to very complex recipes – sometimes
generalized as the “black magic” of catalyst preparation. In the last years, we have worked on
better understanding the synthesis and functionality of the Cu/ZnO/Al
2
O
3
methanol synthesis
catalyst using the well-documented industrially applied preparation route
[1-4]
as starting point.
As a result of this effort, we have elaborated a model of the so-called “chemical memory”
[5-6]
of
catalyst preparation and of the active site in this catalyst system.
[7]
Here, we show how this
knowledge can be used to develop a new family of Cu-based catalysts.
With the help of structure-performance-relationships observed within a series of functional
powder catalysts and DFT calculations, the active site of industrial methanol synthesis was
identified as a combination of a surface defect of Cu and the presence of partially reduced Zn
species at this defect,
[7]
explaining the widely studied “Cu-ZnO synergy”.
[8-9]
Within the
industrial synthesis, high concentrations of these sites can be realized by preparation of
defective Cu nanoparticles and migration of ZnO
x
species onto the Cu surface as a result of
strong-metal-support interaction (SMSI)
[7, 10-12]
and an intimate interface contact of both catalyst
components. At the same time, the total accessible Cu surface area (SA
Cu
) is large, because the
bulk-catalyst is prepared with a porous microstructure
[5, 13]
from a co-precipitated precursor
compound. In this context, ZnO acts as a geometrical spacer between the Cu nanoparticles and
helps to increase and stabilize the Cu dispersion.
[9, 14]
Thus, ZnO has two functionalities in the
final catalyst: (i) as nanoparticles it acts as a physical spacer between the Cu particles stabilizing
the porous microstructure and (ii) as a thin layer at the surface of the Cu particles it is an
essential ingredient for the active site. The work presented here was guided by the idea to
separate these two effects.
A simplified scheme of the relevant properties of Cu/ZnO methanol synthesis catalysts is shown
in Figure 4-1a. Three prerequisites have to be fulfilled in order to generate a high performing
catalyst. The material should have a high SA
Cu
to expose a large number of active sites; the Cu
phase must be defective to achieve a high density of active sites at the surface; and SMSI-
induced synergetic effect of ZnO must be present to activate the defect sites for methanol
synthesis. Only if all three factors come together, in the darkest shaded region of Figure 4-1a,
Chapter 4:
the catalyst will be highly active for methanol synthesis. The defects are ge
careful and delicate preparation method yielding distorted Cu nanoparticles in close contact to
the oxide phase, while the other two properties are governed by function (i) and (ii) of the ZnO
component.
Figure 4-1: a) Schematic represent
at
catalyst. b) Scheme of the role of precursor composition for the Cu dispersion in the final catalyst.
The synthesis route for pre
paring Cu/ZnO/(Al
including T- and pH-
controlled co
sodium carbonate solution followed by aging,
carbonate precursor. This material is calcined and finally activated by reduction of CuO to
metallic Cu. Low amounts of Al
catalyst.
[16-17]
The relevant precursor material has been identified as thin needles of zincian malachite,
(Cu,Zn)
2
(OH)
2
CO
3
.
[5]
The incorporation of Zn
nano-
structuring of the CuO/ZnO aggregates formed upon calcination due to the perfect
distribution of both species in the joint crystal lattice of the precursor compound. This can be
understood as a purely
geometric effect, which is schematically shown in
basis for the functionality (i) of ZnO. Zn
Chapter 4:
Development of Cu-
Catalysts for Methanol Synthesis from CO
the catalyst will be highly active for methanol synthesis. The defects are ge
careful and delicate preparation method yielding distorted Cu nanoparticles in close contact to
the oxide phase, while the other two properties are governed by function (i) and (ii) of the ZnO
at
ions of the necessary ingredients for a high performance methanol synthesis
catalyst. b) Scheme of the role of precursor composition for the Cu dispersion in the final catalyst.
paring Cu/ZnO/(Al
2
O
3
) catalysts follows a multi
controlled co
-precipitation
[6]
of aqueous Cu,Zn,Al nitrate solution with
sodium carbonate solution followed by aging,
[15]
washing, and drying to yield a hydroxide
carbonate precursor. This material is calcined and finally activated by reduction of CuO to
metallic Cu. Low amounts of Al
2
O
3
acts as a structural promoter in the industrially applied
The relevant precursor material has been identified as thin needles of zincian malachite,
The incorporation of Zn
2+
into the cationic lattice of malachite favors the
structuring of the CuO/ZnO aggregates formed upon calcination due to the perfect
distribution of both species in the joint crystal lattice of the precursor compound. This can be
geometric effect, which is schematically shown in
Figure
basis for the functionality (i) of ZnO. Zn
2+
is well suited for this purpose,
because it
Catalysts for Methanol Synthesis from CO
2
and CO
69
the catalyst will be highly active for methanol synthesis. The defects are ge
nerated by the
careful and delicate preparation method yielding distorted Cu nanoparticles in close contact to
the oxide phase, while the other two properties are governed by function (i) and (ii) of the ZnO
ions of the necessary ingredients for a high performance methanol synthesis
catalyst. b) Scheme of the role of precursor composition for the Cu dispersion in the final catalyst.
) catalysts follows a multi
-step procedure
of aqueous Cu,Zn,Al nitrate solution with
washing, and drying to yield a hydroxide
-
carbonate precursor. This material is calcined and finally activated by reduction of CuO to
acts as a structural promoter in the industrially applied
The relevant precursor material has been identified as thin needles of zincian malachite,
into the cationic lattice of malachite favors the
structuring of the CuO/ZnO aggregates formed upon calcination due to the perfect
distribution of both species in the joint crystal lattice of the precursor compound. This can be
Figure
4-1b and is the
because it
exhibits the
Chapter 4: Development of Cu-Catalysts for Methanol Synthesis from CO
2
and CO
70
same charge and an ionic radius similar to Cu
2+
favoring substitution. However, incorporation
into the malachite lattice is limited to < 30% due to solid state chemical constraints
[18]
that are
likely due to the differences in the coordination environment between the Jahn-Teller-ion Cu
2+
(d
9
) and Zn
2+
(d
10
). Mg
2+
is an interesting replacement for Zn
2+
, because its charge matches and
its ionic radius differs, alike Zn
2+
, by less than 2% from that of Cu
2+
. Furthermore,
(Cu,Mg)
2
(OH)
2
CO
3
crystallizes in the rosasite crystal structure, which is closely related to that
of malachite and should open the door for a comparable precursor chemistry between Cu,Zn and
Cu,Mg compounds. Moreover, (Cu,Mg)
2
(OH)
2
CO
3
is naturally occurring as the mineral
McGuinessite
[19-20]
and can incorporate more Mg
2+
than Cu
2+
, which has not been achieved for
zincian malachite. Thus, a more efficient dilution of the Cu
2+
ions might be possible with Mg
2+
compared to Zn
2+
by application of lower amounts of Cu to further promote the nanostructuring
and increase the Cu dispersion.
In this work we compare the classical zincian malachite-derived Cu/ZnO with a new Cu/MgO
catalyst at a fixed molar ratio of Cu to Zn and Mg, respectively, of 80:20. At this ratio, Zn
incorporation into malachite does not exceed the critical Zn concentration in zincian malachite
to assure synthesis of phase-pure precursor compounds resulting in high comparability of the
Cu,Zn and Cu,Mg systems and in uniform catalysts whose properties can be easily traced back
to the precursor compounds. Both precursors were prepared from mixed nitrate solutions by
controlled co-precipitation with sodium carbonate solution and subsequent ageing in the mother
liquor. They are denoted CZ and CM in the following.
4.2 Experimental
4.2.1 Catalyst Preparation
Hydroxide-carbonate precursors of CZ and CM were synthesized by co-precipitation (T =
338 K) from Cu,Zn and Cu,Mg (80:20) nitrate solutions and Na
2
CO
3
solution as precipitating
agent in an automated lab reactor (LabMax, Mettler Toledo). The pH was set to 6.5 for CZ and
9.0 for CM. The precipitates were aged (> 60 min), filtrated, washed and dried. Calcination was
carried out in air at 603 K (2 K min
-1
) for 3 h. One part of the calcined CM was impregnated
with Zn citrate solution, dried and calcined again at the same T.
Chapter 4: Development of Cu-Catalysts for Methanol Synthesis from CO
2
and CO
71
4.2.2 Characterization
X-ray fluorescence (XRF) was performed using a Bruker S4 Pioneer X-ray spectrometer. XRD
was measured on a STOE STADI P transmission diffractometer using Cu K
α1
radiation.
Specific surface areas were determined by N
2
physisorption at liquid nitrogen T in a
Quantachrome Autosorb-6 machine. SA
Cu
was determined in a custom-made setup by applying
N
2
O reactive frontal chromatography (N
2
O-RFC) at 303 K
[21-22]
after reducing the catalysts at
523 K in 5% H
2
. For TEM investigations, a FEI Titan Cs 80-300 microscope operated at 300 kV
and energy-dispersive X-ray (EDX) analyzer was used. Spherical aberrations were corrected by
use of the CEOS Cs-corrector reaching an information limit of 0.8 Å. HRTEM pictures were
processed to obtain the power spectra which were used to measure interplanar distances and
angles for phase identification. The reduced samples were transferred in a glovebox to the
microscope without contact to air.
4.2.3 Catalytic performance
Catalytic tests in CO
2
hydrogenation were carried out in a fixed bed flow reactor. 200 mg (100-
200 µm, mixed with 2 g of crushed SiO
2
chips) were loaded into a 10 mm inner diameter
stainless steel reactor tube. The catalysts were reduced at 573 K (1 K min
−1
) for 1.5 hours in
20% H
2
in He. Upon completion of the reduction, the reactor was cooled to 523 K, a 3:1 H
2
/CO
2
mixture (100 mL min
−1
) containing 4% Ar (as internal standard) was introduced into the reactor,
and the pressure was raised to 30 bars. Online analysis of products was performed with a gas
chromatograph (Agilent 7890A). After the start of the reaction, the catalysts were allowed to
stabilize for 6 hours time on stream at 523 K. CO hydrogenation was performed at the same T
and p in the same set up, but using a 6 mm stainless steel reactor tube and 50 mg of catalyst
diluted with 0.7 g SiO
2
. The feed gas contained 14% CO, 59% H
2
, 4% Ar and balance He.
Performance under synthesis gas conditions was determined after switching the gas composition
gradually to a composition of 6% CO, 8% CO
2
, 59% H
2
and balance He.
Chapter 4: Development of Cu-
Catalysts for Methanol Synthesis from CO
72
4.3
Results and discussion
XRD of the precursors
confirmed the formation of single phase materials with a crystal structure
similar to malachite (
Figure
pronounced shift of the 20-
1
incorporation of non-Jahn-
Teller cations in the lattice of malachite and from the similar position
of the reflections in both patterns a similar degree of substitution can be
suggesting that the non-Jahn
-
structure in both samples. It is noted that CM exhibit
indicative of small crystallites.
compared to CZ
, which exhibits larger and well
larger BET surface area
of the CM precursor
Figure 4-2:
a) XRD patterns of the precursor materials of CZ (green) and CM (red). The pattern of malachite (black;
ICSD: 72-
75) is shown as bar graph. b) XRD patterns of the calcined samples CZ (green), CM (
CuO (black; ICSD: 80-
76) is included as reference.
Figure 4-3:
Scanning electron microscopy images
Catalysts for Methanol Synthesis from CO
2
and CO
Results and discussion
confirmed the formation of single phase materials with a crystal structure
Figure
4-2a). In comparison to the literature patter
n of malachite, a
1
p
eak is seen in both compounds. This is an indication for the
Teller cations in the lattice of malachite and from the similar position
of the reflections in both patterns a similar degree of substitution can be
estimated (
-
Teller ions have been completely incorporated into the malachite
structure in both samples. It is noted that CM exhibit
s significantly broader XRD peaks
indicative of small crystallites.
Also the particle morphology
of CM look
, which exhibits larger and well
-separated particles (Figure 4-
3
of the CM precursor
has been observed (Table 4-1).
a) XRD patterns of the precursor materials of CZ (green) and CM (red). The pattern of malachite (black;
75) is shown as bar graph. b) XRD patterns of the calcined samples CZ (green), CM (
76) is included as reference.
Scanning electron microscopy images
(2.5 keV) of CZ (a) and CM (b)
confirmed the formation of single phase materials with a crystal structure
n of malachite, a
eak is seen in both compounds. This is an indication for the
Teller cations in the lattice of malachite and from the similar position
estimated (
Table 4-1)
Teller ions have been completely incorporated into the malachite
s significantly broader XRD peaks
of CM look
s rather spongy
3
). Accordingly, a
a) XRD patterns of the precursor materials of CZ (green) and CM (red). The pattern of malachite (black;
75) is shown as bar graph. b) XRD patterns of the calcined samples CZ (green), CM (
red) and CMZ (blue).
Chapter 4:
Table 4-1:
Properties of the CZ, CM and CMZ catalysts (prec. = precursor material, calc.
Sample Cu:M ratio
CZ 80:20
[a]
CM 83:17
[a]
CMZ 80:16:4
[b]
[a] molar, determined by XRF, ± 1 mol
%
[b] molar, estimated
[c] Crystallite domain size of (Cu,M)
2
(OH)
[d] Specific Cu surface area
of the reduced catalyst
Figure 4-4: (HR-
)TEM images of the reduced
particles and are used for phase identification
Upon calcination at 603 K, poorly crystalline CuO is formed in both samples as evidenced by
XRD (Figure 4-2
b), while the ZnO and MgO components are mostly X
Chapter 4:
Development of Cu-
Catalysts for Methanol Synthesis from CO
Properties of the CZ, CM and CMZ catalysts (prec. = precursor material, calc.
= calcined material).
D
[c]
/ nm
prec. // calc. SA
BET
/ m
2
g
-1
prec. // calc.
26.6 // 5.8 36 // 83
8.0 // 2.8 81 // 73
8.0 // 3.9 81 // 80
%
(OH)
2
CO
3
(precursor) and CuO/MO (calcined), ± 0.2 nm
derived from XRD peak profiles
of the reduced catalyst
determined by N
2
O-chemisorption, ± 1 m
2
g
-1
)TEM images of the reduced
CZ catalyst.
The insets show power spectra of the neighbouring
particles and are used for phase identification
.
Upon calcination at 603 K, poorly crystalline CuO is formed in both samples as evidenced by
b), while the ZnO and MgO components are mostly X
-
ray amorphous. Again
Catalysts for Methanol Synthesis from CO
2
and CO
73
= calcined material).
SA
Cu[d]
/ m
2
g
cat-1
16.0
20.3
24.2
derived from XRD peak profiles
The insets show power spectra of the neighbouring
Upon calcination at 603 K, poorly crystalline CuO is formed in both samples as evidenced by
ray amorphous. Again
Chapter 4: Development of Cu-
Catalysts for Methanol Synthesis from CO
74
CM exhibits a significantly smaller crystallite size according to the XRD peak width
slightly smaller
specific surface area (
higher specific Cu surface area
Figure 4-5: (HR-)TEM
images of the reduced
TEM investigation of the reduced catalysts showed that the general m
(Figure 4-4a,b) and CM (
Figure
shaped Cu particles separated by differently sized crystallites of ZnO or MgO, re
few Cu
2
O particles are found that probably have formed due to slight re
reduced catalysts upon sample transfer.
CM is consistent with the difference in Cu surface areas
Catalysts for Methanol Synthesis from CO
2
and CO
CM exhibits a significantly smaller crystallite size according to the XRD peak width
specific surface area (
Table 4-1). Furthermore, CM
yields a by more than 30%
higher specific Cu surface area
after reduction.
images of the reduced
CM catalyst.
TEM investigation of the reduced catalysts showed that the general m
icrostructure
Figure
4-5a,b,c) is similar
and characterized by arrangements of round
shaped Cu particles separated by differently sized crystallites of ZnO or MgO, re
O particles are found that probably have formed due to slight re
reduced catalysts upon sample transfer.
The presence of larger Cu particles in CZ compared to
CM is consistent with the difference in Cu surface areas
(Table 4-1).
This result indicate
CM exhibits a significantly smaller crystallite size according to the XRD peak width
, but a
yields a by more than 30%
icrostructure
of CZ
and characterized by arrangements of round
shaped Cu particles separated by differently sized crystallites of ZnO or MgO, re
spectively. A
O particles are found that probably have formed due to slight re
-oxidation of the
The presence of larger Cu particles in CZ compared to
This result indicate
s that
Chapter 4: Development of Cu-Catalysts for Methanol Synthesis from CO
2
and CO
75
MgO is intrinsically a better geometrical spacer compared to ZnO as even at the non-ideal
substitution level of 20% Cu particles can be obtained that with an average size below 10 nm
are similarly small as found in state-of-the-art Cu/ZnO/Al
2
O
3
catalysts. Thus, the structurally
promoting role of ZnO has been successfully replaced with MgO.
High resolution TEM showed that the Cu particles in both samples contain planar defects,
which have been shown to contribute to the methanol synthesis activity in Cu/ZnO catalysts
[7]
(CZ: Figure 4-4c; CM: Figure 4-5d). Thus, the important defectiveness of Cu is probably a
result of the precursor decomposition approach common to both catalysts, which leads to
crystallization of distorted Cu crystallites at relatively mild T.
Both catalysts have been tested in methanol synthesis with various feed gas composition, i.e.
hydrogenation of pure CO
2
, a CO
2
/CO mixture and pure CO. The results are shown in Figure
4-6. In the hydrogenation of pure CO
2
, CZ showed a much higher activity than CM showing
clearly that the methanol synthesis rate is not only a function of the exposed Cu surface area
alone (Figure 4-6a). According to Figure 4-1, the low activity of CM can be explained with the
absence of the synergetic SMSI-effect as MgO is an irreducible oxide that does not show SMSI
in the relevant T regime. The situation is similar if methanol is produced from a typical
synthesis gas mixture with CO
2
and CO in the feed (Figure 4-6b). CZ shows a slightly lower
rate of methanol production compared to the CO
2
/H
2
feed, while CM remains essentially
inactive despite the large exposed Cu surface area. These results strikingly confirm the crucial
synergetic role of the ZnO-promoter that has been subject of many literature reports.
Figure 4-6: Catalytic results of methanol synthesis of the CZ, CM and CMZ catalysts in different feed gas
compositions at 30 bar and 503 K.
Chapter 4: Development of Cu-
Catalysts for Methanol Synthesis from CO
76
Figure 4-7: (HR-
)TEM images of the reduced
With the idea to “switch on” the lacking Cu
reported earlier for model catalysts
Impregnation of the calcined CM with 5 wt% ZnO resulted in a catalyst that was indeed able to
convert CO
2
and the
synthesis gas
4-6a,b). HRTEM
investigation of
particle morphology like
CM (
surf
ace of the Cu/MgO aggregates
from synthesis gas
of CMZ was even higher than that of CZ, which
dispersion in the Cu/MgO-
aggregates. The intrinsic rates per SA
CMZ in this experiment
(not shown)
dispersion instead of ZnO, namely its lowe
Catalysts for Methanol Synthesis from CO
2
and CO
)TEM images of the reduced
CMZ catalyst.
With the idea to “switch on” the lacking Cu
-ZnO synergy by addition of Zn –
reported earlier for model catalysts
[17]
–
a third catalyst, labeled CMZ, was prepared.
Impregnation of the calcined CM with 5 wt% ZnO resulted in a catalyst that was indeed able to
synthesis gas
mixture into methanol at much higher rates than CM
investigation of
CMZ (Figure 4-7) showed similar general
microstructure and
CM (
Figure 4-5) but additionally confirmed
the presence of ZnO at the
ace of the Cu/MgO aggregates
(Figure 4-7c). The weight-
based methanol production rate
of CMZ was even higher than that of CZ, which
is a result of the better Cu
aggregates. The intrinsic rates per SA
Cu
are very similar for CZ and
(not shown)
, which points to another advantage of using MgO for Cu
dispersion instead of ZnO, namely its lowe
r specific weight. At the same molar Cu content of
a similar has been
a third catalyst, labeled CMZ, was prepared.
Impregnation of the calcined CM with 5 wt% ZnO resulted in a catalyst that was indeed able to
mixture into methanol at much higher rates than CM
(Figure
microstructure and
the presence of ZnO at the
based methanol production rate
is a result of the better Cu
are very similar for CZ and
, which points to another advantage of using MgO for Cu
r specific weight. At the same molar Cu content of
Chapter 4: Development of Cu-Catalysts for Methanol Synthesis from CO
2
and CO
77
80% (metal-based) the effective Cu loading of CM is as high as 86 wt%, while that of CZ is 75
wt%. The very high value for CM shows that the microstructure of this kind of catalyst should
not be described as a classical supported system, but its sponge-like microstructure also
resembles an “oxide-stabilized Raney-Cu”.
These results show that the functions of the oxide component can be successfully separated in
Cu-based methanol synthesis catalysts. It was shown for a given catalyst composition as a
proof-of-principle that this approach enables preparation of high-performance catalysts and
leaves additional degrees of freedom for future optimization. In particular, Cu dispersion can be
optimized within the proven malachite-precursor method by increasing the Cu substitution
without being bound to the constraints of the Cu,Zn system. Furthermore, the method of
addition and amount of the synergistic promoter can be varied for a given highly-dispersed
Cu/oxide system to switch on the production of methanol from CO
2
or synthesis gas. In CMZ
areas of relatively large particles of very crystalline ZnO have been observed (Fig. 6c). These
domains probably do not contribute to the synergetic catalytic effect and point to the possibility
to further improve this catalyst by optimization of the ZnO addition.
Interestingly, the catalytic performance of the samples is completely changed when a CO/H
2
feed is used for methanol synthesis. Here CM shows a very high methanol production rate,
which clearly exceeds that of CZ or CMZ in the other feed gases (Figure 4-6c). This result is in
line with previous studies
[23-24]
that have shown that MgO-supported Cu is a powerful CO
hydrogenation catalyst. Interestingly, while it was a prerequisite for methanol production in
CO
2
-containing feeds, the addition of Zn to CM was detrimental in this reaction possibly by
partially covering of the active surface. Thus, in addition of being a very powerful CO
hydrogenation catalyst, the Cu/MgO and derived Cu/MgO/ZnO systems also represent a
suitable material basis for conducting basic studies on the roles of synergy, dispersion and
structural dynamics for methanol synthesis in different feed gases.
In summary, the high comparability of the three catalysts due to the similar general morphology
found by TEM investigation, allows tracing back the differences in activity of the samples to the
influence of the oxide phase(s) ZnO and/or MgO. These two oxides do not only act as structural
promoters, but also determine the preferred pathway of methanol synthesis from CO
2
or CO as
carbon source.
Chapter 4: Development of Cu-Catalysts for Methanol Synthesis from CO
2
and CO
78
4.4 Conclusion
We propose that the presented synthetic approach opens the door to exploit new room for
knowledge-based optimization of the proven Cu/ZnO/Al
2
O
3
catalyst system. It seems in
particular promising to combine the highest possible substitution of Cu
2+
in malachite by a
suitable diluent like Mg
2+
for proper structural promotion with the right amount of a reducible
oxide at the surface of the Cu particle like ZnO for proper synergistic promotion. Furthermore,
the presented materials show potential to fertilize new progress in the long-lasting studies of the
mechanism of methanol synthesis by providing fundamental insight into the role of different
material components. Future studies in our lab will aim at better understanding the observed
differences, which point to fundamental differences in the active sites for CO and CO
2
conversion over the same catalysts, in particular with regard to the involvement of Zn in the
latter, but not in the former. Thus, in addition of being a very powerful CO hydrogenation
catalyst, the Cu/MgO and derived Cu/MgO/ZnO systems also represent a suitable material basis
for conducting basic studies on the roles of synergy, dispersion and structural dynamics for
methanol synthesis in different feed gases.
Acknowledgement
Edith Kitzelmann (XRD measurements), Achim Klein-Hoffmann and Olaf Timpe (XRF),
Gisela Lorenz (BET measurements), Nygil Thomas (help with catalytic measurements) are
acknowledged. Financial support was given by the German Federal Ministry of Education and
Research (BMBF, FKZ 01RI0529, 2005-2008) and the STMWFK (NW-0810-0002, since
2010).
Chapter 4: Development of Cu-Catalysts for Methanol Synthesis from CO
2
and CO
79
4.5 References
[1] C. Baltes, S. Vukojevic, F. Schüth, J. Catal. 2008, 258, 334-344.
[2] D. Waller, D. Stirling, F. S. Stone, M. S. Spencer, Faraday Discuss. 1989, 87, 107-120.
[3] J. L. Li, T. Inui, Appl. Catal. A 1996, 137, 105-117.
[4] J. B. Hansen, P. E. H. Nielsen, in Handbook of Heterogeneous Catalysis, Vol. 6 (Eds.:
G. Ertl, G. Knözinger, F. Schüth, J. Weitkamp), Wiley-VCH, Weinheim 2nd ed., 2008,
pp. 2920-2949.
[5] M. Behrens, J. Catal. 2009, 267, 24-29.
[6] M. Behrens, D. Brennecke, F. Girgsdies, S. Kißner, A. Trunschke, N. Nasrudin, S.
Zakaria, N. F. Idris, S. B. Abd Hamid, B. Kniep, R. Fischer, W. Busser, M. Muhler, R.
Schlögl, Appl. Catal. A 2011, 392, 93-102.
[7] M. Behrens, F. Studt, I. Kasatkin, S. Kühl, M. Hävecker, F. Abild-Pedersen, S. Zander,
F. Girgsdies, P. Kurr, B.-L. Kniep, M. Tovar, R. W. Fischer, J. K. Nørskov, R. Schlögl,
Science 2012, 336, 893-897.
[8] Y. Kanai, T. Watanabe, T. Fujitani, T. Uchijima, J. Nakamura, Catal. Lett. 1996, 38,
157-163.
[9] M. S. Spencer, Top. Catal. 1999, 8, 259-266.
[10] Y. Kanai, T. Watanabe, T. Fujitani, M. Saito, J. Nakamura, T. Uchijima, Catal. Lett.
1994, 27, 67-78.
[11] J. D. Grunwaldt, A. M. Molenbroek, N. Y. Topsoe, H. Topsoe, B. S. Clausen, J. Catal.
2000, 194, 452-460.
[12] R. N. d'Alnoncourt, X. Xia, J. Strunk, E. Löffler, O. Hinrichsen, M. Muhler, Phys.
Chem. Chem. Phys. 2006, 8, 1525-1538.
[13] I. Kasatkin, P. Kurr, B. Kniep, A. Trunschke, R. Schlögl, Angew. Chem. Int. Edit. 2007,
46, 7324-7327.
[14] T. Fujitani, J. Nakamura, Catal. Lett. 1998, 56, 119-124.
[15] S. Zander, B. Seidlhofer, M. Behrens, Dalton Trans. 2012, 41, 13413-13422.
[16] M. V. Twigg, M. S. Spencer, Top. Catal. 2003, 22, 191-203.
[17] M. Kurtz, N. Bauer, C. Buscher, H. Wilmer, O. Hinrichsen, R. Becker, S. Rabe, K.
Merz, M. Driess, R. A. Fischer, M. Muhler, Catal. Lett. 2004, 92, 49-52.
[18] M. Behrens, F. Girgsdies, Z. Anorg. Allg. Chem. 2010, 636, 919-927.
[19] N. Perchiazzi, Z. Kristallogr. 2006, 505-510.
[20] M. Fleischer, L. J. Cabri, Am. Mineral. 1981, 66, 1274-1280.
[21] G. C. Chinchen, C. M. Hay, H. D. Vandervell, K. C. Waugh, J. Catal. 1987, 103, 79-86.
[22] O. Hinrichsen, T. Genger, M. Muhler, Chem. Eng. Technol. 2000, 23, 956-959.
[23] B. Denise, R. P. A. Sneeden, Appl. Catal. 1986, 28, 235-239.
[24] J. C. J. Bart, R. P. A. Sneeden, Catal. Today 1987, 2, 1-124.
Chapter 4: Development of Cu-Catalysts for Methanol Synthesis from CO
2
and CO
80
Supplementary Information
Table S4-1: Internal sample numbers
Sample Precursor Calcined Reduced
CZ 7749 7750 13284
CM 9278 10639 13121
CMZ (9278) 13188 13572
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
81
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
Stefan Zander, Julia Schumann, Igor Kasatkin, Gisela Weinberg, Gregor Koch, Thorsten
Ressler, Patrick Kurr, Benjamin Kniep, Malte Behrens.
Abstract
We report on introduction of gallia as promoter in the Cu,Zn system. Samples with different
promoter content were prepared by co-precipitation. Characterization results (XRF, XRD,
N
2
physisorption, TGMS, TPR, SEM, TEM, UV-Vis, and XANES) during different stadia of
the catalyst preparation process as well as catalytic results in methanol synthesis are presented.
The promoting effect is most effective for low amounts of Ga
3+
(≤ 3 mol%). An increase in
absolute methanol synthesis activity of 60% compared to the binary Ga free system was
observed. Promotion is characterized by a geometric modification which is expressed by a
higher Cu surface area. In contrast, addition of Ga leads to slightly lower intrinsic activities
(related to Cu surface area), probably by modification of the ZnO phase by incorporated Ga
species and consequences for Cu-ZnO synergy. The extent of the geometric and the synergetic
effect depends on incorporation of Zn
2+
and Ga
3+
into the zincian malachite precursor phase and
a linear correlation of the (Zn,Ga) content in this phase with the catalytic activity of the final
catalyst was observed.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
82
5.1 Introduction
Cu/ZnO/(Al
2
O
3
) catalysts are of major industrial interest as they have been successfully applied
in methanol synthesis for over 40 years. Furthermore, they are active in methanol steam
reforming and water-gas-shift reaction. The synthesis route for preparing Cu/ZnO/(Al
2
O
3
)
catalysts follows a multi-step procedure including temperature and pH-controlled co-
precipitation of aqueous Cu,Zn,Al nitrate solution with sodium carbonate solution, aging,
washing, drying, calcination and finally activation by reduction
[1]
.
Metallic copper which is present as nanoparticles is regarded as the active species
[2]
. But there
is still no general agreement about the nature of the active sites in Cu/ZnO catalysts
[3]
.
However, it is known that highly productive catalysts exhibit a high copper surface area
[4]
. This
requires an optimal dispersion of the active copper phase but additionally, the "quality" of the
active Cu surface area plays a decisive role. ZnO is known to act as a geometrical spacer and to
increase the Cu dispersion
[5]
. Thus, it leads to higher exposition of the active surface to the
reaction gas. Beyond this geometrical influence, ZnO was reported to cause synergetic effects
due to the interface contact with the copper phase
[6]
. Proposed active species are Cu-Zn alloy
formed during reduction
[7]
, dissolved Cu
+
in ZnO
[8]
or electron rich Cu at Schottky-junctions
[9]
. In our recently published model of the active site of industrial methanol synthesis over
Cu/ZnO/Al
2
O
3
, the synergetic effect is accounted for by strong metal support interaction (SMSI)
[10]
, which has been observed in high-performance catalysts by HRTEM and XPS. SMSI
between Cu and ZnO has previously been reported in literature and studied by Cu surface area
determination
[11]
, EXAFS
[12]
and IR spectroscopy of CO adsorption
[13]
.
We showed that the intrinsic activity of the exposed Cu surface area scales with the abundance
of stacking faults in Cu nanoparticles
[10]
. This correlation was rationalized by the generation of
high energy sites at the surface at the positions, where the planar defect terminates. Also
residual oxygen in Cu as a result of incomplete reduction might play a role for the defect
structure of active Cu. In this context, the role of promoters like alumina is of interest. Saito et
al.
[14]
reported that addition of metal oxide promoters can have different effects, first the
increase of the Cu dispersion in the case of alumina or zirconia, secondly the improvement of
the specific activity in the case of gallia and chromia. The authors claim that the latter feature is
due to the optimization of the Cu
+
/Cu
0
ratio on the Cu surface under reaction conditions.
Furthermore, Al
2
O
3
is known to enhance the thermal and performance stability
[15-16]
.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
83
We have recently found that an Al content of around 3 mol% is sufficient to obtain a beneficial
promoting effect
[17]
. Al was introduced into the ZnO phase at Zn
2+
sites in tetrahedral
coordination. Alike ZnO, the function of the Al
2
O
3
promoter was divided into a geometrical and
a synergetic contribution. The former affects the Cu dispersion and leads to an increase of the
Cu surface area. The latter promotes the intrinsic activity of Cu and was related to the
incorporation of Al into the ZnO lattice and an influence onto the Cu/ZnO synergy. In this work,
we report on the effects of gallia on Cu/ZnO to study, whether the role of the structural
promoter Al
2
O
3
can be generalized. Part of the data on gallia has already been published
previously
[17]
.
In the literature studies dealing with promoters in the Cu,ZnO system are often carried out with
high fractions of up to 25 mol% and more
[14, 18-25]
. Here, we concentrate on the region of smaller
contents (≤ 4 mol%) for which great promotion of catalytic performance has been observed for
Al
2
O
3
containing Cu/ZnO
[17]
. Additionally, some higher contents are applied for comparison.
We report on introduction of gallia as promoter in the Cu,Zn system (Cu:Zn = 70:30). Aspects
of precursor preparation and microstructural characterization as well as catalytic results in
methanol synthesis are presented.
5.2 Experimental
5.2.1 Sample Preparation
Metal hydroxy carbonate precursors with fixed Cu:Zn ratio (70:30) and different Ga contents up
to 13 mol% (metal base) were synthesized by co-precipitation from acidic Cu,Zn,Ga nitrate
solutions and Na
2
CO
3
solution as basic precipitating agent in an automated lab reactor
(LabMax, Mettler Toledo) under controlled conditions like dosing, stirring, temperature (338 K)
and pH value (6.5). After aging (60 min after pH drop) the slurry was filtrated, the precipitate
washed several times with water and spraydried (Niro minor mobile, T
inlet
= 473 K, T
outlet
=
373 K). Calcination was carried out in air at 603 K (2 K min
-1
) for 3 h. The high Cu:Zn ratio of
70:30 is typically applied in industrial catalyst preparation and aims at a maximal incorporation
of Zn into the zincian malachite precursor phase
[26]
.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
84
5.2.2 Sample Labeling
The designation of the samples was chosen with regard to the nominal Ga content, respectively.
Ga4.0 means a nominal Ga content of 4.0 mol% of all metal species [Cu+Zn+Ga]. The
designation Ga0.0 refers to the binary Cu,Zn reference sample without any trivalent promoter.
5.2.3 Elemental Analysis
All Cu,Zn,Ga calcined oxides were subjected to elemental analysis by XRF (Table 5-2
: Results of
calcined sample characterization and Cu surface areas
). The determined Cu to Zn ratios were near the
nominal ones of 70:30. The measured Ga contents (mol% based on metals) are slightly smaller
than the nominal values. Probably, the deviations are due to the not exactly defined water
content of the metal salts used for the synthesis. Most of all, Ga-nitrate is very hygroscopic.
5.2.4 Characterization
X-ray fluorescence spectroscopy (XRF) was performed after glassing the calcined samples with
Li
2
B
4
O
7
in a Bruker S4 Pioneer X-ray spectrometer. X-ray diffraction (XRD) was applied to the
catalyst precursors and calcined samples. The samples were measured on a STOE STADI P
transmission diffractometer equipped with a primary focusing Ge monochromator (Cu K
α1
radiation) and a linear position sensitive detector (moving mode, step size 0.1 °, counting time
10 s/step, resolution 0.01 °, total accumulation time 634 s). The samples were mounted in the
form of a clamped sandwich of small amounts of powder fixed with a small amount of grease
between two layers of thin polyacetate film. The phase composition was determined by full
pattern refinement in the 2θ range 4-80 ° according to the Rietveld method using the TOPAS
software
[27]
and crystal structure data from the ICSD database. Specific surface areas were
determined by N
2
physisorption in a Quantachrome Autosorb-6 machine after degassing the
samples at 353 K for 2 h. Isotherms were recorded at liquid nitrogen temperature and evaluated
according to the BET method. Thermogravimetric experiments (TGMS) were done on a
NETZSCH Jupiter thermobalance in flowing air. The gas evolution was measured with a
quadrupole mass spectrometer (Pfeiffer Vacuum, Omnistar).
Scanning electron microscopy (SEM) images were taken in a Hitachi S-4800 field emission gun
(FEG) system. Transition electron microscopy (TEM) was performed with a Philips
CM200FEG microscope operated at 200 kV and equipped with an EDX spectrometer. For TEM
investigation, the samples were reduced up to a temperature of 523 K and transferred to the
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
85
microscope in inert atmosphere. The coefficient of spherical aberration was Cs = 1.35 mm, and
the information limit was better than 0.18 nm. High-resolution images with a pixel size of
0.016 nm were acquired at the magnification of 1083000x with a CCD camera, and selected
areas were processed to obtain power spectra (square of the Fourier transform of the image),
which were used for measuring interplanar distances and angles (accuracy ± 1% and ± 0.5 deg,
correspondingly) for phase identification. Projected areas have been measured and equivalent
diameters calculated for 1500-3000 Cu particles in each sample. In all cases the values of
standard error of the mean diameter were ≤ 0.1 nm. Frequency distributions of the particle sizes
fitted well to Lognormal functions. EDX analyses were performed for 5-15 larger aggregates
containing at least several hundred particles in each sample.
Temperature programmed reduction (TPR) was performed with around 40 mg of each sample in
a glass reactor, fixed by means of quartz wool plugs. The reduction was carried out in a CE
instruments TPDRO 1100 machine with 80 mL min
−1
5% H
2
in Ar up to a temperature of 623 K
(6 K min
−1
). The K-values according to Monti and Baiker
[28]
were 100-120 s, the P-values
according to Caballero
[29]
10-12 K. The reduction progress was followed with an internal
thermal conductivity detector. Analysis was performed with regard to the temperature with the
highest H
2
consumption (T
max
) and the total H
2
consumption with respect to the CuO content in
the sample (compared with a pure CuO reference). In the following, the term "reducibility" is
used for the latter feature. The CuO content of the samples was derived from XRF data with the
assumption that only CuO, ZnO and Ga
2
O
3
were present and under neglect of HT-CO
32-
.
UV-Vis spectra were recorded in a Perkin-Elmer Lambda 650 High Performance Spectrometer
equipped with a Harrick Praying Mantis diffuse reflectance attachment. The band gap energy
(direct transition) was calculated by linear extrapolation of the function [F(R
∞
)hν]
1/2
versus hν
to 0, as suggested by Barton et al
[30]
. This procedure results from a linearization of the theory of
direct and indirect band gap transitions in semiconductors.
X-ray absorption spectra were conducted of Ga K-edge (10.367 keV) at the X1 beamline at
HASYLAB at DESY (Hamburg, Germany). Measurements were carried out in transmission
mode. Intensities were detected in ion chambers before samples and behind samples and behind
reference foils. With respect to maximal absorption of the samples at the Ga K-edge samples
were diluted with wax and pressed with a force of 1 ton for 30 seconds to pellets with 13 mm in
diameter. The spectra were calibrated to the position of K-edge of a reference foil (Zn foil for
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
86
Zn K-edge and Ga K-edge). Subsequently, background correction and normalization were
performed. The software Athena 0.8.061 was used
[31]
.
The copper surface area was determined by applying N
2
O reactive frontal chromatography
(N2O-RFC) based on the method proposed by Chinchen et al.
[4]
. Around 100 mg of a sieve
fraction (100-200 µm) of each sample were placed in a stainless steel U-tube reactor and fixed
by means of quartz wool plugs. The prior reduction was carried out in the same device and
conditions as for TPR, but only up to a temperature of 523 K and with a holding time of 30 min.
The reduction progress was additionally followed with a quadrupole mass spectrometer (Pfeiffer
Vacuum, Omnistar). After cooling down to 303 K, the catalyst has been flushed for 30 min in
pure Ar and 15 min in pure He in order to achieve an adsorbate-free reduced catalyst surface.
N
2
O-RFC was performed with 10 mL min
−1
1% N
2
O in He, at which the N
2
O reacts
quantitatively with the Cu surface atoms forming gas-phase N
2
. The specific Cu metal surface
area has been calculated from the formed amount of N
2
using a value of 1.47*10
19
atoms per m
2
for the mean Cu surface atom density. The error of the specific Cu surface area is about
± 1 m
2
g
−1
.
5.2.5 Catalytic testing
Catalytic testing in methanol synthesis was performed using an 8-channel parallel fixed bed
reactor setup working at 60 bars of a synthesis gas mixture (59.5% H
2
, 8.0% CO
2
, 6.0% CO, rest
inert). Gas analytics was done by gas chromatography. 200 mg of each catalyst sample (sieve
fraction 100-200 µm) was loaded to the reactor and reduced prior to the measurement in diluted
hydrogen at 523 K at ambient pressure. After a formation period of 48 hours on stream at
523 K, the catalytic activity was measured for 12 hours and stable conversions were detected.
Afterwards, the reaction temperature was lowered to 483 K and the performance was measured
again for 12 hours. The methanol weight time yields (WTY) were calculated using the methanol
concentration in the outlet gas.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
87
5.3 Results and Discussion
5.3.1 The influence of gallia on the precursor chemistry
Different mixed metal hydroxy carbonate precursor phases can emerge in the course of
precursor preparation, notably Cu
2
(OH)
2
CO
3
(malachite) for pure Cu samples,
(Cu
1-x
Zn
x
)
2
(OH)
2
CO
3
(zincian malachite) with x < 0.3, (Cu
1-y
Zn
y
)
5
(OH)
6
(CO
3
)
2
(aurichalcite)
with y > 0.5, and (Cu,Zn)
6
Al
2
(OH)
16
CO
3
·4H
2
O (hydrotalcite-like phase), only when a
significant amount of Al
3+
is present. The last phase should also be formed with other trivalent
ions instead of Al
3+
, such as Ga
3+
.
The critical role of the precursor chemistry has been emphasized in our earlier work
[26]
. All
parameters applied in each single step of the catalyst preparation influence the bulk and surface
structure and therewith the characteristics and activity of the resulting catalyst. This
phenomenon is also called the "chemical memory" and means the influence of early stage
parameters or rather the characteristics of the precursor phase on the microstructure and activity
of the final catalyst
[32-33]
.
For the industrially applied ternary system (Cu:Zn:Al = 60:30:10), a comprehensive study was
accomplished by Baltes et al. wherein the influence of pH value and temperature during the
precursor preparation was investigated
[23]
. For T = 343 K and pH 6-8, the best catalytic
performance was achieved. A comprehensive understanding of the effect of addition of Al or Ga
can only be achieved if all stages of catalyst preparation, precursor, calcined oxides and active
catalyst, are carefully characterized.
5.3.1.1 XRD analysis
All precursor samples were subjected to XRD analysis for phase identification and
determination of composition and crystal structures. Figure 5-1 shows the XRD patterns of
selected precursors with 1.5, 2.5 and 3.5% nominal Ga content, respectively. Three different
crystalline phases can easily be identified by selected not-overlapping reflections, namely (020)
at 14.8 °2θ for zincian malachite, (400) at 13.0 °2θ for aurichalcite and (003) at 11.6 °2θ for a
hydrotalcite-like phase.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
88
Figure 5-1: Selected XRD patterns of Cu,Zn,Ga hydroxy carbonate precursors with different Ga contents. Bars of the
references malachite (green), aurichalcite (red) and hydrotalcite-like phase (blue) are included.
The phase compositions and domain sizes of the precursors were determined by Rietveld
refinement. As representative example, the refinement result of Ga1.0 is depicted in Figure 5-2.
It is noted that the accurate determination of the exact weight fractions in the phase mixture is
difficult due to the low amounts of some components, the generally low crystallinity and the
high noise of the XRD patterns. However, the general trends seen within the series of samples
are regarded as reliable, while the absolute values of individual samples depend on the fitting
constraints and have to be compared with care. Results of the full Ga series are shown in Figure
5-3. The binary Cu,Zn reference sample Ga0.0 contains about 88% zincian malachite and 12%
aurichalcite as crystalline phases (Figure 5-3a). For increasing Ga content (Ga1.5), the fraction
of aurichalcite is slightly increased up to 19% at the expense of zincian malachite. In Ga2.0, the
zincian malachite fraction is heavily increased and reaches 100% in Ga2.5. But as mentioned,
one has to be very careful with this statement, firstly because of the limits of this evaluation and
secondly because it is only valid for crystalline phases which can be detected by XRD. Further
increase of the Ga content leads to the formation of small but clearly detectable amounts (up to
2%) of a hydrotalcite-like phase. The phase fractions do not change significantly and stay nearly
constant for Ga contents up to 13 mol%.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
89
Figure 5-2: Rietveld refinement for the XRD pattern of a Cu,Zn,Ga hydroxy carbonate precursor sample containing
1.0% Ga, experimental data (black), total calculated curve (red), background (light grey), difference curve (grey),
calculated pattern zincian malachite (orange curve), calculated pattern aurichalcite (blue curve). The thick marks
indicate the positions of the Bragg reflections. The fit quality is comparable for the fits of all other samples.
Also the domain size of zincian malachite can be extracted from the Rietveld refinement in form
of the volume weighted mean column heights. As it can be assumed from the broadening of the
XRD reflections (Figure 5-1), the crystallinity of this phase seems to decrease with increasing
Ga content. Indeed, in the Ga series, the domain size (Figure 5-3b) steadily is decreasing from
Ga0.0 (23.1 nm) to Ga4.0 (6.3 nm) and then remains constant. Only Ga1.5 does not follow this
trend (22.4 nm). Analysis of a reproduced sample prepared at the same conditions confirms the
result and indicates that this effect is real. The smallest domain size for zincian malachite is
reached for a Ga content of 3 mol%. At higher Ga concentrations, only a low amount of
hydrotalcite-like phase is formed and the zincian malachite stays nanocrystalline.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
90
Figure 5-3: Results of XRD full pattern refinement of Cu,Zn,Ga hydroxy carbonate precursors. Top: phase fraction
of zincian malachite (green), aurichalcite (red) and hydrotalcite-like phase (blue); Top: domain size of zincian
malachite (black) and d
value (grey); errors of domain sizes are smaller than used symbols.
An important feature to estimate the potential of a catalyst is the position of the 201
reflection
of the precursor phase zincian malachite in the XRD patterns. It is correlated to the amount of
Zn incorporation in this phase as was first reported by Porta et al.
[34]
. The d
value can be
used as a quantitative measure of the incorporation of non Jahn-Teller-distorted ions like Zn
2+
,
Ga
3+
(or Al
3+
) into this phase. Low d
values correspond to high incorporation (the d
value
of pure malachite, Cu
2
(OH)
2
CO
3
, is about 2.863 Å). Compared to Ga0.0 (2.790 Å), the values
become smaller for the Ga promoted samples (Figure 5-3b) and reach a minimum for Ga2.5 to
Ga3.5 (2.754 Å). This value corresponds to a Zn incorporation of 31% assuming that the
correlation of the binary system shown in
[35-37]
is also valid for the ternary precursors. Again,
Ga1.5 delivers a runaway value. Ga4.0 and Ga5.0 show some scattering but in Ga8.0 and
Ga13.0 the value seems to level off.
Remarkably, the domain (crystallite) sizes of zincian malachite and the Zn (Ga) content in
zincian malachite (obtained from d
values) show a very similar qualitative trend within the
Ga series (Figure 5-3b). Plotting these two features against each other delivers an approximately
linear correlation (Figure 5-4). It can be concluded, that the degree of Zn (Ga) incorporation in
the zincian malachite directly influence the crystallite size and can be controlled by the Ga
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
91
content at a constant Cu:Zn ratio. An inverse correlation is found for the measured BET surface
areas (Figure 5-4). For Ga0.0, the BET surface area is about 60 m
2
g
−1
and is heavily raised for
increasing Ga contents up to 150 m
2
g
−1
for Ga3.0 whereupon a stable value of around
145 m
2
g
−1
is reached for higher Ga contents. Again Ga1.5 does not follow the trend.
Figure 5-4: Zincian malachite domain sizes (black) and BET surface areas (red) of Cu,Zn,Ga hydroxy carbonate
precursors in dependence from Zn (Ga) incorporation into zincian malachite and d
value, respectively.
These results from XRD analysis of the precursors suggest that the limit for the Zn (Ga)
incorporation into the zincian malachite structure is reached for a concentration of the promoter
at around 2.5-4.0 mol% corresponding to a high phase fraction of zincian malachite and a low
crystallite size. Beyond this concentration, the d
value cannot be further decreased. In
contrast, it is slightly increased and settles down at a constant value probably due to competing
incorporation of Zn and Ga in the hydrotalcite-like phase. Thus, low concentrations of Ga
promote the incorporation of the Zn into zincian malachite. With increasing promoter content,
the fraction of Zn-rich aurichalcite is reduced, leading to a decrease of the d
value indicating
a higher Zn (Ga) substitution of zincian malachite. For promoter contents below 2.5 mol%, no
crystalline phases originally containing M
3+
ions were found, suggesting that either Ga
3+
is
present as an amorphous phase which cannot be detected by XRD, or introduced into the
aurichalcite or the zincian malachite structure. Insertion into zincian malachite is known for Al
3+
[17]
and should be transferable to Ga
3+
as well. In this case, the excess of positive charge by Ga
3+
ions on Cu
2+
or Zn
2+
sites has to be compensated. Possible mechanisms are the formation of
cation or proton vacancies. For Ga contents higher than 2.5 mol%, the M
3+
ions probably cannot
be further inserted into the zincian malachite phase due to the charge mismatch.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
92
Unlike the results for Al as promoter
[17]
, the fraction of the hydrotalcite-like phase stays
surprisingly stable at only 2 wt%. No GaOOH was detected. This indicated, that at high Ga
contents, a Ga containing phase is formed which is X-ray amorphous. Using Al as a promoter,
large fractions of a hydrotalcite-like phase also have been observed when exceeding a critical
promoter content
[17]
. The critical promoter contents for high incorporation of Zn and Ga in
zincian malachite and a large fraction of this phase are in the range of 2.5-4.0 mol% for both,
Ga and Al.
5.3.1.2 Scanning electron microscopy
The precursor sample Ga13.0 was investigated with SEM to study the Ga distribution and find
evidence for a segregated Ga-rich phase. As expected, the sample is inhomogeneous. Figure
5-5a shows an overview about a typical region of the sample. The backscattered electron
microscopy image of the same area delivers a mean atomic number contrast (Figure 5-5b) and
shows a brighter area which should correspond to a higher mean atomic number.
Altogether, the predominantly observed morphology of the Cu,Zn,Ga precursor in spot 1
(Figure 5-5c) is consistent with an earlier study on the Cu,Zn,Al system, where zincian
malachite was described as needles of 20 nm × 200 nm
[26]
. Energy-dispersive X-ray
spectroscopy (Figure 5-5e) of six different regions with the mentioned morphology showed an
average local elemental composition of 64 ± 2% (Cu), 25 ± 1% (Zn) and 11 ± 2% (Ga) which
agrees to the results from XRF in Table 5-1 (63:25:12).
But aside this finding, some smaller regions with different morphology, e.g. platelets or
particles, were detected. In spot 2 (Figure 5-5d), no needles were observed but the particle looks
sponge-like. The local composition showed high Ga contents at the expense of Cu which
explains the brighter area of spot 2 in Figure 5-5b. The Ga-richest composition was 33:26:41
where the M
2+
to M
3+
ratio (≈ 3:2) does not fit to that of a hydrotalcite-like phase (3:1). Since
EDX cannot be applied to infinitely small regions, it cannot be finally clarified whether this is a
matter of a single phase or a superposition of a Cu rich phase (zincian malachite) and an
unidentified Ga rich phase. Interestingly, the Zn content in all measured regions was constant
within 24-28 mol% and thus independent of the Cu and Ga fraction.
Figure 5-5: Electron microscopy
images (2.5 keV) and EDX
showing an overview (a), the same overview in backscattered electron mode (b), zincian malachite needles and some
platelets (c), Ga-
rich area (d) and local elemental distribution from SEM
Chapter 5:
Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
images (2.5 keV) and EDX
-results
of the Ga13.0 hydroxy carbonate precursor
showing an overview (a), the same overview in backscattered electron mode (b), zincian malachite needles and some
rich area (d) and local elemental distribution from SEM
-EDX (e).
Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
93
of the Ga13.0 hydroxy carbonate precursor
showing an overview (a), the same overview in backscattered electron mode (b), zincian malachite needles and some
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
94
Table 5-1: Results of precursor sample characterization
Label Sample
number
(internal)
BET
surface
area
[m
2
g
-1
]
Phase composition [wt%] Domain
size
zincian
malachite
[nm]
[Å]
Thermal Analysis
Zincian
malachite Aurichalcite Hydrotalcite-
like
phase
Mass
loss
[wt%]
T
max[a]
[K]
HT-CO
2[b]
[%]
Ga0.0 7399 61 88 12 0 23.1 2.790 27.7 723 52
Ga0.5 10220 88 85 15 0 17.2 2.772 28.4 747 50
Ga1.0 10210 87 84 16 0 15.9 2.772 28.3 760 50
Ga1.5 10222 77 81 19 0 22.4 2.789 28.1 756 52
Ga2.0 10216 118 94 6 0 11.1 2.759 28.3 775 49
Ga2.5 10224 137 100 0 0 9.2 2.754 28.4 789 45
Ga3.0 10212 149 99 0 1 9.3 2.754 28.8 776 45
Ga3.5 10482 146 98 0 2 7.3 2.756 28.3 788 46
Ga4.0 12045 143 98 0 2 6.3 2.763 28.1 785 37
Ga5.0 12063 147 100 0 0 6.9 2.754 27.4 792 39
Ga8.0 12079 143 99 0 1 6 2.761 26.8 783 38
Ga13.0 12158 143 99 0 1 5.8 2.762 26.5 781 31
[a] Temperature of maximum CO
2
emission
[b] CO
2
emission above 673 K relative to overall CO
2
emission according to MS signal
5.3.1.3 Thermal analysis
Thermal decomposition of the precursor is necessary to obtain nano-sized CuO and ZnO
particles. This calcination step can be followed by thermogravimetric measurement (TG)
combined with evolved gas analysis (EGA). In contrast to the calcination, which is performed
up to a temperature of 603 K, the TG-EGA experiments were executed from 303-973 K. The
anions in the Cu,Zn,X hydroxy carbonate are decomposed under emission of water and carbon
dioxide. Exemplarily, the results for Ga2.5 are depicted in Figure 5-6, showing the development
of the mass loss and the normalized H
2
O and CO
2
traces. XRD patterns of the samples thermally
treated at 973 K revealed crystalline CuO and ZnO (not shown). The domain sizes of both
phases decreased by a factor of around three within the Ga series indicating the geometrical
effect of the promoter. The XRD patterns of Ga8.0 and Ga13.0 showed weak and broad
reflections of ZnGa
2
O
4
.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
95
Figure 5-6: TG-MS results of Ga2.5 hydroxy carbonate precursor: mass loss (black), MS traces of H
2
O (blue) and
CO
2
(green).
The theoretical mass losses for pure zincian malachite (Cu
1-x
Zn
x
)
2
(OH)
2
CO
3
and aurichalcite
(Cu
1-y
Zn
y
)
5
(OH)
6
(CO
3
)
2
account for 28% and 26%, respectively, and show only a slight
dependence on the Cu:Zn ratio because of the similar molar masses of Cu and Zn. However, the
Ga incorporation and charge compensation effects in these phases as well as the varying amount
of incorporated or physisorbed water were neglected during this calculation. The theoretical
mass loss for the hydrotalcite-like phase (Cu
1-z
Zn
z
)
6
(Ga)
2
(OH)
16
CO
3
·4H
2
O is 20%. Thus, the
highest mass loss is expected for pure zincian malachite. Indeed, Figure 5-7a shows the highest
mass loss of 28.8% for the sample Ga3.0, which is in the regime of phase-pure zincian
malachite according to XRD. For Ga contents higher than 3 mol%, the mass loss is
monotonically decreasing down to 26.5% (Ga13.0). EGA shows that the decomposition of
Cu,Zn,Ga hydroxy carbonates mainly proceeds in three steps (Figure 5-6). After the release of
physisorbed, incorporated or interlayer water (range I, up to 403 K), the second step is
characterized by simultaneous emission of H
2
O and CO
2
(range II, ca. 403-673 K). In the third
step, only CO
2
is emitted at high temperatures (range III, ca. 673-873 K). The origin of this last
decomposition step is the presence of temperature stable carbonate species (HT-CO
32-
) which
are probably located at the interface between the formed CuO and ZnO
[33, 38]
. The role of this
residual carbonate, which is still present after calcination at not too high temperatures, is
debated and it has been proposed that it can stabilize oxidized copper species in the reduced
catalyst and increase the activity
[39]
. The abundance and the stability of these species are
suggested as measures for the amount and the quality of the interfaces and grain boundaries of
CuO/ZnO aggregates, respectively. Accordingly, pure malachite Cu
2
(OH)
2
CO
3
and hydrozincite
Zn
5
(OH)
6
(CO
3
)
2
do not show the emission of these species
[33, 36, 38]
.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
96
Figure 5-7: TG-MS results of Cu,Zn,Ga hydroxy carbonate precursors: mass loss after heating to 973 K (a),
temperature of highest CO
2
emission rate (b) and CO
2
emission above 673 K relative to overall CO
2
emission
according to MS trace (c).
The HT-CO
32-
decomposition temperature first increases with Ga content from 723 (Ga0.0) up
to a temperature of 792 K (Ga2.5) and then stays nearly constant indicating very stable HT-CO
2
(Figure 5-7b). This indicates stronger interaction of CuO and ZnO with higher Zn (Ga)
incorporation into zincian malachite.
The HT-CO
32-
amount can be calculated from the fraction of HT-CO
2
related to the overall CO
2
emission in a semi-quantitative manner. This fraction is relatively constant around 50% for
Ga0.0 to Ga2.0 and then starts to diminish down to 31% in the Ga series (Figure 5-7c). Thus,
the HT-CO
32-
cannot be intrinsic to the unidentified Ga-phase but originates from the synthetic
Cu,Zn hydroxy carbonate. The decrease rather suggests that the Ga-rich by-phase contains
loosely bound carbonate itself and contributes to the CO
2
emission in range II (Figure 5-6).
These results show that the effect of the Ga promoter on precursor chemistry is also reflected in
the thermal properties of the carbonate, which is still present in the catalyst after calcination.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
97
5.3.2 Calcined samples
5.3.2.1 XRD analysis
Exemplarily, the XRD pattern of Ga2.5 after calcination is shown in Figure 5-8. All patterns
were analyzed by Rietveld refinement (not shown), revealing CuO as the main phase and only
small amounts (up to 1 wt%) of crystalline ZnO. Crystalline ZnGa
2
O
4
spinel, which is
commonly formed during calcination of hydrotalcite phase, was not found. Due to the
homogeneous metal distribution in the precursor, a good dispersion of CuO and ZnO should be
achieved. The CuO domain sizes are listed in Table 5-2. For Ga0.0 it is 4.7 nm and decreases
down to 2.9 nm in the region between Ga2.0 and Ga5.0 where it reaches a minimum of 2.4 nm
for Ga3.0. Higher Ga contents do not much affect the CuO domain size. A roughly inverse trend
is observed for the BET surface areas (Table 5-2). This is expected because small CuO domains
are a result of better nanostructuring which leads to higher BET surface areas.
Figure 5-8: XRD pattern of calcined Ga2.5 sample.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
98
Table 5-2: Results of calcined sample characterization and Cu surface areas
Label Sample
number
(internal)
Ga-
content
[mol%]
(XRF)
Cu:Zn
ratio
(XRF)
BET
surface
area
[m
2
g
-1
]
Domain
size
CuO
[nm]
TPR results Cu surface
area
[m
2
g
CuO-1
]
T
maxa
[K]
H
2
cons.
b
[%]
Ga0.0 7400 0 71:29 110 4.7 481 101 25.3
Ga0.5 10221 0.4 71:29 96 5.3 474 104 38.4
Ga1.0 10211 0.8 71:29 97 3.9 476 95 43.4
Ga1.5 10223 1.2 71:29 91 5.3 475 101 38.2
Ga2.0 10217 1.5 70:30 117 2.9 479 96 47.9
Ga2.5 10225 2 71:29 117 3.3 480 101 48.1
Ga3.0 10213 2.5 71:29 108 2.4 475 96 51.5
Ga3.5 10483 2.9 71:29 115 3.7 474 101 45.7
Ga4.0 12046 3.5 70:30 117 3.3 474 103 51.1
Ga5.0 12065 4.4 71:29 131 2.7 483 100 45.9
Ga8.0 12081 7.2 71:29 118 3.4 480 100 50.6
Ga13.0 12159 12 71:29 126 3.2 476 97 47.8
a
Temperature of the highest H
2
consumption rate according to TCD signal
b
H
2
consumption relative to the amount of CuO contained in the sample
5.3.2.2 The influence of gallia on ZnO
To gain more insight into the state of the Ga promoter in the samples, K-edge X-ray absorption
spectroscopy was applied for selected samples and five oxidic Ga references (α-gallia, β-gallia,
γ-gallia, ZnGa
2
O
4
and ZnO doped with 3 mol% of Ga). To confirm that Ga is incorporated in
ZnO in the ZnO/3%Ga reference sample, UV-Vis measurements were performed with pure ZnO
and ZnO/3%Ga in order to confirm the change of the optic properties due to doping. The band
gap energy (direct transition) of pure ZnO (3.28 eV) was decreased to 3.13 eV for ZnO/3%Ga
(Figure 5-9). This band gap energy reduction was already seen by eye, because ZnO/3%Ga has
a yellow color whereas gallia is white and ZnO only has a slightly yellow touch. Change of
optic properties by doping ZnO with small amounts of Ga or Al has been reported in literature
[40-43]
and naturally is accompanied also by the change of electrical properties. The calcined
sample Ga2.5 was chosen because it was derived from a phase pure precursor material and
should not contain large amounts of segregated Ga-phases. However, the results were found to
be similar to those of Ga3.5 reported in ref.
[17]
, whose precursor was not phase-pure.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
99
Figure 5-9: Determination of absorption edge energies from UV-Vis measurements by the intercept of a linear fit.
The experimental Ga-K-XANES spectrum of Ga2.5 was simulated as a linear combination of
oxidic Ga reference spectra using the software Athena 0.8.061
[31]
according to the method of
least squares fit. The fitting procedure was applied in the fitting region of -20 to 50 eV (related
to the Ga-K-edge of 10367 eV) for all possible 21 combinations of the five references. γ-gallia
seemed not to contribute and was excluded. The results of the remaining 11 fits (Table 5-3)
were sorted by fit quality, represented as R-values, whereas the lowest R-value refers to the best
fit. The E
0
-shifts were smaller than 0.3 eV in all cases. The fit results show that in principle each
of the remaining Ga oxide references (α-gallia, β-gallia, ZnGa
2
O
4
, ZnO doped with 3 mol% Ga)
can be present in the sample Ga2.5 because the R-values of fits 1-4 are similar. But it is
apparent that no satisfying fit is possible without using the Ga-doped ZnO and ZnGa
2
O
4
spinel
reference. Especially simulating the region around 10388 eV needed the contribution of this
reference which is visible in Figure 5-10 showing the best linear combination fit.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
100
Figure 5-10: Ga K-edge XANES of the calcined Ga2.5 sample (black) and results of the linear combination fit in the
range of -20 to 50 eV (related to the Ga-K-edge) using experimental spectra of Ga oxide reference materials.
Table 5-3: Results of linear combination fit of the Ga-K-edge XANES of the calcined Ga2.5 sample by Ga-oxide-
reference spectra (the lowest R-value corresponds to the best fit).
Fit-Nr. R-value (*10
-3
) α-Ga
2
O
3
β-Ga
2
O
3
ZnGa
2
O
4
ZnO/3%Ga
1 0.50 19% 0% 45% 36%
2 0.50 19% - 45% 36%
3 0.55 - 13% 59% 29%
4 0.64 - - 64% 36%
5 1.11 52% - - 48%
6 1.11 52% 0% - 48%
7 1.20 - 35% 65% -
8 1.20 0% 35% 65% -
9 2.49 14% - 86% -
10 3.76 - 47% - 53%
11 4.08 51% 49% - -
In agreement with the results reported in ref.
[17]
, the Ga promoter in the calcined samples seems
to be present in different oxidic species, one of them Ga incorporated into ZnO where Ga
3+
is
tetrahedrally coordinated by oxygen. Hence, Ga is not only increasing the Zn incorporation into
the zincian malachite precursor but also modifies the ZnO component in the resulting catalyst
by partial substitution of Zn
2+
with Ga
3+
. This leads to a change of different properties, like
redox behavior, electronic structure and defect chemistry. The redox properties are changed as it
can be seen from a comparison of TPR profiles whereas pure ZnO shows no signal but in the
case of ZnO doped with 3 mol% Ga probably a small fraction of ZnO is reduced (Figure 5-11).
The change of the electronic structure is obvious from the different band gaps (Figure 5-9). A
modified defect chemistry of doped ZnO has been reported in literature
[44-45]
and is expected to
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
101
be present in our Ga containing samples as well. Altogether, the changed properties will affect
the intrinsic activity of the Cu/Zn(Ga)O probably via modified Cu-ZnO synergy.
Figure 5-11: TPR profiles of ZnO (black) and ZnO/3%Ga (blue).
5.3.2.3 Temperature programmed reduction
The reduction behavior of the calcined samples was investigated by temperature programmed
reduction (TPR) in hydrogen. It was not possible to describe the reduction profiles with a single
peak because at least two shoulders were observed (Figure 5-12). Reasons might be reduction of
CuO in multiple steps
[46]
, reduction of multiple CuO species, e.g. from different precursor
phases
[47]
, or reduction of other components than CuO. The last two possibilities seem unlikely
at least for the catalyst derived for phase pure zincian malachite precursors.
Reduction profiles were analyzed and the results are summarized in Table 5-2. No big change or
clear trend were observed neither for the temperature of the highest H
2
consumption (474-
483 K) nor for the reducibility of CuO (95-104% ± 5%). This is a striking example that the
careful characterization of the early preparation stages may reveal a much greater wealth of
information compared to the later steps of Cu/ZnO catalyst synthesis.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
102
Figure 5-12: TPR profiles of Cu,Zn,Ga calcined samples.
5.3.3 Activated samples
5.3.3.1 Transmission electron microscopy
Selected reduced samples were subjected to TEM analysis: Ga0.0 as reference, Ga2.5, the
sample with the lowest d
value and Ga13.0. In all cases a porous arrangement of roundish
copper particles separated by zinc oxide particles was found indicating that the nanoparticulate
microstructure of the calcined samples was conserved after reduction (Figure 5-13). The Cu
particles sizes were around 11.1 nm (Table 5-4). The lowest value was found for Ga2.5. The
local elemental composition was determined with TEM-EDX at different locations of the
samples. The average results are shown in Table 5-4 and show a good agreement with the
composition obtained by XRF (Table 5-2). Nevertheless, Ga13.0 shows a conspicuously large
standard deviation for elemental distribution of Cu and Ga, which is in agreement with the
findings from SEM of the precursor. From a triangular TEM-EDX composition diagram it
becomes apparent that there is a positive correlation of the Zn and the Ga contents (Figure
5-14a; dotted line). For an independently varying fraction of a pure Ga and a "binary" Cu/Zn
(70:30) component, a negative correlation would be expected (full line). The data points can be
extrapolated to a Zn content of 33% on the binary Zn-Ga line, which equals the spinel
com
position. It can be concluded that the elemental composition of the individual single spots is
a superposition of
two different phases, the first originating from zincian malachite (Cu:Zn =
70:30) and the second is ZnGa
crystallites on the Cu particles (
from separated ZnO and gallia
the local elemental distribution in the precursor
with respect to the tendency of spinel composition, it is quite probable that the precursor of the
spinel was an amorphous Zn,Ga hydrotalcite
unidentified precursor phase.
Figure 5-13:
TEM image of the reduced Ga2.5 sample showing the typical arrangement.
Table 5-4:
Results of reduced sample characterization and activity measurements
Label
TEM results
TEM-EDX
[mol%]
Cu Zn
Ga0.0 69.8 (2.0) 30.2 (2.0)
Ga1.0 - -
Ga2.0 - -
Ga2.5 67.6 (1.6) 30.0 (1.7)
Ga3.0 - -
Ga3.5 - -
Ga13.0 59.7 (9.2) 25.1 (1.7)
Chapter 5:
Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
position. It can be concluded that the elemental composition of the individual single spots is
two different phases, the first originating from zincian malachite (Cu:Zn =
70:30) and the second is ZnGa
2
O
4
spinel (Zn:Ga = 67:33). The latte
r was observed as small
crystallites on the Cu particles (
Figure 5-14b). The formation of ZnGa
2
O
4
by solid state reaction
from separated ZnO and gallia
is unlikely at the low temperatures applied. Although the trend of
the local elemental distribution in the precursor
sample (Figure 5-5e) is
not that
with respect to the tendency of spinel composition, it is quite probable that the precursor of the
spinel was an amorphous Zn,Ga hydrotalcite
-
like phase what would answer the question of the
TEM image of the reduced Ga2.5 sample showing the typical arrangement.
Results of reduced sample characterization and activity measurements
TEM results
Interface ratio
of Cu particles
[%]
MeOH
productivity
(relative)
[%]
Cu particle size
[nm]
Ga
0 11.1 (4.5) 41
- - -
- - -
2.4 (0.1) 10.5 (4.3) 19
- - -
- - -
15.2 (4.4) 11.5 (4.4) 11
Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
103
position. It can be concluded that the elemental composition of the individual single spots is
two different phases, the first originating from zincian malachite (Cu:Zn =
r was observed as small
by solid state reaction
is unlikely at the low temperatures applied. Although the trend of
not that
pronounced
with respect to the tendency of spinel composition, it is quite probable that the precursor of the
like phase what would answer the question of the
MeOH
productivity
(relative)
[%]
Intrinsic
activity
(relative)
[%]
100 100
134 80
151 82
165 90
162 83
163 95
- -
Chapter 5:
Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
104
Figure 5-14:
Results from transmission electron microscopy of reduced Ga,Zn,Ga samples. Ga13.0: elemental
composition from TEM-
EDX showing a strong tendency of ZnGa
particles covered with ZnGa
2
O
4
spinel and ZnO (b). Ga2.5: e
homogeneous distribution (c) and TEM image of crystalline
In contrast, TEM investigation of Ga2.5 showed homogeneous local elemental distribution
(Figure 5-14
c) and images with some small crystallites of
with the XANES results where
is assumed that this phase is formed during precursor calcination by segregation. Another
possibility is the formation of α
Ga. No crystalline ZnGa
2
O
4
spinel was found in HRTEM images.
5.3.3.2 Cu surface areas
Because the overall Cu content in the reduced samples slightly decreases with increasing Ga
content, Cu surface areas from N
Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
Results from transmission electron microscopy of reduced Ga,Zn,Ga samples. Ga13.0: elemental
EDX showing a strong tendency of ZnGa
2
O
4
spinel formation (a) and TEM image of Cu
spinel and ZnO (b). Ga2.5: e
lemental composition from TEM
homogeneous distribution (c) and TEM image of crystalline α
-gallia (d).
In contrast, TEM investigation of Ga2.5 showed homogeneous local elemental distribution
c) and images with some small crystallites of α
-gallia (Figure 5-
14
with the XANES results where α
-
gallia was found to be present according to the two best fits. It
is assumed that this phase is formed during precursor calcination by segregation. Another
possibility is the formation of α
-gallia during reduc
tion which can lead to a separation of Cu and
spinel was found in HRTEM images.
Because the overall Cu content in the reduced samples slightly decreases with increasing Ga
content, Cu surface areas from N
2
O-RFC
were calculated for a better comparison in relation to
Results from transmission electron microscopy of reduced Ga,Zn,Ga samples. Ga13.0: elemental
spinel formation (a) and TEM image of Cu
lemental composition from TEM
-EDX showing
In contrast, TEM investigation of Ga2.5 showed homogeneous local elemental distribution
14
d) in agreement
gallia was found to be present according to the two best fits. It
is assumed that this phase is formed during precursor calcination by segregation. Another
tion which can lead to a separation of Cu and
Because the overall Cu content in the reduced samples slightly decreases with increasing Ga
were calculated for a better comparison in relation to
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
105
the contained CuO before reduction based on XRF under assumption that only CuO, ZnO and
Gallia were present. A pronounced difference of the Cu surface areas of Ga0.0 (25.3 m
2
g
CuO−1
)
and Ga0.5 (38.4 m
2
g
CuO−1
) can be observed in Figure 5-15a. Hence, already small amounts of
Ga improve the Cu dispersion or the gas accessibility. With increasing Ga contents up to
3 mol%, the Cu surface area is increased up to 52 m
2
g
CuO−1
. For Ga contents higher than
3 mol%, the values scatter in the region between 46 and 51 m
2
g
CuO−1
. With regard to the
increased Cu surface areas of Ga2.5 and Ga13.0 compared to Ga0.0 at a very similar particle
size (from TEM), the effect of increased Cu surface area can be explained by less embedment of
the Cu particles in the oxidic phase which results in better gas accessibility due to increased
porosity. A correlation between Cu surface areas and the CuO domain sizes was found for the
Ga containing samples (Figure 5-15b).
Figure 5-15: Cu surface areas of reduced Cu,Zn,Ga samples with respect to the calcined sample mass (a) and
contained CuO mass (b). The error was estimated to be ± 1 m
2
g
−1
in the top graph. CuO domain sizes of the calcined
samples are given for comparison (c). Errors for domain sizes were smaller than the used symbols.
From the TEM Cu particle size and known Cu content, a theoretical Cu surface area (assuming
isolated roundish particles) was calculated. Comparison with experimentally obtained Cu
surface areas from N
2
O-RFC delivered the interface ratio of Cu particles and indicated their
degree of embedment. Therein, contact with other Cu particles or the oxidic matrix is possible.
Although the absolute values are very inaccurate, the trend of decreasing interface ratio (equals
decreased embedment) with increasing Ga content is discernable (Table 5-4).
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
106
5.3.4 Methanol synthesis activity
The activity in methanol synthesis was measured for selected Ga samples. The promoting effect
of Ga on the catalytic performance is clearly visible: Starting from Ga0.0 (Figure 5-16), the
activity increased and reached a maximum at Ga2.5 which is 60% higher than the value for the
unpromoted sample Ga0.0. After this, the activity did not change significantly for Ga contents
up to 3.5 mol%.
Figure 5-16: Methanol productivity of selected Cu/ZnO/Ga
2
O
3
catalysts in methanol synthesis relative to the
unpromoted sample Ga0.0 (blue bars) and content of non-Jahn-Teller ions (Zn
2+
and Ga
3+
) in the zincian malachite
structure (red points) calculated from d
values of the binary Cu,Zn system according to
[35]
.
The corresponding d
values of the precursors are converted into contents of non-Jahn-Teller-
ions (Zn
2+
and Ga
3+
) in the zincian malachite according to the correlation found for the binary
Cu,Zn system
[35]
. These values are added in Figure 5-16 and show the same trend like the
activities so that a linear correlation can be assumed.
The correlation between activity and Cu surface area (not shown), however, shows little but
significant scattering. This can be explained by different intrinsic activities of the exposed Cu
surface areas, calculated by dividing the activity by the Cu surface area (Figure 5-17). The
highest intrinsic activity is found for Ga0.0. In principle, the density of catalytically active sites
on the surface of the Cu particles is described and different intrinsic activities have often been
reported for different Cu based catalytic systems
[11, 48]
.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
107
Figure 5-17: Intrinsic activities (related to the Cu surface area) of Cu/ZnO/Ga
2
O
3
catalysts in methanol synthesis
relative to the unpromoted sample Ga0.0.
In the case of Ga, the promoting effect, which leads to increased activity, can only be explained
by the strong increase of the accessible Cu surface area up to 52 m
2
g
CuO−1
(Ga3.0) compared to
25 m
2
g
CuO−1
(Ga0.0) which means a doubling. This geometrical effect is promoted by Cu
2+
dilution in the zincian malachite precursor with non-reducible cations. On the other hand, the
overall activity is only improved by 60% (Ga3.0) which implies a lower intrinsic activity
compared to Ga0.0. This effect is probably related to the modification of the ZnO phase by
incorporated Ga which might have a negative effect on Cu-ZnO synergy. Another explanation
would be that the intrinsic activity is related to the degree of embedment. Less embedment
means also less contact to the ZnO phase. The generation of active centers which require the
presence of ZnO is probably reduced.
The measured activity has to be regarded as a convolution of Cu surface area and intrinsic
activity. In this sample series, the resulting improved activity due to Ga promotion is achieved
due to a beneficial geometrical dispersion effect although the intrinsic activity is lowered. Both,
geometrical and synergetic function of Ga promotion can be traced back to incorporation of
Ga
3+
in the zincian malachite precursor, explaining the correlation of d
and activity data.
5.4 Conclusions
In summary, it was shown that Ga
3+
, similar to Al
3+
, has a promoting effect on Cu/ZnO
methanol synthesis catalysts. The promoting effect is most effective if only low amounts of Ga
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
108
– around 3 mol% – are added to the preparation. Higher Ga loadings lead to a Ga segregation
and formation of amorphous ZnGa
2
O
4
spinel in the final catalyst. Thus, the microstructure of
the catalyst becomes inhomogeneous with Ga-rich domains of presumably lower catalytic
activity. Low amounts of Ga
3+
can be incorporated into the zincian malachite precursor phase
and lead to an increase in methanol synthesis activity of 60% compared to the binary Ga free
system. The promoting effect is mainly traced back to a geometric contribution: The Cu
dispersion is increased by Ga due to better dilution of Cu
2+
in the precursor by Zn
2+
and Ga
3+
,
leading to a better interdispersion of metallic and oxidic components and a higher porosity of
the resulting Cu/ZnO aggregates. This effect leads to an increase of the accessible Cu surface
area by 100%. Apparently, the presence of Ga slightly diminishes the intrinsic activity of the
exposed Cu surface area . This effect is probably related to the fraction of Ga species that are
found to be incorporated into the ZnO phase after calcination of the precursor, leading to a
modification of the properties of ZnO associated with a negative effect on the well-known Cu-
ZnO synergy. The lack of synergy might also be enhanced by less embedment of the Cu
particles leading to less interface contact with ZnO. The functionality of Ga promotion depends
critically on the homogeneous distribution of Ga. The best distribution is achieved by
incorporation into the zincian malachite precursor phase and a linear correlation of the (Zn,Ga)
content in this phase with the catalytic activity of the final catalyst was observed.
Acknowledgement
Frank Girgsdies (help with XRD pattern analysis), Edith Kitzelmann (XRD measurements),
Achim Klein-Hoffmann and Olaf Timpe (XRF), Gisela Lorenz (BET measurements), Genka
Tzolova-Müller (UV-Vis), Antje Ota and Liandi Li for providing the Ga oxide reference
samples, Juliane Scholz, Alexander Müller and Anke Walter for help with XAS measurement
are acknowledged. Stefanie Kühl is acknowledged for valuable discussions, HASYLAB/DESY
(Hamburg) for allocation of beamtime. Financial support was given by the German Federal
Ministry of Education and Research (BMBF, FKZ 01RI0529, 2005-2008) and the STMWFK
(NW-0810-0002, since 2010). Robert Schlögl is greatly acknowledged for valuable discussions
and his continuous support.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
109
5.5 References
[1] D. Waller, D. Stirling, F. S. Stone, M. S. Spencer, Faraday Discuss. 1989, 87, 107-120.
[2] K. C. Waugh, Catal. Today 1992, 15, 51-75.
[3] J. B. Hansen, P. E. H. Nielsen, in Handbook of Heterogeneous Catalysis, Vol. 6 (Eds.:
G. Ertl, G. Knözinger, F. Schüth, J. Weitkamp), Wiley-VCH, Weinheim 2nd ed., 2008,
pp. 2920-2949.
[4] G. C. Chinchen, K. C. Waugh, D. A. Whan, Appl. Catal. 1986, 25, 101-107.
[5] T. Fujitani, J. Nakamura, Catal. Lett. 1998, 56, 119-124.
[6] R. Burch, R. J. Chappell, S. E. Golunski, Catal. Lett. 1988, 1, 439-443.
[7] Y. Kanai, T. Watanabe, T. Fujitani, M. Saito, J. Nakamura, T. Uchijima, Catal. Lett.
1994, 27, 67-78.
[8] R. G. Herman, K. Klier, G. W. Simmons, B. P. Finn, J. B. Bulko, T. P. Kobylinski, J.
Catal. 1979, 56, 407-429.
[9] J. C. Frost, Nature 1988, 334, 577-580.
[10] M. Behrens, F. Studt, I. Kasatkin, S. Kühl, M. Hävecker, F. Abild-Pedersen, S. Zander,
F. Girgsdies, P. Kurr, B.-L. Kniep, M. Tovar, R. W. Fischer, J. K. Nørskov, R. Schlögl,
Science 2012, 336, 893-897.
[11] M. Kurtz, N. Bauer, C. Buscher, H. Wilmer, O. Hinrichsen, R. Becker, S. Rabe, K.
Merz, M. Driess, R. A. Fischer, M. Muhler, Catal. Lett. 2004, 92, 49-52.
[12] J. D. Grunwaldt, A. M. Molenbroek, N. Y. Topsoe, H. Topsoe, B. S. Clausen, J. Catal.
2000, 194, 452-460.
[13] R. N. d'Alnoncourt, X. Xia, J. Strunk, E. Löffler, O. Hinrichsen, M. Muhler, Phys.
Chem. Chem. Phys. 2006, 8, 1525-1538.
[14] M. Saito, T. Fujitani, M. Takeuchi, T. Watanabe, Appl. Catal. A 1996, 138, 311-318.
[15] M. V. Twigg, M. S. Spencer, Top. Catal. 2003, 22, 191-203.
[16] M. Kurtz, H. Wilmer, T. Genger, O. Hinrichsen, M. Muhler, Catal. Lett. 2003, 86, 77-
80.
[17] M. Behrens, S. Zander, P. Kurr, N. Jacobsen, J. Senker, G. Koch, T. Ressler, R. W.
Fischer, R. Schlögl, in Performance Improvement of Nano-Catalysts by Promoter-
Induced Defects in the Support Material: Methanol Synthesis over Cu/ZnO:Al,
submitted to J. Am. Chem. Soc.
[18] J. Sloczynski, R. Grabowski, A. Kozlowska, P. Olszewski, M. Lachowska, J. Skrzypek,
J. Stoch, Appl. Catal. A 2003, 249, 129-138.
[19] S. D. Jones, H. E. Hagelin-Weaver, Appl. Catal. B 2009, 90, 195-204.
[20] J. P. Breen, J. R. H. Ross, Catal. Today 1999, 51, 521-533.
[21] Y. Nitta, O. Suwata, Y. Ikeda, Y. Okamoto, T. Imanaka, Catal. Lett. 1994, 26, 345-354.
[22] S. K. Ihm, S. W. Baek, Y. K. Park, J. K. Jeon, ACS Sym. Ser. 2003, 852, 183-194.
[23] C. Baltes, S. Vukojevic, F. Schüth, J. Catal. 2008, 258, 334-344.
[24] P. Kurr, I. Kasatkin, F. Girgsdies, A. Trunschke, R. Schlögl, T. Ressler, Appl. Catal. A
2008, 348, 153-164.
[25] H. Fan, H. Zheng, H. Li, Front. Chem. Eng. China 2010, 4.
[26] M. Behrens, J. Catal. 2009, 267, 24-29.
[27] A. Coelho, 4.2 ed., Bruker AXS GmbH, Karlsruhe, Germany, 2003-2009.
[28] D. A. M. Monti, A. Baiker, J. Catal. 1983, 83, 323-335.
[29] P. Malet, A. Caballero, J. Chem. Soc. Faraday T. 1 1988, 84, 2369-2375.
[30] D. G. Barton, M. Shtein, R. D. Wilson, S. L. Soled, E. Iglesia, J. Phys. Chem. B 1999,
103, 630-640.
[31] B. Ravel, M. Newville, J. Synchrotron Rad. 2005, 12, 537-541.
[32] M. S. Spencer, Catal. Lett. 2000, 66, 255-257.
Chapter 5: Promoting Methanol Synthesis Catalysts:
Correlations between Microstructure and Activity in Cu/ZnO/Ga
2
O
3
110
[33] B. Bems, M. Schur, A. Dassenoy, H. Junkes, D. Herein, R. Schlögl, Chem. Eur. J.
2003, 9, 2039-2052.
[34] P. Porta, S. Derossi, G. Ferraris, M. Lojacono, G. Minelli, G. Moretti, J. Catal. 1988,
109, 367-377.
[35] S. Zander, B. Seidlhofer, M. Behrens, submitted to Inorg. Chem.
[36] M. Behrens, F. Girgsdies, A. Trunschke, R. Schlögl, Eur. J. Inorg. Chem. 2009, 1347-
1357.
[37] M. Behrens, F. Girgsdies, Z. Anorg. Allg. Chem. 2010, 636, 919-927.
[38] G. J. Millar, I. H. Holm, P. J. R. Uwins, J. Drennan, J. Chem. Soc. Faraday Trans.
1998, 94, 593-600.
[39] L. M. Plyasova, T. M. Yureva, T. A. Kriger, O. V. Makarova, V. I. Zaikovskii, L. P.
Soloveva, A. N. Shmakov, Kinet. Catal. 1995, 36, 425-433.
[40] H. Gomez, M. de la L. Olvera, Mat. Sci. Eng. B Solid 2006, 134, 20-26.
[41] L. Zhao, G. Shao, S. Song, X. Qin, S. Han, Rare Metals 2011, 30, 175-182.
[42] M. Zhou, H. Zhu, Y. Jiao, Y. Rao, S. Hark, Y. Liu, L. Peng, Q. Li, J. Phys. Chem. C
2009, 113, 8945-8947.
[43] J. D. Ye, S. L. Gu, S. M. Zhu, S. M. Liu, Y. D. Zheng, R. Zhang, Y. Shi, Appl. Phys.
Lett. 2005, 86.
[44] N. Roberts, R. P. Wang, A. W. Sleight, W. W. Warren, Phys. Rev. B 1998, 57, 5734-
5741.
[45] H. Matsui, H. Saeki, H. Tabata, T. Kawai, J. Electrochem. Soc. 2003, 150, G508-G512.
[46] M. M. Günter, B. Bems, R. Schlögl, T. Ressler, J. Synchrotron Rad. 2001, 8, 619-621.
[47] G. Fierro, M. LoJacono, M. Inversi, P. Porta, F. Cioci, R. Lavecchia, Appl. Catal. A
1996, 137, 327-348.
[48] M. Behrens, A. Furche, I. Kasatkin, A. Trunschke, W. Busser, M. Muhler, B. Kniep, R.
Fischer, R. Schlögl, Chem. Cat. Chem. 2010, 2, 816-818.
Chapter 6: Final Summary and Conclusion
111
Chapter 6: Final Summary and Conclusion
The results presented in this work give insights into preparation of methanol synthesis catalysts
by systematic investigation of the precursor chemistry. In particular, they allow to better
understand the crucial role of the zincian malachite, (Cu,Zn)
2
(OH)
2
CO
3
, precursor phase and its
properties on the performance of the resulting catalysts.
Zincian malachite was prepared by co-precipitation, followed by aging, filtrating, washing and
drying. Subsequent calcination and reduction led to the catalytically active Cu/ZnO catalyst. Co-
precipitation (Cu:Zn = 70:30) was performed in a pH- and temperature-controlled (338 K)
manner and enabled homogeneous distribution of the metal ions in the amorphous suspended
solid, a Cu,Zn hydroxide carbonate, which transformed into crystalline product during aging.
The influence of synthesis parameters, especially in the early stages of catalyst preparation was
investigated.
The aging step of Cu,Zn hydroxy carbonates is critical with regard to the incorporation of Zn
into zincian malachite and was investigated by in-situ energy dispersive x-ray diffraction and in-
situ UV-Vis spectroscopy. To study the aging process independently, it had to be decoupled
from the prior co-precipitation step by a continuous preparation. The obtained “unaged”
amorphous precursor phase was transformed to crystalline zincian malachite under controlled
conditions by aging in solutions of similar composition to the mother liquor. By varying the pH-
values (5.0-8.0), two different aging mechanisms were found. Low pH-values (5.0-6.5) showed
direct co-condensation of Cu
2+
and Zn
2+
. This mechanism led to higher Zn incorporation as
indicated by the shifted position of the 201
reflection. The second pathway was observed for pH
≥ 7 and showed simultaneous initial crystallization of Cu-rich malachite and a transient Zn-
storage phase, sodium zinc carbonate. This intermediate re-dissolved and allowed for
enrichment of Zn into malachite at pH ≥ 7 at later stages of aging. As a function of different
aging conditions, a variation of the Zn content in zincian malachite between ca. 24 and 29% was
observed despite the same nominal Zn-content in the starting material of 30% indicating that a
varying fraction of Zn was present in an undetected phase “Zn↓” acting as a sink for Zn.
Variation of temperature (at pH 7) only led to gradual changes. Thus, the acidity of the aging
medium was identified as the most critical synthesis parameter to determine the final Zn-content
in zincian malachite. Interestingly, Zn incorporation was independent of the crystallization
mechanism. Even in the absence of Na
+
, suppressing the transient crystallization of the sodium
zinc carbonate storage phase, a lower degree of Zn incorporation was observed in the final
Chapter 6: Final Summary and Conclusion
112
sample at pH 7, although the reaction was following the direct co-condensation mechanism. The
effect of individual synthesis parameters like temperature or acidity during catalyst preparation
can be rationalized on basis of the complex chemistry of precursor aging: They should be
optimized to give a low amount of Zn↓ and a maximal Zn-substitution in malachite approaching
the nominal Cu:Zn ratio of the synthesis.
Application of different pH-values in the range of pH 6-9 during co-precipitation and aging in a
batch synthesis also has a highly reproducible influence on the precursor chemistry. Rietveld
refinement was performed on the XRD patterns of the precursors. For pH 6.0, large fractions of
the undesired Zn-rich by-phase aurichalcite were found besides zincian malachite as the main
phase. Application of pH values ≥ 6.5 led to higher phase fraction of zincian malachite at the
expense of aurichalcite with the consequence, that more Zn was introduced into the zincian
malachite phase. Samples prepared at pH 7.5 and higher showed a split up signal of the 201
reflection indicating inhomogeneous distribution of Zn within two different zincian malachite
phases. Samples prepared at 6.0 ≤ pH ≤ 7.0 showed a better homogeneity of the Zn distribution.
Thus, precursors in this sample series can be characterized by the degree of Zn incorporation
into the zincian malachite phase and also the homogeneity of the Zn distribution within this
compound. The largest CuO domain sizes were found for calcined samples prepared at pH 6.0.
Cu surface areas, which are a prerequisite for the performance of the reduced Cu/ZnO catalysts
revealed similar values in the range of 18 to 20 m
2
g
-1
. Only samples prepared at pH 8.5 showed
a larger Cu surface area of around 25 m
2
g
-1
.
The abovementioned results revealed the complexity of the interplay of synthesis parameters
during catalyst preparation by co-precipitation. The properties of the precursor materials
obtained by aging of the co-precipitate influence the structural properties, which in turn will
affect the performance of the final catalysts. Unfortunately, directly tracking back the catalytic
performance to the synthesis pH in a simple synthesis parameter–structure–performance
relationship is difficult as variation of the parameter pH induced numerous simultaneous
changes in the precursor material that lead to different and partially compensating effects on the
resulting catalyst.
ZnO is known to act as a spacer for the single Cu particles in the Cu/ZnO catalyst and to enable
Cu-ZnO synergy which beneficially affects the activity. MgO was investigated to act as a
substitute for ZnO. The geometric influence turned out to be better compared to ZnO but the
synergetic effect of Cu and ZnO during methanol synthesis from CO
2
/CO/H
2
was unequaled.
Both geometric and synergetic effects were combined by preparation of Cu/MgO/ZnO sample
Chapter 6: Final Summary and Conclusion
113
which exhibited a higher activity than Cu/ZnO and Cu/MgO. Changing the feed gas to CO/H
2
,
Cu/MgO was most active.
Industrial methanol synthesis catalysts are promoted by small amounts of refractory oxide,
typically Al
2
O
3
. In the present thesis, the effect of Ga as a promoter in the Cu,Zn catalytic
system for methanol synthesis was investigated by preparing a sample series with increasing Ga
concentration. Already small Ga contents up to 3 mol% changed the characteristics of the
samples dramatically. Despite the charge mismatch, some Ga
3+
was incorporated into the
zincian malachite precursor phase. Higher Ga loadings led to a Ga segregation and formation of
amorphous ZnGa
2
O
4
spinel in the final catalyst. Thus, the microstructure of the catalyst became
inhomogeneous with Ga-rich domains of presumably lower catalytic activity. After calcination,
some of the Ga was incorporated in the ZnO which was verified by X-ray absorption near edge
structure spectroscopy. After reduction, the Cu surface area was doubled and the methanol
synthesis activity increased by 60% compared to the binary Ga free Cu,Zn reference system.
The promoting effect was mainly traced back to a geometric contribution: The Cu dispersion
was increased by Ga due to better dilution of Cu
2+
in the precursor by Zn
3+
and Ga
3+
, leading to
a better interdispersion of metallic and oxidic components and a higher porosity of the resulting
Cu/ZnO aggregates. This effect led to an increase of the accessible Cu surface area. Apparently,
the intrinsic activity of the exposed Cu surface area was lowered by the presence of Ga. This
effect is related to the fraction of Ga species incorporated into the ZnO phase after calcination of
the precursor, leading to a modification of the properties of ZnO associated with a negative
effect on the well-known Cu-ZnO synergy. This might also be enhanced by less embedment of
the Cu particles leading to less interface contact with ZnO. The functionality of Ga promotion
depended critically on the homogeneous distribution of Ga. The best distribution was achieved
by incorporation into the zincian malachite precursor phase and a linear correlation of the
(Zn,Ga) content in this phase with the catalytic activity of the final catalyst was observed.
In summary, these systematic studies on an applied and highly complex catalytic system like
Cu/ZnO/X provide a better understanding of the chemistry underlying catalyst preparation and,
most important, reveal relationships between synthesis parameters, microstructure and activity
that help to explain the role of the structural promoter phase X. These findings shall not only
contribute to the fundamental knowledge about this catalyst, but also guide the way to a more
rational catalyst design in the future.
115
Appendix
xxi
Appendix
Curriculum vitae
Personal information
Name: Stefan Zander (né: Kißner)
Date of birth: 27.05.1980
Place of birth: Berlin, Germany
Family status: married, 1 child
Education
08/2008 - 08/2012 Ph.D. Thesis
Fritz-Haber-Institut der Max-Planck-Gesellschaft in Berlin, Department
of Inorganic Chemistry, Prof. Dr. Robert Schlögl, Title: “Preparation
and Characterization of Cu/ZnO Catalysts for Methanol Synthesis”
(including two months of paternity leave).
08/2007 - 11/2007 Scientific assistant
Technical University Berlin, Institute of Technical Chemistry, Prof. Dr.
Marion Ansorge-Schumacher
12/2006 - 07/2007 Diploma thesis
Technical University Berlin, Institute of Technical Chemistry, Prof. Dr.
Marion Ansorge-Schumacher, Title: “Assayentwicklung für die
biokatalysierte Abspaltung von Schutzgruppen”
10/1999 - 07/2007 Study of chemistry
Technical University Berlin - qualification: Diplom-Chemiker
Appendix
xxii
Publications
M. Behrens, S. Kißner, F. Girgsdies, I. Kasatkin, F. Hermerschmidt, K. Mette, H. Ruland, M.
Muhler, R. Schlögl
“Knowledge-based development of a nitrate-free synthesis route for Cu/ZnO methanol synthesis
catalysts via formate precursors”
Chem. Comm. 2011, 47, 1701-1703.
M. Behrens, D. Brennecke, F. Girgsdies, S. Kißner, A. Trunschke, N. Nasrudin, S. Zakaria, N.
F. Idris, S. B. A. Hamid, B. Kniep, R. Fischer, W. Busser, M. Muhler, R. Schlögl
“Understanding the complexity of a catalyst synthesis: Co-precipitation of mixed Cu,Zn,Al
hydroxycarbonate precursors for Cu/ZnO/Al
2
O
3
catalysts investigated by titration experiments”
Appl. Catal. A, 2011, 392, 93-102.
M. Behrens, F. Studt, I. Kasatkin, S. Kühl, M. Hävecker, F. Abild-Pedersen, S. Zander, F.
Girgsdies, P. Kurr, B. Kniep, M. Tovar, R. W. Fischer, J. K. Nørskov, R. Schlögl
“The Active Site of Methanol Synthesis over Cu/ZnO/Al
2
O
3
Industrial Catalysts”
Science, 2012, 336, 893-897.
S. Zander, B. Seidlhofer, M. Behrens
“In-situ EDXRD Study of the Chemistry of Aging of Co-precipitated Mixed Cu,Zn
Hydroxycarbonates – Consequences for the Preparation of Cu/ZnO Catalysts”
Dalton T., 2012, 41, 13413-13422.
S. Zander, E. Kunkes, M. E. Schuster, J. Schumann, G. Weinberg, R. Schlögl, M. Behrens
“Development of Cu-Catalysts for Methanol Synthesis from CO
2
and CO”
(to be submitted)
S. Zander, J. Schumann, I. Kasatkin, G. Weinberg, G. Koch, T. Ressler, P. Kurr, B. Kniep, M.
Behrens
“Methanol Synthesis Promoting Catalysts: Correlations between Microstructure and Activity in
Cu/ZnO/Ga
2
O
3
”
(to be submitted)
S. Zander, I. Kasatkin, G. Weinberg, P. Kurr, B. Kniep, M. Behrens
Appendix
xxiii
“Correlations between Microstructure and Activity of Cu/ZnO Catalysts for Methanol Synthesis
- Influence of the pH value during Synthesis of Cu,Zn Hydroxy Carbonates”
(to be submitted)
Appendix
xxiv
Poster presentations
EuropaCat X, „Catalysis: across the disciplines“ (Glasgow, Scotland) 28.08.-02.09.2011
„Preparation and Characterization of Cu/ZnO/Ga
2
O
3
Catalysts”
S. Zander, I. Kasatkin, G. Koch, T. Ressler, P. Kurr, R. Schlögl, M. Behrens
44. Jahrestreffen Deutscher Katalytiker (Weimar, Germany) 16.-18.03.2011
„Cu/ZnO/Ga
2
O
3
Catalysts Preparation for Methanol Synthesis “
S. Zander, B. Zhang, R. Schlögl, M. Behrens
GDCh Tagung, Materialkonzepte für Katalyse und Sensorik, (Berlin, Germany) 20.-22.09.2010
„Cu/ZnO catalyst preparation via formate precursors“
S. Kißner, F. Hermerschmidt, K. Mette, F. Girgsdies, B. Zhang, R. Schlögl, M. Behrens
43. Jahrestreffen Deutscher Katalytiker (Weimar, Germany) 10.-12.03.2010
„Cu/ZnO catalyst preparation via formate precursors“
Stefan K., F. Hermerschmidt, K. Mette, F. Girgsdies, R. Schlögl, M. Behrens
„Neutron Diffraction Study of different Cu based methanol synthesis catalysts“
S. Kühl, S. Kißner, F. Girgsdies, M. Tovar, D. Wallacher, M. Behrens
SNI 2010 – Deutsche Tagung für Forschung und Synchrotronstrahlung, Neutronen und
Ionenstrahlen an Großgeräten (Berlin, Germany) 24.-26.02.2010
„Neutronenbeugung an Cu basierten Methanol-Synthese-Katalysatoren“
S. Kühl, S. Kißner, F. Girgsdies, M. Tovar, D. Wallacher, M. Behrens
3rd Workshop on Industry-Academia partnership in catalysis (Berlin, Germany) 26.-27.10.2009
„New insights into the precursor chemistry of Cu/ZnO/Al
2
O
3
catalysts - decoupling
precipitation and aging“
S. Kißner, B. Seidlhofer, W. Bensch, R. Schlögl, M. Behrens
EuropaCat IX „Catalysis for a sustainable world“ (Salamanca, Spain) 30.08.-04.09.2009
„New insights into the precursor chemistry of Cu/ZnO/Al
2
O
3
catalysts - decoupling
precipitation and aging“
S. Kißner, B. Seidlhofer, W. Bensch, M. Behrens
Appendix
xxv
42. Jahrestreffen Deutscher Katalytiker (Weimar, Germany) 11.-13.03.2009
„Preparation of Cu/ZnO/Al
2
O
3
catalysts applying continuous precipitation“
S. Kißner, M. Behrens, R. Schlögl
