Stable amorphous georgeite as a precursor to a high-activity catalyst

Article abstract of DOI:10.1038/nature16935

Copper and zinc form an important group of hydroxycarbonate minerals that include zincian malachite, aurichalcite, rosasite and the exceptionally rare and unstable - and hence little known and largely ignored - georgeite. The first three of these minerals are widely used as catalyst precursors for the industrially important methanol-synthesis and low-temperature water-gas shift (LTS) reactions, with the choice of precursor phase strongly influencing the activity of the final catalyst. The preferred phase is usually zincian malachite. This is prepared by a co-precipitation method that involves the transient formation of georgeite; with few exceptions it uses sodium carbonate as the carbonate source, but this also introduces sodium ions - a potential catalyst poison. Here we show that supercritical antisolvent (SAS) precipitation using carbon dioxide (refs 13, 14), a process that exploits the high diffusion rates and solvation power of supercritical carbon dioxide to rapidly expand and supersaturate solutions, can be used to prepare copper/zinc hydroxycarbonate precursors with low sodium content. These include stable georgeite, which we find to be a precursor to highly active methanol-synthesis and superior LTS catalysts. Our findings highlight the value of advanced synthesis methods in accessing unusual mineral phases, and show that there is room for exploring improvements to established industrial catalysts.

Full text of DOI:10.1038/nature16935

LEttER
doi:10.1038/nature16935
Stable amorphous georgeite as a precursor to a
high-activity catalyst
1
1
2,3
4,5
1
1
Simꢀꢁ A. Kꢀꢁdraꢂ , Paul J. Smiꢂꢃ , Peꢂer P. Wells , Pꢃilip A. Cꢃaꢂer , James h. Carꢂer , David J. Mꢀrgaꢁ ,
6
6
1,5
7
1
1
Elisabeꢂꢂa M. Fiꢀrdalisꢀ , Jakꢀb B. Wagꢁer , tꢃꢀmas E. Davies , Li Lu , Jꢀꢁaꢂꢃaꢁ K. Barꢂley , Sꢂuarꢂ h. taylꢀr ,
Micꢃael S. Speꢁcer , Cꢃrisꢂꢀpꢃer J. Kiely , Gꢀrdꢀꢁ J. Kelly , Cꢀliꢁ W. Park , Maꢂꢂꢃew J. Rꢀsseiꢁsky & Graꢃam J. huꢂcꢃiꢁgs
1
7
8
8
5
1
Copper and zinc form an important group of hydroxycarbonate georgeite that is effectively dry from the point of precipitation and
minerals that include zincian malachite, aurichalcite, rosasite and thus has limited contact with water (which facilitates ageing and trans-
the exceptionally rare and unstable—and hence little known and formation to malachite).
1
largely ignored —georgeite. The first three of these minerals are
When we used mixed copper and zinc acetate solutions (with a 2/1
2–4
widely used as catalyst precursors for the industrially important or 1/1 copper/zinc ratio) for the SAS precipitation, zincian georgeite
methanol-synthesis and low-temperature water–gas shift (LTS) (with a 2/1 or 1/1 copper/zinc ratio, as judged by ICP-MS) formed
5–7
reactions , with the choice of precursor phase strongly influencing with a density, IR spectrum and XRD pattern nearly identical to those
2,3,8–10
the activity of the final catalyst. The preferred phase
is usually of copper-only georgeite (Extended Data Figs 1 and 2); no additional
zincian malachite. This is prepared by a co-precipitation method phases—such as smithsonite (ZnCO3)—were discernible by XRD.
1
1
that involves the transient formation of georgeite ; with few Scanning transmission electron microscopy (STEM) analysis of zin-
12
exceptions it uses sodium carbonate as the carbonate source, but cian georgeite revealed predominantly amorphous material, although a
crystallites with diameters less than
this also introduces sodium ions—a potential catalyst poison. Here small fraction (<10 vol%) of CuO
x
we show that supercritical antisolvent (SAS) precipitation using 2nanometres were observed (Extended Data Fig. 3; analysis of geor-
carbon dioxide (refs 13, 14), a process that exploits the high diffusion geite itself was not possible because of its instability under vacuum). In
rates and solvation power of supercritical carbon dioxide to rapidly addition, the SAS-derived zincian georgeite had few impurities (such
expand and supersaturate solutions, can be used to prepare copper/ as sodium ions; Extended Data Fig. 2).
zinc hydroxycarbonate precursors with low sodium content. These
We therefore found that it was possible to synthesize stable zincian
include stable georgeite, which we find to be a precursor to highly georgeite by SAS in sufficient quantities to evaluate the catalytic per-
active methanol-synthesis and superior LTS catalysts. Our findings formance of copper/zinc oxide, derived from zincian georgeite, in the
highlight the value of advanced synthesis methods in accessing methanol-synthesis and LTS reactions. As a control, we synthesized
unusual mineral phases, and show that there is room for exploring a zincian-malachite precursor—considered the optimal precursor
3
improvements to established industrial catalysts.
for model copper/zinc-oxide catalysts —by co-precipitation (see
For the SAS precipitation process, we initially used copper(ꢀꢀ) Methods). The catalysts used here were not stabilized with alumina,
17,18
acetate solutions (Extended Data Fig. 1) with ethanol as the solvent; an important promoter and stabilizer of the commercial catalyst
.
the result was amorphous copper(ꢀꢀ) acetate, which we character- Before use, the precursors were subjected to calcination to remove the
ized by X-ray diffraction (XRD) and Fourier transform-infrared carbonate, followed by in situ reduction to form the active catalyst. The
(
FT-IR) spectroscopy. Adding 5 vol% water as a co-solvent produced optimal calcination temperature was determined using thermal gravi-
a blue precipitate with an IR spectrum exhibiting a broad OH band metric analysis (TGA) with evolved gas analysis (EGA), which showed
−
1
2–
−1
at 3,408cm and CO
3
bands at 1,470, 1,404 and 829cm , which SAS-prepared georgeite to exhibit three distinct mass losses, and mala-
are inconsistent with malachite but closely match the spectrum iden- chite to exhibit a single mass loss, associated with concurrent removal
1
tifying the rare georgeite phase . Helium pycnometry gave a density of water and carbon dioxide (Fig. 1). The multistep decomposition of
−3
of 3.1gcm for the SAS-prepared phase; as with mineralogical geor
-
georgeite and zincian georgeite (the 1/1 copper/zinc sample shown in
geite, the SAS-prepared phase was amorphous by XRD. We further Extended Data Fig. 2) can be separated into three regions: water loss
studied the precipitate by inductively coupled plasma mass spec- at 80–100°C, water and carbon dioxide loss at 190–210°C, and carbon
trometry (ICP-MS) and carbon–hydrogen–nitrogen (CHN) analysis dioxide loss at 315–420°C (carbonate decomposition). On the basis of
(
Extended Data Fig. 2), which gave an ideal formula for the precipi- this analysis, the precursors were calcined at 300°C in air for 6hours:
.
tate of Cu
7
(CO
3
)
5
(OH)
4
5H
2
O. We regard this to be a close match to this is below the temperature associated with the evolution of carbon
.
1
the Cu
5
(CO
3
)
3
(OH)
4
6H
2
O formula reported for georgeite , taking dioxide, and residual carbonate species have been shown to improve
19
into account that slight variations in composition are anticipated for the activity of methanol-synthesis catalysts . The calcined material
1
5
an amorphous phase . The SAS-prepared and mineralogical forms was then reduced in situ in dilute hydrogen at 225°C.
2
−
2+
of georgeite both have a higher CO
3
/Cu molar ratio than mal-
/Cu values being 0.7, 0.6 and 0.5, respectively); mercial catalysts, showing that the catalyst derived from SAS-prepared
this contrasts with previous assertions that georgeite and malachite zincian georgeite is more active than both the catalyst formed from
Figure 2 contrasts the activity of our model catalysts with that of com-
2
−
2+
achite (the CO
3
1
6
(
Cu2CO3(OH)2) are iso-compositional because georgeite rapidly zincian malachite and the standard commercial catalyst. In the case
transformed into malachite. We did not observe such instability for of methanol synthesis, for which our model catalysts showed mark-
SAS-prepared georgeite, as the rapid solvent extraction produced edly higher initial productivity (normalized for copper surface area;
1Cardiff Catalysis Institute, School of Chemistry, Cardiff University, Main Building, Park Place, Cardiff CF10 3AT, UK. 2The UK Catalysis Hub, Research Complex at Harwell, Harwell, Oxon OX11
0
FA, UK. 3Kathleen Lonsdale Building, Department of Chemistry, University College London, Gordon Street, London WC1H 0AJ, UK. 4Diamond Light Source, Didcot OX11 0DE, UK. 5Department
of Chemistry, University of Liverpool, Crown Street, Liverpool L69 7ZD, UK. 6Center for Electron Nanoscopy, Technical University of Denmark, Fysikvej 307, DK-2800 Kgs Lyngby, Denmark.
7Department of Materials Science and Engineering, Lehigh University, 5 East Packer Avenue, Bethlehem, Pennsylvania 18015, USA. 8Johnson Matthey, PO Box 1, Belasis Avenue, Cleveland
TS23 1LB, UK.
0
0 M o n t h 2 0 1 6 | V o L 0 0 0 | n A t U R E | 1
©
2016 Macmillan Publishers Limited. All rights reserved

reSeArCH Letter
a
b
c
1
00
90
80
70
60
1
00 200 300 400 500
Temperature (°C)
100 200 300 400
Temperature (°C)
100 200 300 400
Temperature (°C)
Figure 1 | Thermal gravimetric analysis (TGA) with evolved gas analysis zincian georgeite (red lines). c, EGA of malachite (black lines) and zincian
(
EGA) of georgeite and malachite. a, TGA of georgeite (black solid
malachite (red lines). In b and c, the solid lines indicate CO2 (fragment
with mass 44) and dashed lines indicate H2O (fragment with mass 18).
line), zincian georgeite (red solid line), malachite (black dotted line) and
zincian malachite (red dotted line). b, EGA of georgeite (black lines) and
Extended Data Fig. 4), the enhancement of activity was short-lived. using 30% less copper than is present in a catalyst derived from a 2/1
But in the LTS reaction, the SAS-prepared zincian-georgeite-derived copper/zinc georgeite (Extended Data Fig. 4).
catalyst was far more active and stable over the entire test period;
The stability and quantities of georgeite that are available through
indeed, enhanced activity was apparent at all contact times investigated SAS allowed us to explore structural origins for its enhanced activity.
Fig. 2). These observations demonstrate that the availability of stable X-ray-absorption near-edge structure (XANES) data for copper-only
zincian georgeite as a precursor gives access to highly active copper/ georgeite and malachite precursors (Fig. 3a) indicate that both materials
(
zinc-oxide catalysts, with the LTS activity and stability maintained contain Cu2 in a distorted octahedral environment , in agreement
+
20
a
14
260
250
b
100
95
90
85
80
75
70
65
12
240
230
220
210
10
8
6
4
2
0
2
1
00
90
180
0
10 20 30 40 50 60 70 80
Time (h)
0
20
40
60
80 100 120
Time (h)
c
100
90
80
70
60
50
40
30
50
100
150
200
250
300
Gas hourly space velocity (h–1 ×1,000)
Figure 2 | Comparison of our model catalysts to commercial catalysts.
a, Methanol synthesis (shown as mass-normalized, time-on-line
methanol production). The dashed line denotes the representative
reactor-bed temperature for methanol synthesis. Reaction conditions
for methanol synthesis: 190–250 °C (shaded area shows reaction under
equilibrium conditions at 250 °C), 25 bar, gas composition CO/CO2/H2/
2 2 2 2
27.5 bar, gas composition H O/CO/CO /H /N =50/2/8/27.5/12.5,
−
1 −1
MHSV =75,000 l kg
h . c, LTS conversions at different gas hourly
space velocities. (Extended Data Fig. 4 shows LTS mass-normalized data,
copper mass-normalized productivities/activities, initial copper surface
area normalized productivities/activities and product impurities.) Red
triangles, zincian-georgeite-derived catalyst; blue squares, co-precipitated
zincian-malachite-derived catalyst; open squares, industrial standards
(composition of the industrial methanol-synthesis catalyst by wt% is CuO/
ZnO/Al O =60/30/10; composition of the industrial LTS catalyst by wt%
−
1
h
−1.
N2 =6/9.2/67/17.8, mass hourly space velocity (MHSV) =7,200 l kg
The selectivities of all of the catalysts to methanol are greater than
9
9.96%. b, Low-temperature water–gas shift (LTS) reaction (shown
2
3
as time-on-line CO conversion). The dashed line represents the
maximum equilibrium conversion. Reaction conditions for LTS: 220 °C,
2 3
is CuO/ZnO/Al O =50/33/17). Error-bar calculations are discussed in
Methods.
2
| n A t U R E | V o L 0 0 0 | 0 0 M o n t h 2 0 1 6
©
2016 Macmillan Publishers Limited. All rights reserved

Letter reSeArCH
a
b
c
1
1
1
1
0
0
0
0
0
.6
.4
.2
.0
.8
.6
.4
.2
.0
770 nm
825 nm
8,980 9,000 9,020 9,040
0
1
2
3
4
5
6
200
400
600
800
Energy (eV)
R (Å)
Wavelength (nm)
d
e
Zincian georgeite
Georgeite
Malachite
Georgeite
0
5
10
15
20
0
5
10
15
20
r (Å)
r (Å)
Figure 3 | X-ray absorption fine-edge spectroscopy (XAFS) and X-ray
pair distribution function (PDF) analysis. a, Copper K-edge XAFS
information on local structure, for malachite (red line) and georgeite
transition bands (based on peak maxima) marked as small vertical lines
(zincian phases are shown in Extended Data Fig. 5a). d, X-ray PDF data for
georgeite and zincian georgeite. D(r) is the probability of finding a pair of
atoms, weighted by the scattering power of the pair, at a given distance, r.
The probability is scaled to r to emphasize order at greater distances. e,
Top, observed PDF G(r) data (G(r) is equivalent to D(r), only without
scaling by r) for georgeite. Bottom, calculated PDF G(r) data for all of the
atom pairs of malachite (black), together with the partial contributions
from the most strongly contributing atom pairs: C–O (red), Cu–O
(green) and Cu–Cu (blue). The dotted line represents a contribution from
crystallographically equivalent Cu atoms in malachite (see text).
(
black line). b, Fourier transform (FT) of extended X-ray absorption
2
2
fine-edge spectra (EXAFS) k weighted χ data (Fourier transform k
χ (R)), with associated fitting parameters for malachite (red line) and
georgeite (black line). Variation in the magnitude of the FT is plotted
with distance R (Å) from the Cu absorber; k denotes the fitting window
from the χ data. The fit (shown with dashed lines) was modelled on four
shorter Cu–O bond distances and two longer Jahn–Teller distorted Cu–O
bond distances (Extended Data Table 1). c, Ultraviolet–visible spectra
for malachite (red line) and georgeite (black line), with centres of d–d
with diffuse reflectance ultraviolet–visible spectroscopy (DRS) and (Fig. 3e) shows that georgeite is not simply a nanoscale form of mala-
extended X-ray absorption fine structure (EXAFS) data (Fig. 3b and chite (a phenomenon noted for other amorphous minerals, such as iron
23
Extended Data Table 1). But our analysis also suggests subtle differences sulphide ): the complete absence of the 15.27Å correlation between
in coordination geometries. Specifically, DRS locates the optical band crystallographically identical copper atoms in georgeite shows that
associated with the metal centre(s) in georgeite at 770nm, whereas the local ordering operates well below the length scales associated with
combined bands from the two crystallographic copper centres in mal- the malachite unit cell. PDF matching to other copper phases such
2
4
25
26
achite occur at 825nm (ref. 21). Although XANES data for georgeite as aurichalcite , azurite and rosasite confirmed georgeite to be a
and malachite are similar, the Fourier transform of the EXAFS χ data material distinct from any known copper hydroxycarbonate (Extended
(
Fig. 3b) shows that malachite has contributions further out from the Data Fig. 5).
copper absorber that are correlated to the scattering effects of copper–
After synthesis, the zincian georgeite and malachite precursors were
copper neighbours, whereas there are very few features in the georgeite calcined at 300°C to form copper-oxide/zinc-oxide intermediates, and
EXAFS data beyond 2Å and no evidence of metal–metal correlation. then reduced to the active catalyst phase. We find that the intermediate
The addition of zinc did not affect metal geometry in georgeite, with no materials produced on calcination and the final-state catalysts both
discernible changes appearing in the copper EXAFS data or zinc K-edge exhibit a microstructure that reflects the structural characteristics of
XANES or EXAFS χ-plots of zincian georgeite (such changes would their precursor phase. In the case of SAS-prepared zincian georgeite,
be expected in the case of impurity phases such as ZnCO3) (Extended calcination produced disordered copper oxide/zinc oxide with nano-
Data Fig. 5).
crystals of 3–4nm in diameter, as observed by STEM and PDF analysis
X-ray pair distribution function (PDF) analysis, which does not suf- (Extended Data Figs 6 and 7). The extent of the disorder resulted in
fer from phenomena related to scattering-path phasing, indicated that XRD and XANES analysis showing no discernible metal-oxide con-
georgeite and zincian georgeite have strikingly similar and strong short- tribution, although subtle changes in the EXAFS data were observed
range order up to 5Å (Fig. 3d); subtle differences appear between 5Å (Extended Data Fig. 8). Disordered copper-oxide species have previ-
and 10Å, with georgeite exhibiting less long-range order. The extended ously been reported to have copper-edge XANES data that fit with a
27
order seen upon addition of zinc could be responsible for the enhanced Jahn–Teller distorted octahedron , as with the calcined zincian geor-
16
stability of zincian georgeite prepared by co-precipitation . A com- geite. We found that the nanocrystallinity of the SAS-prepared zin-
parison of the georgeite PDF with a calculated PDF of malachite22 cian georgeite persisted until the residual high-temperature carbonate
0
0 M o n t h 2 0 1 6 | V o L 0 0 0 | n A t U R E | 3
©
2016 Macmillan Publishers Limited. All rights reserved

reSeArCH Letter
Georgeite
Malachite
5
nm
5 nm
FFT of selected area
FFT
FFT of selected area
FFT
1
.8 Å
2.1 Å
2
.1 Å
1.8 Å
2.1 Å
55º
55º
55º
71º
2
.1 Å
5
nm–1
5 nm–1
Cu zone axis [0 1 1]
Cu zone axis [0 1 1]
Figure 4 | Characterization, by environmental transmission electron
microscopy, of the microstructure of the reduced final-state catalysts
derived from zincian georgeite and zincian malachite. Calcined
materials were reduced under a H2 pressure of 2 mbar and a temperature
of 225 °C. The representative images shown in the top panels reveal
distinct Cu nanoparticles distributed on ZnO for both samples. This
was confirmed by fast Fourier transform (FFT) analysis, shown in the
bottom panels. The reduced malachite sample generally has larger Cu
nanoparticles than the georgeite sample. Also, there is a substantially
greater Cu–ZnO interaction in the georgeite-derived sample than in the
malachite-derived catalyst. The georgeite sample has a complex mix of
Cu and poorly defined ZnO particles (as revealed by diffuse rings in the
large-scale FFTs); the malachite sample shows better-defined, phase-
separated Cu and ZnO. The oxidation state of Cu was monitored by means
of electron energy-loss spectroscopy, which revealed that Cu was in its
metallic state in both samples under the reduction conditions (Extended
Data Fig. 9f). The Cu zone axis is the direction (as determined by analysis
of the FFT) along which the Cu crystallite is being observed.
decomposed to form XRD-discernible copper-oxide and zinc-oxide copper and zinc oxide in the catalyst that was derived from disordered
phases above 425°C (Extended Data Fig. 7). Conversely, calcination of zincian georgeite.
zincian malachite at temperatures greater than 300°C produced crys-
Methanol-synthesis activity is well known to correlate strongly
28
talline, XRD- and XANES-observable, copper-oxide and zinc-oxide with copper surface area , and a large interfacial area between copper
phases (Extended Data Fig. 7). A linear combination fit of the XANES and zinc oxide has been reported to produce highly active methanol-
2
9
data showed that 60% of the 300 °C calcined zincian malachite synthesis catalysts , containing active sites that are associated with
8
(
Extended Data Fig. 8) was associated with copper oxide and zinc oxide, surface-defect copper sites decorated by zinc . However, we find that
whereas these phases were not present in the calcined zincian georgeite. 40 hours of LTS conditions markedly decrease the copper surface
2
−1
STEM showed that 300°C calcined zincian malachite (Extended Data areas of catalysts derived from both zincian georgeite (53± 3m g to
2
−1
2
−1
2 −1
Fig. 6) was composed of copper-oxide and zinc-oxide grains with much 17± 1m g ) and zincian malachite (38± 2m g to 19± 1m g )—
longer-range crystalline order than the corresponding calcined zincian yet the catalysts were still active after this time period, clearly indicat-
georgeite.
ing that copper surface areas do not adequately correlate with catalyst
The nanocrystalline nature of calcined zincian georgeite translates activity in this reaction. We surmise that the high activity of our catalyst
2
−1
to a high copper surface area (53m g ) and small mean copper crys
-
can be explained by a combination of a greater content of exposed
tallite size (7.4± 0.7nm) when the material is reduced. Environmental copper, and an intimate interface contact between copper and zinc
transmission electron microscopy (ETEM) carried out under reducing oxide, inherited from the zincian-georgeite precursor. Moreover, the
conditions shows a complex mixture of copper particles and excep- SAS method produces very pure zincian georgeite with a low content
tionally small, disordered zinc-oxide grains (the mean zinc-oxide of sodium ions, without requiring an aqueous washing step; this high
crystallite size was 2.8 ± 1.6 nm, as judged by XRD), with notable purity might also contribute to both the high activity and the stability
interactions between copper and zinc oxide (Fig. 4). By comparison, of the zincian-georgeite-derived LTS catalyst.
the more crystalline zincian-malachite-derived copper oxide/zinc
Online Content Methods, along with any additional Extended Data display items and
Source Data, are available in the online version of the paper; references unique to
2
−1
oxide had a lower copper surface area (35m g ) and a microstructure
composed of distinctly larger copper particles (mean crystallite size these sections appear only in the online paper.
was 14.2± 3.8nm, by XRD) and zinc oxide particles (mean crystallite
ꢀeceived 11 September; accepted 8 December 2015.
size was 5.6 ± 3.6 nm by XRD). This difference in particle size and
structural order is reflected in the smaller copper–copper and zinc–
zinc coordination numbers noted for the zincian-georgeite-derived
catalyst from in situ EXAFS analysis (Extended Data Fig. 9). On a
macroscopic level, strong interactions between copper and zinc oxide,
as shown by XPS observations of partial ZnOx coverage of copper on
Published online 15 February 2016.
1
.
Bridge, P. J., Just, J. & Hey, M. H. Georgeite, a new amorphous copper carbonate
from the Carr Boyd Mine, Western Australia. Mineral. Mag. 43, 97–98 (1979).
Spencer, M. S. The role of zinc oxide in Cu/ZnO catalysts for methanol
synthesis and the water-gas shift reaction. Top. Catal. 8, 259–266 (1999).
Behrens, M. Meso- and nano-structuring of industrial Cu/ZnO/(Al O )
2 3
catalysts. J. Catal. 267, 24–29 (2009).
2.
3.
4.
8
reduction , were observed for both catalysts (Extended Data Fig. 9).
Fujitani, T. & Nakamura, J. The effect of ZnO in methanol synthesis catalysts on
However, ETEM shows considerably more interactions between
Cu dispersion and the specific activity. Catal. Lett. 56, 119–124 (1998).
4
| n A t U R E | V o L 0 0 0 | 0 0 M o n t h 2 0 1 6
©
2016 Macmillan Publishers Limited. All rights reserved

Letter reSeArCH
5.
Chinchen, G. C., Denny, P. J., Jennings, J. R., Spencer, M. S. & Waugh, K. C.
Synthesis of methanol. Part 1. Catalysts and kinetics. Appl. Catal. 36, 1–65
25. Belokoneva, E. L., Gubina, Y. K. & Forsyth, J. B. The charge density distribution
3 3 2 2
and antiferromagnetic properties of azurite Cu [CO ] (OH) . Phys. Chem.
Miner. 28, 498–507 (2001).
(1988).
6
.
.
Rhodes, C., Hutchings, G. J. & Ward, A. M. Water-gas shift reaction: finding the
mechanistic boundary. Catal. Today 23, 43–58 (1995).
Short, G. D., Chinchen, G. C. & Williamson, J. G. Synthesis of methanol; finely
divided oxides of zinc, aluminum, magnesium; with metallic copper. US patent
26. Perchiazzi, N. & Merlino, S. The malachite-rosasite group: crystal structures of
glaukosphaerite and pokrovskite. Eur. J. Mineral. 18, 787–792 (2006).
2
27. Belin, S. et al. CuAu/SiO catalysts for the selective oxidation of propene to
7
acrolein: the impact of catalyst preparation variables on material structure and
catalytic performance. Catal. Sci. Technol. 3, 2944–2957 (2013).
4,788,175 (1988).
8
.
.
Behrens, M. et al. The active site of methanol synthesis over Cu/ZnO/Al
industrial catalysts. Science 336, 893–897 (2012).
Baltes, C., Vukojevic, S. & Schueth, F. Correlations between synthesis, precursor,
and catalyst structure and activity of a large set of CuO/ZnO/Al O catalysts for
2 3
methanol synthesis. J. Catal. 258, 334–344 (2008).
2
O
3
28. Chinchen, G. C., Waugh, K. C. & Whan, D. A. The activity and state of the copper
surface in methanol synthesis catalysts. Appl. Catal. 25, 101–107 (1986).
29. Behrens, M. et al. The potential of microstructural optimization in metal/oxide
catalysts: higher intrinsic activity of copper by partial embedding of copper
nanoparticles. Chem. Cat. Chem. 2, 816–818 (2010).
9
1
1
1
0. Bems, B. et al. Relations between synthesis and microstructural properties of
copper/zinc hydroxycarbonates. Chemistry 9, 2039–2052 (2003).
1. Pollard, A. M. et al. Georgeite and azurite as precursors in the preparation of
coprecipitated copper/zinc oxide catalysts. Appl. Catal. A 85, 1–11 (1992).
2. Prieto, G., de Jong, K. P. & de Jongh, P. E. Towards ‘greener’ catalyst
Acknowledgements We thank C. Brookes, L. Van de Water, H. Stanness and
C. Ramson for technical assistance. We thank Johnson Matthey and the
Engineering and Physical Sciences Research Council (EPSRC) for funding
though a CASE award, and acknowledge funding from the UK Technology
Strategy Board and the EPSRC and UK Catalysis Hub (grants EP/K014714/1,
EP/K014714/1, EP/K014668/1, EP/K014706/1, EP/H000925/1 and
EP/I019693/1). We used the B18 beamline at the Diamond Light Source
(allocation number SP8071) with the help of D. Gianolio and G. Cibin; and we
used the I15 beamline, which contributed to the PDF results. This research
used the resources of the Advanced Photon Source (APS)—a US Department
of Energy (DOE) Office of Science User Facility, operated for the DOE Office of
Science by Argonne National Laboratory under contract DE-AC02-06CH11357;
we thank K. Chapman for collecting PDF data at beamline 11-ID-B, APS. C.J.K
acknowledges funding from the National Science Foundation Major Research
Instrumentation program (GR#MRI/DMR-1040229). M.J.R. is a Royal Society
Research Professor. We thank the A.P. Møller and Chastine Mc-Kinney Møller
Foundation for contributing to the establishment of the Center for Electron
Nanoscopy in the Technical University of Denmark. Source data for Figs 1–3
are available at http://dx.doi.org/10.17035/d.2015.0008102108.
2 3
manufacture: reduction of wastewater from the preparation of Cu/ZnO/Al O
methanol synthesis catalysts. Catal. Today 215, 142–151 (2013).
1
3. Tang, Z.-R. et al. New nanocrystalline Cu/MnO
supercritical antisolvent precipitation. ChemCatChem 1, 247–251
2009).
x
catalysts prepared from
(
14. Reverchon, E. Supercritical antisolvent precipitation of micro- and nano-
particles. J. Supercrit. Fluids 15, 1–21 (1999).
1
1
5. Rogers, A. P. A review of the amorphous minerals. J. Geol. 25, 515–541 (1917).
6. Pollard, A. M., Thomas, R. G., Williams, P. A., Just, J. & Bridge, P. J. The synthesis
and composition of georgeite and its reactions to form other secondary
copper(II) carbonates. Mineral. Mag. 55, 163–166 (1991).
1
1
1
2
7. Behrens, M. et al. Performance improvement of nanocatalysts by promoter-
induced defects in the support material: methanol synthesis over Cu/ZnO:Al.
J. Am. Chem. Soc. 135, 6061–6068 (2013).
8. Campbell, J. S. Influences of catalyst formulation and poisoning on the activity
and die-off of low temperature shift catalysts. Ind. Eng. Chem. Process Des. Dev.
9, 588–595 (1970).
9. Schur, M. et al. Continuous coprecipitation of catalysts in a micromixer:
nanostructured Cu/ZnO composite for the synthesis of methanol. Angew.
Chem. Int. Edn 42, 3815–3817 (2003).
0. Rothe, J., Hormes, J., Bonnemann, H., Brijoux, W. & Siepen, K. In situ X-ray
absorption spectroscopy investigation during the formation of colloidal copper.
J. Am. Chem. Soc. 120, 6019–6023 (1998).
Author Contributions S.A.K., J.K.B., S.H.T., M.S.S., G.J.K., C.W.P., M.J.R., C.J.K.
and G.J.H. designed the experiments. S.A.K. and P.J.S. prepared samples for
TGA/EGA analysis and IR spectroscopy; S.A.K., P.J.S. and J.H.C. carried out the
catalysis evaluation and determination of copper surface area; S.A.K., P.J.S. and
P.A.C. carried out the XRD analysis; P.A.C. carried out the PDF analysis; P.P.W.,
S.A.K. and P.J.S. carried out the XAFS and interpretation; L.L. and C.J.K. carried
out the TEM and interpretation; E.M.F., J.B.W., C.J.K. and S.A.K. carried out
the ETEM and interpretation; D.J.M., P.J.S. and S.A.K. carried out the XPS and
interpretation. S.A.K., C.J.K., J.B.W., G.J.K., M.J.R. and G.J.H. wrote and edited the
manuscript. G.J.H. directed the research.
2
2
2
1. Klokishner, S. et al. Cation ordering in natural and synthetic (Cu1-xZn
x 2 3 2
) CO (OH)
and (Cu1-xZn (CO (OH) . J. Phys. Chem. A 115, 9954–9968 (2011).
)
x 5
)
3 2
6
2. Behrens, M. & Girgsdies, F. Structural effects of Cu/Zn substitution in the
malachite-rosasite System. Z. Anorg. Allg. Chem. 636, 919–927 (2010).
3. Michel, F. M. et al. Short- to medium-range atomic order and crystallite size of
the initial FeS precipitate from pair distribution function analysis. Chem. Mater.
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The authors declare no competing financial
interests. Readers are welcome to comment on the online version of the paper.
Correspondence and requests for materials should be addressed to
G.J.H. (Hutch@cardiff.ac.uk) and S.A.K. (KondratSA@cardiff.ac.uk).
1
7, 6246–6255 (2005).
4. Harding, M. M., Kariuki, B. M., Cernik, R. & Cressey, G. The structure of
aurichalcite, (CuZn) (OH) (CO determined from a microcrystal. Acta
Crystallogr. B 50, 673–676 (1994).
2
5
6
3 2
)
0
0 M o n t h 2 0 1 6 | V o L 0 0 0 | n A t U R E | 5
©
2016 Macmillan Publishers Limited. All rights reserved

reSeArCH Letter
MꢁꢂꢃODS
rate 2°Cmin−1). The sample was then cooled to room temperature and a 20–100°
Material preparation. SAS method for preparing georgeite and zincian georgeite. 2θ scan performed. Copper crystallite size was estimated from a peak-broadening
−
1
30
Copper(ꢀꢀ) acetate monohydrate (4 mg ml ) and zinc(ꢀꢀ) acetate dihydrate analysis of the XRD pattern using Topas Academic , and the volume-weighted
−
1
(
2.16mgml ) (Sigma Aldrich≥99% Puriss) were dissolved in ethanol (reagent column height (Lvol) was calculated according to ref. 31.
grade, Fischer Scientific) containing 0 vol%, 5 vol% or 10 vol% deionized water. X-ray absorption fine-edge spectroscopy (XAFS). K-edge XAFS studies of copper
−1
Smithsonite ZnCO
in a 10 vol% water and ethanol solution. SAS experiments were performed using Didcot, UK. Measurements were performed using a quick extended (QE) XAFS
apparatus manufactured by Separex. CO (from BOC) was pumped through the set-up with a fast-scanning silicon (111) double-crystal monochromator. The time
system (held at 110 bar, 40°C) via the outer part of a coaxial nozzle at a rate of resolution of the spectra reported herein was 1 minute per spectrum (kmax =14),
3
was prepared with zinc(ꢀꢀ) acetate dehydrate (2.16mgml
)
and zinc were carried out on the B18 beamline at the Diamond Light Source,
2
−1
6
kgh . The metal salt solution was concurrently pumped through the inner noz- and on average three scans were acquired to improve the signal-to-noise ratio of
−
1
zle, using an Agilent HPLC pump at a rate of 6.4mlmin . The resulting precipitate the data. All ex situ samples were diluted with cellulose and pressed into pellets to
was recovered on a stainless steel frit, while the CO –solvent mixture passed down optimize the effective-edge step of the XAFS data and measured in transmission
2
2
stream, where the pressure was decreased to separate the solvent and CO . The mode using ion-chamber detectors. All transmission XAFS spectra were acquired
precipitation vessel has an internal volume of 1 litre. Precipitation was carried concurrently with the appropriate reference foil (copper or zinc) placed between
out for 120 minutes, and followed by a purge of the system with ethanol–CO2 for It and Iref XAS data processing, and EXAFS analysis were performed using
3
2
33
1
5 minutes, then CO for 60 minutes under 110 bar and 40°C. The system was then IFEFFIT with the Horae package (Athena and Artemis). The FT data were
2
2
s
depressurized and the dry powder collected. Recovered samples were placed in a phase-corrected for oxygen. The amplitude-reduction factor, , was derived from
0
vacuum oven at 40°C for 4 hours to remove any residual solvent. Approximately 1.5 g EXAFS data analysis of a known copper reference compound, namely tenorite
of georgeite is prepared during the 120-minute duration of solution precipitation. (CuO). For the fitting of the local coordination geometry of georgeite and mala-
Co-precipitation method for preparing malachite and zincian malachite precursors chite, the Jahn–Teller distorted copper–oxygen bond was difficult to observe
for the standard methanol-synthesis catalysts. The procedure was performed via a because of thermal disorder, and so was fixed in the model.
semi-continuous process, using two peristaltic pumps to maintain pH. Copper(ꢀꢀ) X-ray pair distribution function (PDF) analysis. Synchrotron X-ray PDF data were
nitrate hydrate and zinc(ꢀꢀ) nitrate hydrate solutions in deionized water were pre- collected on the 11-ID-B beamline at the Advanced Photon Source, Argonne
pared with copper/zomc molar ratios of 1/0, 1/1 and 2/1. The premixed metal National Laboratory, and on the I15 beamline at the Diamond Light Source.
−1
solution (5 l, ~0.5gml ) was preheated along with a separate 5 l solution of 1.5M Powder samples were packed into kapton capillaries with an internal diameter
sodium carbonate. The mixed metals were precipitated by combining the two of 1mm. Room-temperature powder X-ray diffraction data were collected at a
solutions concurrently at 65°C, with the pH being held between 6.5 and 6.8. The wavelength of 0.2114Å (11-ID-B) and 0.1620Å (I15) using the Rapid Acquisition
3
4
−1
precipitate would spill over from the small precipitation pot into a stirred ageing PDF method . The scattering data (0.5≤Q≤22Å ) were processed into PDF
35
vessel, held at 65°C. The precipitate was aged for 15 minutes after all precursor data using the program GudrunX . Two notations of PDF data are presented in
solutions had been used. this manuscript: G(r) and D(r) (ref. 36). The total radial distribution function, G(r),
The precipitate was than filtered and washed to minimize sodium content. The is the probability of finding a pair of atoms, weighted by the scattering power of the
sample was washed with 6 l of hot deionized water, and the sodium content mon- pair, at a given distance, r; it is the Fourier transform experimentally determined
itored using a photometer. The washing process was repeated until the sodium total structure factor. D(r) is re-weighted to emphasize features at high r, such
content showed no change. The sample was then dried at 110°C for 16 hours. that D(r)=4πrρ0G(r), where ρ0 is the average number density of the material (in
−1
Samples were calcined for 6 hours at 300°C or 450°C in a tube furnace under static atoms Å ).
−
1
air. The ramp rate used to reach the desired temperature was 1°Cmin
.
Fourier transform-infrared spectroscopy (FT-IR). Analysis was carried out using
Catalyst testing. Catalyst testing was performed with 0.5g of the calcined catalyst, a Jasco FT/IR 660 Plus spectrometer in transmission mode over a range of
–1
pelleted and ground to a sieve fraction of 0.6–1 mm for both the methanol-syn- 400–4,000cm . Catalysts were supported in a pressed KBr disk.
thesis and the LTS test reactions. The catalysts were reduced in situ using a 2% H Raman spectroscopy. Raman spectra were obtained using a Renishaw inVia spec-
gas mixture at 225°C (ramp rate 1°Cmin–1), before the reaction gases were trometer equipped with an argon ion laser (λ=514nm).
introduced. Data reported for the industrial standards are the mean value from Thermal gravimetric analysis (TGA). TGA measurements were performed using
2
/
N
2
−1
four repeat experiments. Error bars are based on the standard deviation from the a SETARAM Labsys analyser with sample masses of about 20mg, at 1°Cmin
−1
mean values.
under air with a flow rate of 20mlmin .
Methanol synthesis. Testing was carried out in a single-stream, six-fixed-bed reactor Evolved gas analysis (EGA). EGA experiments were performed using a Hiden
with an additional bypass line. After reduction, the catalysts were then subjected CATLAB under the same conditions used in the TGA experiments.
1
to synthesized syngas (CO/CO
/H /N
2 2 2
=6.9/2/67/17.8) at 3.5lh− , 25 bar pres- Copper surface area analysis. This was determined by reactive frontal chromatog-
sure and 195°C. In-line gas analysis was performed using an FT-IR spectrometer, raphy. Catalysts were crushed and sieved to a particle size of 0.6–1mm and packed
which detected CO, CO2, H2O and CH3OH. Downstream of the catalyst beds, into a reactor tube. The catalyst was purged under helium for 2 minutes at 70°C
−1
2
knockout pots collected effluent produced from the reaction. The contents were before being heated under a 5% H reduction gas to 230°C (8°C min ) for 3 hours.
2
collected after each test run and analysed using gas chromatography to evaluate the The catalyst was then cooled to 68°C under helium before the dilute 2.5% N O
−1
selectivity of catalysts. The total system flow was maintained using the bypass line. reaction gas was added with a flow rate of 65mlmin . The formation of N
LTS. Testing was performed in six parallel fixed-bed reactors with a single stream the surface oxidation of the copper catalyst by N O was measured downstream
feed and an additional bypass line. After reduction, the catalysts were subjected using a thermal conductivity detector (TCD). Once the surface of the copper is
to synthetic syngas (CO/CO /H /N =1/4/13.75/6.25) at 27.5bar pressure and fully oxidized, there is a breakthrough of N O, detected on the TCD. From this,
20 °C. The reactant gas stream was passed though vaporized water to give a the number of oxygen atoms that are chemisorbed on the copper surface can be
water composition of 50 vol%. This gives a total gas flow of H O/CO/CO /H determined. The number of exposed surface copper atoms and the copper surface
=50/2/8/27.5/12.5. The standard mass hourly space velocity (MHSV) used area can then be derived. Quoted surface areas are calculated using discharged
2
from
2
2
2
2
2
2
2
2
2
/
N
2
−1
−1
37,38
for testing was 75,000kh kg . In-line IR analysis was carried out to measure sample mass. Note that recent work
CO conversion. Selectivity was determined by the methanol content within the to partial pressures of H2 exceeding 0.05bar, then partial reduction of ZnO at the
knockout pots downstream of the reactors. Space and mass velocity variation tests copper interface can occur. This will affect copper surface area results, owing to
has shown that, if the catalyst is exposed
were performed at 65 hours’ time-on-line by altering the flow for each catalyst bed.
Relative activities were calculated by altering the flow for each catalyst bed in order as a H
to achieve 75% CO conversion after 75 hours’ time-on-line. The total system flow to copper surface area
was maintained using the bypass line. Copper surface area analysis after exposure to LTS conditions. Copper surface area
Characterization methods. Powder X-ray diffraction (XRD). Measurements were analysis of the fresh catalysts were carried out on a Quantachrome ChemBET 3000.
N
2
O oxidizing both copper and ZnO
x
. In these cases, alternative techniques, such
2
thermal conductivity detector, will give more accurate data with respect
37,38
.
performed using a PANalytical X’pert Pro diffractometer with a Ni-filtered CuK
α
Sample (100mg) was packed into a stainless-steel U-tube and purged with high-
−1
radiation source operating at 40kV and 40mA. Patterns were recorded over the purity helium for 5 minutes. Samples were reduced using 10% H
2
/Ar (30mlmin
)
range of 10–80° 2θ using a step size of 0.016°. All patterns were matched using the heated to 140°C at 10°Cmin− , before heating further to 225°C at 1°Cmin−1.
International Centre for Diffraction Data (ICDD) database. An in situ Anton Parr The resulting catalyst was held at this temperature for 20 minutes to ensure that
XRK900 cell (with an internal volume of ~0.5 l) was used to monitor the formation complete reduction took place. Note that, under this partial pressure of H2, it has
1
37,38
of metallic copper during the reduction of the CuO/ZnO materials. A flow of 2% been reported that ZnO species in contact with copper can partially reduce
.
-1
H2/N2 (60mlmin ) was passed through the sample bed while the cell was heated Residual H
2
was flushed from the system by switching the gas line back over to
to 225°C (ambient temperature to 125°C, ramp rate10°C min−1; 125–225°C, ramp helium (80mlmin ), while holding the sample at 220°C for another 10 minutes.
−1
©
2016 Macmillan Publishers Limited. All rights reserved

Letter reSeArCH
The temperature was then reduced to 65°C for N
2
O pulsing (BOC AA Grade). temperature in an H2 atmosphere for both the georgeite and malachite samples,
A programme of 12 pulses of 113μl N2O with a 5-minute stabilization time between and recorded using a Gatan US 1000 charge-coupled-device camera.
pulses was carried out, followed by three pulses of N
O was trapped before reaching the detector using a molecular sieve 5A (pelleted, to collect XPS spectra, using a monochromatic Al K
.6mm, Sigma Aldrich) trap. The copper surface area was determined from the 120W. Data were collected with pass energies of 160eV for survey spectra, and
amount of N2 emitted and the post-reaction analysis of catalyst mass, as follows: 40eV for the high-resolution scans. The system was operated in the Hybrid mode,
Copper surface area (m g ) = N2 volume (ml) × NA × 2/Catalyst mass using a combination of magnetic immersion and electrostatic lenses and acquired
2
for calibration. Unreacted X-ray photoelectron spectroscopy (XPS). A Kratos Axis Ultra DLD system was used
N
2
α
X-ray source operating at
1
2
−1
1
9
−2
2
(
g)×24,000(ml)×(1.4710 ) (atomsm
where N is the Avogadro constant, 6.022×10 (atoms). The key assumptions are compensation system was used to minimize charging of the sample surface, and the
that the amount of N2 emitted amounts to half a monolayer’s coverage of oxygen, resulting spectra were calibrated to the C(1s) line at 284.8eV; all spectra were taken
)
over an area approximately 300×700 μm . A magnetically confined charge-
23
A
1
9
−2
−9
and that the surface density of copper is 1.47×10 (atomsm ). The volume of with a 90° take-off angle. A base pressure of ~1×10 Torr was maintained during
N2 produced was quantified using a TCD. collection of the spectra. Gas treatments were performed in a Kratos catalysis cell,
After copper surface area analysis, the samples were kept under N2 and trans- which mimics the conditions of a normal reactor vessel, allowing the re-creation
ferred to an LTS reactor. Ageing was carried out in a single fixed-bed reactor of reactor conditions and analysis of the chemical changes taking place on the
equipped with a bypass line. CO, N2, CO2 and H2 were introduced to the catalyst catalyst surface. Briefly, the cell consists of a fused quartz reactor vessel contained
−8
bed via mass-flow controllers (Bronkhorst). Water of high-performance liquid within a stainless-steel vacuum chamber (base pressure ~10 Torr after baking).
−1
chromatography (HPLC) grade was passed through a liquid-flow controller Samples were heated at a controlled ramp rate of 2°Cmin to a temperature of
(
Bronkhorst) and then into a controlled evaporator mixer (Bronkhorst) that was 225°C using a eurotherm controller. The catalysts were exposed to an atmosphere
−1
heated to 140°C. N
2
was fed through the vaporized water to give a dilute syngas of 2% H2 in nitrogen with a flow rate of 30mlmin , controlled using MKS mass
/H /N =25/1/4/13.75/56.25). This mixture was intro- flow controllers during the heating ramp, during the 20-minute isotherm at 225°C,
mixture (H
2
O/CO/CO
2
2
2
duced at 220°C after re-reduction of the catalyst. The gas flows were controlled and also while the catalyst was cooled to 25°C. The samples were analysed before
−1
−1
to achieve an MHSV of 30,000lh kg . After ageing for 40 hours, the samples and after gas treatment without breaking the vacuum.
were transferred back to the Quantachrome ChemBET Chemisorption analyser, Inductively coupled plasma mass spectrometry (ICP-MS) and carbon–hydrogen–
whereby, after re-reduction, the copper surface areas of the aged catalysts were nitrogen (CHN) analysis. These analyses were provided as a commercial service
measured, as described above.
by Warwick Analytical Services.
Scanning transmission electron microscopy (STEM). Samples for STEM character- Helium pycnometry. This analysis was provided as a commercial service by MCA
ization were dry-dispersed onto holey carbon TEM grids. They were examined Services.
in an aberration-corrected JEOL ARM-200CF scanning transmission electron
microscope operating at 200kV in bright-field and dark-field STEM imaging
30. Coelho, A. A. TOPAS Academic: General Profile and Structure Analysis Software
for Powder Diffraction Data (Bruker AXS, Karlsruhe, 2010).
modes. Reliable electron microscopy results could be obtained only from the set
of zincian-georgeite- and zincian-malachite-derived materials, as these were found
to be stable under the vacuum environment of the microscope, and largely unaf-
fected by electron-beam irradiation. By way of contrast, the copper-only georgeite
precursor materials were highly unstable under vacuum conditions (even without
electron-beam irradiation), turning from a blue to dark-green colour, probably
owing to the loss of occluded water. The corresponding zincian-georgeite materials
showed no such colour transformation under vacuum.
31. Balzar, D. et al. Size-strain line-broadening analysis of the ceria round-robin
sample. J. Appl. Cryst. 37, 911–924 (2004).
32. Newville, M. IFEFFIT: interactive XAFS analysis and FEFF fitting. J. Synchrotron
Radiat. 8, 322–324 (2001).
3
3. Ravel, B. & Newville, M. ATHENA, ARTEMIS, HEPHAESTUS: data analysis for
X-ray absorption spectroscopy using IFEFFIT. J. Synchrotron Radiat. 12,
5
37–541 (2005).
3
4. Chupas, P. J. et al. Rapid-acquisition pair distribution function (RA-PDF)
analysis. J. Appl. Cryst. 36, 1342–1347 (2003).
Environmental transmission electron microscopy (ETEM). Samples for reduc-
tion during characterization by ETEM were dry-dispersed on heater chips
35. Soper, A. K. & Barney, E. R. Extracting the pair distribution function from
white-beam X-ray total scattering data. J. Appl. Cryst. 44, 714–726 (2011).
(
DensSolution trough hole) and then mounted on a DensSolution SH30 heat-
3
6. Keen, D. A. A comparison of various commonly used correlation functions for
describing total scattering. Angew. Chem. Int. Edn 53, 7043–7047 (2014).
7. Fichtl, M. B. et al. Counting of oxygen defects versus metal surface sites in
methanol synthesis catalysts by different probe molecules. J. Appl. Cryst.
34, 172–177 (2001).
ing holder. The holder with sample was inserted into an FEI Titan 80-300 envi-
ronmental transmission electron microscope operated at 300kV (ref. 39) . The
reduction of the samples was performed in situ as follows, for both the georgeite
3
2
and the malachite precursors: a flow of H was let into the microscope, building
up a pressure of 2mbar. The sample was heated from room temperature to 150°C
38. Kuld, S., Conradsen, C., Moses, P. G., Chorkendorff, I. & Sehested, J.
Quantification of zinc atoms in a surface alloy on copper in an industrial-type
methanol synthesis catalyst. Angew. Chem. Int. Edn 53, 5941–5945 (2014).
−
1
using a heating ramp rate of 10°Cmin . The final heating to 225°C was done at
°Cmin− . The oxidation state of copper was monitored by electron energy-loss
1
1
3
9. Hansen, T. W., Wagner, J. B. & Dunin-Borkowski, R. E. Aberration corrected and
monochromated environmental transmission electron microscopy: challenges
and prospects for materials science. Mater. Sci. Technol. 26, 1338–1344
(2010).
spectroscopy (EELS) during the reduction treatment using a Gatan Tridiem 866
spectrometer attached to the microscope. After an extended in situ treatment of
several hours at 225°C, phase-contrast lattice imaging was performed at elevated
©
2016 Macmillan Publishers Limited. All rights reserved

reSeArCH Letter
positions in wavenumbers (cm− ). *The presence of this band shows that
SAS precipitation with no additional water co-solvent produced some
georgeite as well as amorphous copper acetate. We attribute the formation
of georgeite to the small amount of water present from the monohydrated
starting salt. d, IR spectrum of mineralogical georgeite, reproduced with
the permission of the Mineralogical Society of Great Britain and Ireland
from ref. 11. e, FT-IR spectra of copper/zinc samples. f, XRD of copper/
zinc samples.
1
Extended Data Figure 1 | FT-IR and XRD characterization of SAS
copper and copper/zinc acetate precipitates. Key: i, SAS-prepared
copper acetate; ii, SAS-prepared georgeite; iii, malachite prepared by co-
precipitation; iv, 2/1 copper/zinc malachite prepared by co-precipitation;
v, SAS-prepared 2/1 copper/zinc georgeite; vi, SAS-prepared 1/1 copper/
zinc georgeite; vii, SAS-prepared zmithsonite (ZnCO3). a, XRD analysis of
copper-only samples. b, FT-IR analysis of copper-only samples. c, FT-IR
band assignment of copper-only samples, with (reference) designated as
received copper(ꢀꢀ) acetate monohydrate. Values given are for IR band
©
2016 Macmillan Publishers Limited. All rights reserved

Letter reSeArCH
Extended Data Figure 2 | Elemental composition of copper and
copper/zinc samples, with supplementary TGA analysis. a, Elemental
composition of SAS-prepared georgeite and co-precipitated malachite
samples. Elemental composition was determined by CHN analysis and
a skeletal density that negates buoyancy effects. b, Comparison of
experimental and calculated mass losses for georgeite and malachite from
TGA measurements. *Calculated from the assumption of final products
being 2/1 copper oxide/zinc oxide. The lower-than-expected mass losses
for the zincian phases could be associated with the inclusion of copper
ions in the zinc oxide lattice. c, TGA analysis of 2/1 copper/zinc georgeite
(red line) and 1/1 copper/zinc georgeite (black line).
a
ICP-MS. Densities were determined by helium pycnometry. Values
b
from ref. 11. Values from ref. 16. *Density determined by sink–float
(
SF) method. The helium pycnometry used in our present study provides
©
2016 Macmillan Publishers Limited. All rights reserved

reSeArCH Letter
Extended Data Figure 3 | Representative STEM micrographs of the
zincian georgeite precursor. Left, dark-field (DF) STEM micrograph.
Right, bright-field (BF) micrograph. The general morphology of the
zincian-georgeite precursor is shown in the DF-STEM micrograph. It
typically consists of very characteristic, irregularly shaped agglomerates,
about 100–200 nm in diameter, that are composed of ‘amorphous’ non-
faceted particles about 40 nm wide. Closer inspection by BF-STEM shows
that these non-faceted particles consist of an amorphous matrix phase
in which are embedded largely disconnected, sub-2-nm crystallites of
ordered material exhibiting clear lattice fringes. The amorphous matrix,
probably containing the carbonate and hydroxyl species, is by far the
majority phase, while the nanocrystallites make up less than 10% of
the material by volume. This observation is consistent with our other
characterization data, as signals from the matrix phase would dominate
the XAFS analysis, whereas the nanocrystallites are too small in dimension
to be detected by XRD. Analysis of fringe spacings and interplanar angles
of individual nanocrystallites from such images suggests that some of the
grains could be CuO (tenorite, where the copper is fourfold coordinated by
oxygen), while others fit better to Cu2O (cuprite, where the copper has a
coordination number of 2). No convincing matches of the lattice fringes to
either ZnO or ZnCO3 could be found.
©
2016 Macmillan Publishers Limited. All rights reserved

Letter reSeArCH
Extended Data Figure 4 | Further catalyst testing in methanol-synthesis
and LTS reactions. Filled red triangles, 2/1 Cu/Zn georgeite; open
triangles, 1/1 Cu/Zn georgeite; open squares, industrial standard; filled
blue squares, 2/1 Cu/Zn malachite (this test was terminated after 190 °C
because of poor activity). a, b, Methanol-synthesis data, normalized
to total catalyst mass (a) and copper mass (b), at 190–250 °C. The
dashed line shows the representative reactor-bed temperature. Reaction
conditions were as follows: pressure 25 bar; gas composition CO/CO2/
27.5 bar; gas composition H O/CO/CO /H /N =50/2/8/27.35/12.5;
2
2
2
2
−
1 −1
MHSV =75,000 l kg
h . f, Concentration of methanol collected in
the condensate pot after the LTS reaction, as determined by GC analysis.
g, Methanol-synthesis productivities and LTS activities, normalized by
copper surface area. *Copper surface area analysis determined by N O
2
§
reactive frontal chromatography before testing. Methanol-synthesis data
acquired at 190 °C, with steady state being at 18 hours’ time-on-line. ^LTS
data acquired at 220 °C, with steady state being at 40 hours time-on-line.
Note that LTS simulation testing showed that copper surface area dropped
−
1 −1
H2/N2 =6/9.2/67/17.8; MHSV =7,200 l kg
h . c, Concentration of
2
−1
2 −1
byproducts that were collected in the condensate pot after the methanol-
synthesis reaction, as determined by gas-chromatographic (GC) analysis.
Byproducts are: ethanol (blue), propanol (red), butanol (green), iso-
butanol (purple), and methyl iso-butyl ketone (turquoise). d, e, LTS-
reaction data, normalized to total catalyst mass (d) and copper mass
markedly after 40 hours (to 17 m g and 19 m g for zincian-
georgeite-derived and zincian-malachite-derived catalysts, respectively).
The inverse correlation between copper surface area and initial
productivity also suggests that loss of surface area occurs rapidly during
the reaction and that the initial rate data are therefore likely to be
inaccurate.
(
e). Reaction conditions were as follows: temperature 220 °C; pressure
©
2016 Macmillan Publishers Limited. All rights reserved

reSeArCH Letter
Extended Data Figure 5 | Spectroscopic analysis of the addition of
zinc to georgeite. a, Diffuse reflectance UV–vis spectra of zincian
malachite (red) and zincian georgeite (black). b, Copper K-edge EXAFS
with SAS-prepared smithsonite. d, Zinc K-edge EXAFS (χ) comparison of
zincian georgeite with SAS-prepared smithsonite. e, Comparison of observed
georgeite PDF data with simulated data for crystalline hydroxycarbonate
minerals with similar compositions to that of georgeite—namely aurichalcite,
azurite, rosasite, zincian malachite and malachite.
(
χ) comparison of zincian georgeite (with a 4/1 or 2/1 copper/zinc ratio)
with georgeite. c, Zinc K-edge XANES comparison of zincian georgeite
©
2016 Macmillan Publishers Limited. All rights reserved

Letter reSeArCH
Extended Data Figure 6 | Representative DF-STEM and BF-STEM
grain diameter of 3–4nm, which is just below the detection limit for XRD.
Analysis of the fringe spacings and interplanar angles from individual grains
suggests that the material is now mainly an intimate mixture of zinc and
copper oxides; the small amount of disordered material corresponds to the
residual occluded carbonate material, as detected by TGA/EGA analysis
of this material. b, BF-STEM of calcined zincian georgeite. c, DF-STEM of
calcined zincian malachite. d, BF-STEM of calcined zincian malachite.
micrographs of zincian georgeite and zincian malachite, calcined at
3
00°C. a, DF-STEM of calcined zincian georgeite. Higher-magnification
imaging reveals that, after much of the carbonate and hydroxyl content is
lost by calcination, most of the disordered matrix material in the precursor
has crystallized, and only a small amount of amorphous material remains.
The crystallized material is entirely in a nanocrystalline form, with a mean
©
2016 Macmillan Publishers Limited. All rights reserved

reSeArCH Letter
Extended Data Figure 7 | X-ray diffraction analysis of calcined zincian
georgeite and zincian malachite. a, PDF of zincian georgeite, as prepared
and after calcination at 120 °C, 200 °C, 250 °C, 300 °C and 450 °C. There
is little change in the observed PDF up to 250 °C, other than a slight peak
broadening, which can be attributed to a reduction in short-range order.
The dashed line shows the position of the C–O peak, which is retained
until temperatures higher than 300 °C. b, PDF and c, Rietveld fits of
zincian georgeite after calcination at 450 °C. The measured data are shown
as open circles; the fit is a solid black line; the difference is a grey line. Both
techniques determine the product to be a mixture of copper oxide and
zinc oxide (weight ratio of 68/32 by PDF; 67.7(4)/32.3(4) by Rietveld).
d, Ex situ XRD patterns following calcination of zincian georgeite for
2 hours at 250 °C, 300 °C and 450 °C. Open circles, copper oxide; filled
squares, zinc oxide. e, f, In situ XRD analysis of zincian georgeite (d) and
zincian malachite (e) during calcination between 300 °C and 500 °C under
an atmosphere of static air, with XRD scans every 25 °C.
©
2016 Macmillan Publishers Limited. All rights reserved

Letter reSeArCH
Extended Data Figure 8 | Copper K-edge XAFS analysis of zincian
georgeite and zincian malachite calcined at 300°C. a, EXAFS (χ)
comparison of copper oxide, calcined zincian georgeite, zincian georgeite
and georgeite. b, EXAFS (R) comparison of copper oxide (green),
calcined zincian georgeite (black) and zincian georgeite (blue). c–e, Linear
combination fit of copper oxide and zinc oxide with zincian malachite
calcined at 300°C.
©
2016 Macmillan Publishers Limited. All rights reserved

reSeArCH Letter
Extended Data Figure 9 | In situ characterization of final-state, reduced
copper/zinc-oxide catalysts derived from zincian georgeite and zincian
malachite, calcined at 300 °C. a, XRD of zincian-georgeite-derived
catalyst after in situ hydrogen reduction (2% H2/N2 at 225 °C for 1 hour).
Fine dotted lines indicate zinc-oxide reflections; dashed lines indicate
metallic copper. b, Copper K-edge EXAFS Fourier transorm of final
reduced catalysts derived from zincian georgeite and zincian malachite.
c, Copper K-edge EXAFS fit of reduced catalysts. d, Zinc K-edge EXAFS
Fourier transform of final reduced catalysts derived from zincian georgeite
and zincian malachite. e, Zinc K-edge EXAFS fit of reduced catalysts.
f, Electron energy-loss spectra (EELS) of georgeite and malachite,
respectively, during reduction in the ETEM experiment. The EELS data
were acquired in 2 mbar H2 at 225 ºC, and show the fine structure of the
0
Cu L2,3 ionization edges, which are characteristic of metallic Cu .
g, EXAFS fitting data for copper and zinc K-edge data. Fitting parameters
2
for K-edge copper: amplitude-reduction factor (S0 ) =0.91, as deduced
from copper-foil standard; fit range 3<k<11.2, 1<R<5.5 (with k denoting
the fitting window from the χ data, and R, the path length, denoting the
fitting from the Fourier transform of the χ data); number of independent
points =23; *denotes multiple scattering path. Fitting parameters for
2
K-edge Zn: s =0.90 as deduced from a ZnO2 standard; fit range
0
3.3<k<9.5, 1<R<4.8; number of independent points =16; *denotes
multiple scattering path. h, XPS analysis of calcined and reduced catalysts
derived from zincian georgeite and zincian malachite.
©
2016 Macmillan Publishers Limited. All rights reserved

Letter reSeArCH
ꢁxtended Data ꢂable 1 | ꢁXAFS fitting parameters for georgeite and malachite
Fitting parameters for malachite: S02 =0.72 as deduced from a copper-oxide standard; fit range 3<k<12, 1<R<3.5; number of independent points=14. Fitting parameters for georgeite: s =0.72 as
2
0
2
deduced from a copper-oxide standard; fit range 3<k<12, 1<R<3.5; number of independent points=14. Abs Sc; scattering path, N; coordination number (first shell); R, path length, σ : Debye–Waller
factor, Ef; change in edge energy, Rfactor; goodness of fit.
©
2016 Macmillan Publishers Limited. All rights reserved