1
In situ XRD and HRTEM studies on the evolution of the Cu/ZnO methanol synthesis catalyst during its reduction and re-oxidation

2
During the formation of the CuZnO solid solutions, the foreign anions in anion-modified (a.m.-) oxides give rise to: (i) extended stacking faults of (002) ZnO lattice plane, which are occupied by copper ions in the form of small clusters; and (ii) vacant inner holes of the a.m.-ZnO crystal.

3
The main part of copper ions in the clusters is reduced to Cu0 with hydrogen at 473 K. According to HRTEM studies, the reduction of Cu0.08Zn0.92O results in the formation of copper metal species of two types: (i) particles of 3–10 nm in size on the a.m.-ZnO surface; and (ii) small (no more than 3 nm in size) atomic copper metal clusters in defect voids of the a.m.-ZnO structure.

4
The copper metal clusters are coherent inclusions in the bulk of the ZnO, and the large copper metal particles are epitaxially bonded to the surface of the ZnO matrix.

5
Copper metal particles on the surface of a.m.-ZnO are reoxidized to Cu+2 at 523 K in the helium flow containing 0.05 vol% oxygen, and they come back to the extended stacking faults.

6
The copper metal clusters in the holes of a.m.-ZnO are inaccessible to the oxygen and are not reoxidized.

Introduction

7
Methanol is used as the principal feedstock for production of many organic compounds, among which are formaldehyde, acetic acid, methyl-methacrilate MMA, dimethyl-terephthalate DMT, methyl-tert-butylether MTBE and chloromethanes.

8
Due to its high energetic density and no problem in its storage and handling, methanol is also considered an important liquid power-supplier in fuel cells.

9
Copper–zinc and copper–zinc–aluminium oxide catalysts, which are very selective and active enough at low-pressure conditions, are mainly used for the methanol synthesis.

10
A variety of techniques were applied by numerous researchers1–10 for studying the active state of the copper–zinc catalysts but the nature of the active sites is still disputable in literature.

11
The present research is focused on the structure of the active state of the methanol synthesis catalysts by means of studying the model copper–zinc catalysts, including the as-prepared sample, as well as those after hydrogen reduction and mild re-oxidation.

12
The chosen samples are the solid solutions of copper ions in anion-modified zinc oxide Cu0.08Zn0.92O and solid solutions of copper and aluminium ions in anion-modified zinc oxide Cu0.2Al0.1Zn0.7O.

13
They are known to be the precursors for active catalysts for synthesis of methanol.11–14

14
It has been shown before that these solid solutions of copper (and aluminium) cations in the zinc oxide structure can only be obtained via low-temperature (not higher than 773–823 K) decomposition of Cu(Al)Zn hydroxycarbonates of hydrozincite or aurichalcite clay structure12,15 to produce OH- and CO3-containing zinc oxide (anion-modified, a.m.-ZnO).

15
The ad-anions provide stabilization of copper cations in the zinc oxide structure, i.e. dissolution of the copper cations in the a.m.-ZnO.16

16
Al3+ cations dissolved in zinc oxide cause an increase of the clusters in number thus preventing formation of the copper oxide phase as the copper concentration increases.13

17
ES studies of the solid solutions at 5000–50000 cm−1 showed copper cations forming oxygen-containing small clusters Cu2+–O–Cu2+ with the strongly distorted, almost flat-square, oxygen coordination of Cu2+.17

18
The hydrogen treatment at temperatures below 573 K results in the reduction of the majority of the copper ions from the a.m.-ZnO followed by the copper diffusion to the a.m.-ZnO surface.

19
XRD data allow the hypothesis that metallic copper forms 3–4 nm crystallites which are epitaxially bonded to the ZnO.18

20
The epitaxy of Cu0 crystallites over ZnO support was directly shown by the in situ HREM.9

21
The data of refs. 18 and 19 also give the grounds for the supposition that the reduction of copper cations from a.m.-ZnO solid solution is reversible: the epitaxial Cu0 particles are easily re-oxidizable and can return to the a.m.-ZnO structure under the action of low partial pressures of oxygen.

22
The reversible reduction–oxidation of copper–zinc oxide was also observed recently in .ref. 20

23
Meanwhile, the air treatment transforms the epitaxial Cu0 particles to CuO phase.

24
The goal of the present study is to give a better understanding of the structural evolution of the methanol synthesis catalysts during their reduction and re-oxidation by means of in situ XRD and HRTEM.

25
The nature of the copper species migration is in the focus of the investigation.

26
The catalytic activity of both Cu/ZnO and CuAl/ZnO catalysts is proportional to the copper concentration in the a.m.-ZnO.11

27
This allows to assume a similar nature of the Cu/ZnO and CuAl/ZnO catalysts action in the methanol synthesis, while introduction of Al3+ contributes to increasing the solubility of Cu2+ in a.m.-ZnO and improves the thermal stability of the catalyst.

28
Since the electron microscopic studies can give more clear results for the double oxide system (i.e., without aluminium), these studies were first performed for the Cu0.08Zn0.92O sample.

29
Meanwhile, we considered it important to present in this paper the in situ XRD data for the reduction–oxidation evolution of the Cu0.2Al0.1Zn0.7O sample as well.

30
The study uses exactly the same samples of Cu0.08Zn0.92O and Cu0.2Al0.1Zn0.7O, which were studied earlier in refs. 11–13 and 17–19.

Experimental procedures

31
Samples of a.m.-CuO, a.m.-ZnO, a.m.-Cu0.08Zn0.92O and a.m.-Cu0.2Al0.1Zn0.7O were prepared by calcination of the corresponding hydroxycarbonates in flowing air at 623 K for 4 h.

32
Hydroxycarbonate precursors of Cu/ZnO and CuAl/ZnO catalysts were synthesized via the continuous co-precipitation method.

33
Solution of sodium carbonate and mixture of appropriate quantities of 0.5 M solutions of copper, zinc and aluminum nitrates were added dropwise at 343–348 K and pH = 6.8–7.0 to an appropriate buffer solution.

34
Dwell time was 30 min.

35
Copper nitrate 3-hydrate (Saki Chemical Plant, Russia, >99% pure), zinc nitrate 6-hydrate (Donetsk Chemical Plant, Ukraine, >98% pure) and aluminum nitrate 9-hydrate (Donetsk Chemical Plant, Ukraine, >99% pure) were used.

36
The precipitates were thoroughly washed to remove Na+ and NO3 ions, filtered out, and dried under IR-lamp overnight.

37
Reductive treatments were performed in flowing hydrogen at 473 K for 1 h, heating rate −2 K min−1.

38
The re-oxidations were performed in the flowing mixture of helium and oxygen (0.05% O2) at 2 K min−1, 523 K, until the termination of oxygen absorption.

39
A Siemens D-500 and a Bruker-D8 diffractometers (Germany) with Cu Kα radiation were used for in situ X-ray diffraction (XRD) studies.

40
A high-temperature reactor chamber21 was linked to the gas feeding system.

41
At first, in this chamber the full diffraction patterns were acquired for the as-prepared samples.

42
Then, the samples were reduced in flowing hydrogen at 2K min−1 473 K until the diffraction patterns (monitored by the intensity of Cu0 (111) line) stopped changing, and the full diffraction patterns were acquired.

43
Then the reduced samples were reoxidized in the flowing mixture of helium and oxygen (0.05% O2) at 523 K until the diffraction patterns stopped changing.

44
Samples treated under present conditions were studied using high resolution transmission electron microscopy (HRTEM).

45
Electron micrographs were obtained with JEM-2010 instrument with lattice resolution 1.4 Å and accelerating voltage 200 kV.

46
Periodic images of lattices structures were analyzed using digital Fourier transformation.

47
Local elemental analysis was performed with EDX method on Energy-dispersive X-ray Phoenix spectrometer equipped with Si (Li) detector with energy resolution not worse than 130 eV.

48
Samples were fixed on ≪holey≫ carbon films supported on copper or molybdenum grids and investigated with the electron microscope.

49
In situ thermogravimetric (STA) data were obtained by means of the Netzsch STA 409 thermobalance.

50
The sensitivity of the weight measurements was about 0.01 mg (ca. 0.02% of the sample weight).

51
The heating rate was 5 K min−1).

Results and discussion

Studies of the as-prepared samples

XRD and STA studies

52
Fig. 1 shows the diffraction patterns of as-prepared samples: a.m.-CuO (curve 1), a.m.-ZnO (curve 2), a.m.-Cu0.08Zn0.92O (curve 3).

53
The diffraction pattern of as-prepared a.m.-Cu0.2Al0.1Zn0.7O is presented by curve 1 in Fig. 2.

54
The lines of zinc oxide are only seen in the diffraction patterns acquired both for a.m.-Cu0.08Zn0.92O and a.m.-Cu0.2Al0.1Zn0.7O samples, as well as for a.m.-ZnO.

55
So, only one phase with ZnO structure is observed in a.m.-Cu0.08Zn0.92O and a.m.-Cu0.2Al0.1Zn0.7O.

56
No copper oxide phase is seen.

57
The STA method provides additional evidence for the absence of a copper oxide phase.

58
Fig. 3 shows a TPR profile recorded for the a.m.-Cu0.08Zn0.92O sample in flowing hydrogen.

59
Two effects are observed at 435 and 680 K. According to XRD data, the first effect is caused by the reduction of copper and formation of metallic particles.

60
The weight loss of 1.44% weight corresponds to the reduction of ca. 92% of the Cu2+.

61
The second weight loss at ca. 680 K is related to the removal of the ad-anions and the recrystallization of the metallic copper particles as well as of the a.m.-ZnO phase giving the regular zinc oxide.

62
The peculiarities of the transformation of the anion-modified oxides into regular oxides were discussed earlier in refs. 22 and 23.

63
The existence of the weight loss at 680 K in hydrogen flow proves that the sample Cu0.08Zn0.92O after calcination at 625 K contains the ad-anions, i.e. represents itself as the anion-modified mixed oxide.

64
The weight loss at above 750 K is accounted for by the reduction of ZnO and evaporation of metallic zinc.

Electron microscopic studies

65
HRTEM micrograph of anion-modified zinc oxide, a.m.-ZnO, is shown in Fig. 4 (left).

66
There are seen (002) planes of ZnO crystal lattice with a lattice spacing of 0.26 nm (the (002) planes are oriented normally to the plane of the micrograph).

67
Microvoids are seen in the crystal bulk as rounded 2–3 nm spots of slightly light contrast.

68
There is no visible anomalous contrast or distortion of the planes due to imperfection of the crystal lattice.

69
Fig. 4 (right) presents a HRTEM image of a.m.-Cu0.08Zn0.92O as-prepared.

70
As was said above, this is the solid solution of copper ions in a.m.-ZnO.

71
Here, two types of the structural defects are observed in the micrograph.

72
The interplanar spacing ZnO (002) is unchanged (0.26 nm), i.e. the planes are also perpendicular to the plane of image.

73
However, some planes are seen broken and shifted as a result of (002) plane shear dislocations.

74
These shearing dislocations cause the formation of stacking faults of (002) lattice planes to substitute for the atomic plane alternation characteristic of the hexagonal lattice.

75
The stacking faults are visible as dark bands along planes (002).12

76
The stacking faults are linked with the large pores of 5 to 10 nm in size, exposed to the external crystal surface.

77
The defects of the second type are microvoids that are seen in the crystal bulk as light-contrasted 2–3 nm rounded spots.

78
These defects are similar to those observed in a.m.-ZnO.

79
From the electron micrography data, when considered along with the data on formation of the sample under consideration, it can be proposed that the observed elongated defects such as stacking faults are the residence regions of impurity anions and copper cations bound to them.

80
The microvoids may form in the crystal bulk as a result of the vacancy sink in the anion lattice during formation of the a.m.-ZnO crystal.

81
The microvoids don’t seem to include copper ions.

82
So, the performed study of Cu/ZnO catalysts has confirmed the formation of copper and copper and aluminium ion solutions in a.m.-ZnO via the low-temperature decomposition of the appropriate hydroxocarbonate.

83
A.m.-Cu0.08Zn0.92O and a.m.-Cu0.2Al0.1Zn0.7O samples represent themselves as solid solutions on the basis of a.m.-ZnO and don’t contain the CuO phase.

84
The HRTEM data allow discussing the structural peculiarities of a.m.-ZnO based solid solutions and the regularities of copper ions allocation.

85
It is worth emphasizing once more that there is no visible anomalous contrast or distortion of the planes due to imperfection of the crystal lattice in a.m.-ZnO as-prepared, but microvoids are only seen in the crystal bulk of ZnO as the rounded 2–3 nm spots.

86
Meanwhile, in a.m.-Cu0.08Zn0.92O as-prepared sample shearing dislocations cause the formation of stacking faults of (002) lattice planes in addition to the microvoids, similar to those observed in a.m.-ZnO.

87
Supposedly, these are the observed stretched defects such as stacking faults, where the ad-anions and copper cations are allocated.

Studies of patterns after reduction by H2

XRD studies

88
Diffraction patterns of a.m.-Cu0.08Zn0.92O and a.m.-Cu0.2Al0.1Zn0.7O after reduction in flowing hydrogen at 473 K are shown in Figs. 1 (curve 4) and 2 (curve 2), accordingly.

89
The lines assigned to copper metal appear.

90
The coherent scattering domains allow the estimation of the Cu0 crystallite size as 3.5–5.0 nm and 3–4 nm, accordingly.

91
The observed Cu0 XRD band intensities ratio shows the plausible imperfection of copper metal particles.

Electron microscopic studies

92
Fig. 5 shows the micrographs of a.m.-Cu0.08Zn0.92O reduced with hydrogen.

93
The general view of the picture is changed: (i) stacking faults have less contrast; and (ii) two kinds of Cu0 particles are observed.

94
Some of Cu0 particles are 3 to 10 nm in size on the surface of zinc oxide and there is also the fraction of small clusters almost equal in size (2–3 nm) which are seen as dark spots in the microvoids of zinc oxide.

95
The small clusters are shown at a higher magnification in Figs. 5 (right) and 6 (left).

96
The dark spots in the HRTEM micrograph are seen to be made up of Moiré fringes parallel to ZnO (002) lattice plane with periodicity of about 1 nm.

97
They result from superposition of ZnO (002) and cluster Cu0 (111) planes (d002 ZnO = 0.26 nm, d111Cu = 0.21 nm).

98
Hexagonal packing is characteristic of these planes in both crystal lattices.

99
The fact that Moiré fringes of the fine Cu0 particles are parallel to ZnO (002) lattice plane indicates the uniform orientation and the coherent lattices of Cu0 intrusions and the ZnO crystal matrix.

100
Two types of reflections are seen in the Fourier pattern (Fig. 6, right) of the coherent inclusion: one related to ZnO (the reflections are shown by downward arrows) and another – to Cu0 (upward arrows).

101
Reflections of Cu0 111 and 100 are broadened that indicates the presence of stacking faults of (111) metal planes.

102
This structural feature of Cu0 clusters may happen because the FCC (face centred cubic) structure of Cu is incorporated into HP (hexagonal packing) structure of ZnO.

103
So, it follows from the obtained XRD results that the reduction of both a.m.-Cu0.08Zn0.92O and a.m.-Cu0.2Al0.1Zn0.7O leads to the copper metal formation in the form of imperfect particles of small size.

104
It is in good agreement with our earlier data on Cu/ZnO catalysts.

105
HRTEM studies make it possible to reveal that the reduction of a.m.-Cu0.08Zn0.92O results in the formation of copper metal species of two types: (i) particles of 3–10 nm in size on the a.m.-ZnO surface; and (ii) small (no more than 3 nm in size) metal copper clusters in microvoids of the a.m.-ZnO structure.

106
The copper clusters are coherent inclusions in the bulk of zinc oxide, and the large particles are epitaxially bonded to the surface of the ZnO matrix in accordance with earlier data.18

107
It should be noted that if such a microvoid is situated in the subsurface layer of ZnO, it seems to be observable by LEIS as a metallic copper particle covered with ZnO.

108
Such a state was observed in ref. 7.

Studies of samples oxidized in the flow of oxygen-containing (0.05 vol% O2) helium after reduction with hydrogen

XRD studies

109
Diffraction patterns of a.m.-Cu0.08Zn0.92O and a.m.-Cu0.2Al0.1Zn0.7O samples after reduction with hydrogen and oxidation by the oxygen–helium mixture are shown in Figs. 1 and 2, respectively.

110
The changes observed in the diffraction patterns of reduced samples in the course of oxidation (Fig. 1, curves 5–7 and Fig. 2, curves 3–4) are the same: the lines with d = 0.208 nm decrease in intensity to disappear within ca. 2 h.

111
Some difference in the behavior of the samples is that the line with d = 0.208 nm is shifted to d = 0.212 nm as its intensity decreases in the diffraction pattern of the a.m.-Cu0.2Al0.1Zn0.7O sample.

112
The line with d = 0.212 nm is the most intense in the pattern of Cu2O.

113
After re-reduction with hydrogen of the samples at 500 K, the diffraction pattern (Fig. 2, curve 5) is identical to those of the primarily reduced samples (Fig. 1, curve 4 and Fig. 2, curve 2).

114
When the samples are re-oxidized again (re-re-oxidation) in the flowing oxygen-containing helium, the characteristic lines of copper metal disappear again (Fig. 2, curves 6–8).

115
Thus, Figs. 1 and 2 illustrate the experimental observations of the reversible changes in copper state in the samples under study.

116
Copper ions in the structure of a.m.-ZnO are reduced with hydrogen to form copper metal, then Cu0 atoms are oxidized in the flowing oxygen-containing helium to form Cu2+ ions, which return back to the zinc oxide structure.

117
The obtained results on the evolution of the diffraction patterns during re-oxidation of the reduced a.m.-Cu0.08Zn0.92O and a.m.-Cu0.2Al0/1Zn0.7O samples are the direct proof of the reversible red-ox copper behaviour.

118
This phenomenon for Cu/ZnO sample has been shown by us before18,19 and recently reported elsewhere.20

119
For a.m.-Cu-Al-ZnO, it is for the first time shown in the present work.

Electron microscopic studies

120
Fig. 7 shows a HRTEM micrographs of the a.m.-Cu0.08Zn0.92O sample oxidized in the flow of oxygen containing (0.05 vol%) helium after the reduction with hydrogen.

121
In the micrographs, some elongated defects of stacking faults type are seen in the a.m.-ZnO structure, which are similar to the defects in the pre-reduced sample (see Fig. 4, right) and 2–3 nm clusters of copper metal in the micro-voids of the a.m.-ZnO crystal (the second type of copper metal in the reduced samples).

122
The surface copper metal particles (first type of copper metal in the reduced samples) are not observed.

123
It is important that many defective regions, containing stacking faults of ZnO (002) planes, are present.

124
They are located in regions near the copper clusters (Fig. 7).

125
Thus, HRTEM studies make it possible to reveal that oxidation of surface copper metal particles by oxygen (at its low partial pressure) brings major copper back to the bulk of a.m.-ZnO crystal.

126
The copper cations are localized at the stacking faults in the a.m.-ZnO crystal.

127
Copper metal particles of 2–3 nm in size in the microvoids in the bulk of the a.m.-ZnO structure are inaccessible to the oxygen.

128
The portion of preserved Cu0 after the treatment in O2-containing mixture is not so high, since it is not detected by XRD.

129
Hence, HREM data also argue for the reversible reduction of copper–zinc samples, i.e. the main part of copper particles formed in the sample upon treatment with hydrogen can be re-oxidized to return to the structure of defective zinc oxide.

130
The quantitative evaluation of the amount of Cu0 in the re-oxidized samples needs the use of the other investigation techniques, e.g. the measurements of the temperature dependence of the magnetic susceptibility.

131
We expect to present the relevant data on the magnetic susceptibility of the CuZn and CuAlZn samples in one of our next publications.

Conclusions

132
The solid solutions of copper and aluminum ions in anion-modified ZnO are formed in the course of low temperature decomposition of appropriate hydroxocarbonates.

133
Copper metal as a defect particles of small size is formed as a result of the reduction with hydrogen.

134
Copper cations reduction is reversible: the return of copper back to the a.m.-ZnO structure occurs during re-oxidation with oxygen at low partial pressures.

135
The HRTEM studies allow to reveal the sites of copper cations placed in the solid solution.

136
Important details of the reduction and reoxidation processes can be understood, among which are the following:

137
(i).

138
For as-prepared samples it is shown that two types of defects can be observed in a.m.-Cu0.08Zn0.92O: shearing dislocations causing the formation of stacking faults of ZnO (002) lattice planes and microvoids in the crystal bulk in the form of 2–3 nm rounded spots.

139
Only one type of defects is observed in a.m.-ZnO – microvoids in the crystal bulk of ZnO rounded 2–3 nm spots, the same as in a.m.-Cu0.08Zn0.92O.

140
It may be thought that the observed stretched defects such as stacking faults are the residence regions of impurity anions and copper cations linked to them.

141
Microvoids in the crystal bulk are not the residence places of copper cations.

142
(ii) The reduction with hydrogen at temperatures below 573 K results in reduction of copper cations from stacking faults to Cu0 atoms and their diffusion either to the surface of the anion-modified zinc oxide or to free microvoids nearby the elongated defects.

143
The particles on the a.m.-ZnO surface are of 3–10 nm in size and are epitaxially bonded to the surface of the a.m.-ZnO matrix.

144
The copper metal atoms in the microvoids form clusters limited by the size of the microvoids, i.e. less than 2–3 nm.

145
The copper metal clusters are coherent inclusions in the bulk of zinc oxide.

146
(iii) The oxidation of surface copper metal particles by oxygen (at its low partial pressure) brings major copper back to the bulk of a.m.-ZnO crystal.

147
The copper atoms are localized at the stacking faults in the a.m.-ZnO structure.

148
Copper metal particles in the form of 2–3 nm clusters in the microvoids of the bulk of the a.m.-ZnO structure are inaccessible to the oxygen and are not oxidized.

149
The portion of preserved Cu0 after the treatment in O2-containing mixture is not so high, since it is not detected by XRD.