2.1. X-ray diffractometer

The X-ray diffractometer is configured in Debye-Scherrer mode. It is composed of a molybdenum tube (_K1 = 0.709 Å, _K2 = 0.71359 Å, E = 18 keV) and an INEL CPS 120 curved detector (Evain et al., 1993; Ballon et al., 1983) (Fig. 2). An elliptic mirror from AXO Dresden GmbH (http://www.axo-dresden.de/mainframe_products.htm ), 60.5 mm long, 200 mm wide and 10.5 mm high, suppresses the Mo K radiation and focuses the beam on the detector. The detector has a curvature radius of 250 mm. It contains 6 bar (1 bar = 10⁵ Pa) of an argon-ethane mixture (0.85:0.15). A solid blade anode and a segmented cathode are mounted inside the detector to intercept the X-rays. The detector resolution was measured with a certified LaB₆ sample (NIST, Maryland, USA) and is equal to 0.21. The molybdenum tube provides energetic X-rays, allowing the characterization of thick samples under an absorbing atmosphere.

Figure 2 A schematic view of the X-ray diffraction system.

Diffracted X-rays are collected simultaneously by an INEL curved detector, in order to decrease the analysis time and to retain enough space for both the in situ and in operando cell and the DRIFT spectrometer. Since the centre of the detector is located above the sample, diffraction cones are collected at two points (Fig. 3). Finally, the recorded diffraction patterns are added for signal exploitation.

Figure 3 A raw diffraction diagram before folding.

2.2. DRIFT spectrometer

The DRIFT spectrometer is an IRcube OEM FT-IR spectrometer provided by Bruker (Bruker Optics, http://www.bruker.com/ ). The mercury cadmium telluride detector allows the analysis of wavenumbers between 12 000 and 600 cm^-1 with a resolution between 1 and 4 cm^-1. It is cooled by liquid nitrogen and has a running autonomy of 8 h. Two sets of plane and parabolic mirrors made of silica and gold, and nickel-aluminium, respectively, are used to focus the IR beam from the source to the reaction cell and from the reaction cell to the detector (Fig. 4). Both mirrors were specially developed for this system to minimize the loss of signal. Moreover, to avoid signal perturbation by atmospheric gases such as CO₂ or water vapour, the optical path is flushed with gaseous nitrogen.

Figure 4 A schematic view of the DRIFT spectroscopy system.

DRIFT spectroscopic analysis gives information on the surface of the studied material and, more precisely, on the adsorbed species and products during in situ/in operando treatment.

The reflection mode is mainly used to characterize adsorbent species and has not been adapted to quantification.

2.3. Reaction cell

The reaction cell, which mimics a fixed-bed catalytic reactor, was specially developed to characterize catalysts simultaneously by XRD and DRIFT spectroscopic measurements under working conditions at industrially relevant temperatures and pressures. Thus, the analysis chamber is able to support pressures up to 18 bar and temperatures up to 873 K under atmospheric pressure. Moreover, different gases can be used in the current configuration for in situ analysis, namely H₂, CO or N₂. The gas enters the analysis chamber flowing up and leaves the cell by passing down through the sample (Fig. 5).

Figure 5 A schematic view of the reaction cell.

A heating resistor rolled into a cylinder allows thermal treatments. A porous silica fritted disc placed in the cylinder acts as a support for the powder sample (labelled catalyst in Fig. 5), from a few milligrams to 200 mg.

Specific windows allow the X-ray and IR beams to reach the sample. For X-ray measurements, a set of beryllium windows, each 1.5 mm thick, is used, located at the top and bottom of the cell. For IR analysis, ZnSe windows, each 3 mm thick, are installed perpendicular to the X-ray plane.

2.4. Online gas chromatograph

A gas chromatograph (GC) was installed online after a condenser to allow the online analysis of gases. The gas analyser contains a flame ionization detector (FID) and two thermal conductivity detectors (TCDs). The analytical schema is presented in Fig. 6. The Pona column and FID allow the separation and detection of hydrocarbons such as paraffins, olefins and oxygenates. The first TCD, with Ar as the vector gas and PPQ and MS5A (from Agilent) as chromatographic columns, enables the detection and quantification of hydrogen. The second TCD is used with helium and associated with PPU and MS5A columns (from Agilent) for N₂, CO, CO₂, light hydrocarbon and O₂ analysis.

Figure 6 A schematic view of the online gas chromatograph. Abbreviations are defined in the text.

3.1. Experimental

Thermal decomposition of a calcium oxalate sample (Carlo Erba) was carried out at atmospheric pressure under a flow of 16 ml min^-1 of N₂. The heating ramp was 5 K min^-1 from 293 to 853 K. For XRD and DRIFT spectroscopy analyses, sample characterization was performed every 20 K. About 50 mg of calcium oxalate in powder form was placed on the fritted disc inside the reaction cell. A thermocouple was inserted into the calcium oxalate powder. During the analysis, the temperature was kept constant. 200 DRIFT spectra (resolution 4 cm^-1) were recorded and represented on the Kubelka-Munk scale. The background used for the IR measurements was recorded at room temperature on dehydrated KBr (Merck). XRD measurements were carried out for 200 s (resolution of 4096 channels). The XRD pattern of the fritted disc and the beryllium windows was subtracted from the sample patterns.

In parallel, thermal decomposition of calcium oxalate was studied by TGA. The evolution of the sample weight was followed by a Sétaram TGA thermoscale. Calcium oxalate (46.18 mg) in powder form was placed in the basket of the balance. The experiment was carried out under nitrogen (50 ml min^-1) with a heating ramp of 10 K min^-1 from 298 to 1073 K.

3.2. Results and discussion

The thermal decomposition of calcium oxalate CaC₂O₄·H₂O powder was followed using TGA (Fig. 8). Two decomposition steps can be observed. The first occurs around 483 K (peak temperature) and represents a loss of 5.68 mg (10.8 wt%), corresponding to the loss of the water (step 1). The second decomposition step occurs around 743 K (peak temperature) and represents a loss of 8.80 mg (21.72 wt%). This weight loss can be explained by the formation of CO and calcium carbonate, CaCO₃ (step 2). These experimental weight losses are in good agreement with the theoretical ones.

Figure 8 Thermogravimetric analysis plot for calcium oxalate (46 mg, 10 K min-1).

After being studied by TGA, the thermal decomposition of calcium oxalate was followed inside the reaction cell by XRD and DRIFT spectroscopy. Fig. 9 shows the evolution of the XRD patterns of calcium oxalate as a function of temperature. The initial product is stable up to 413 K. It presents diffraction lines at 2 = 6.85, 7.01, 11.15, 13.74, 16.34, 17.38 and 19.68° (Fig. 10), characteristic of CaC₂O₄·H₂O (Hocart et al., 1965). At 413 K it starts to decompose into another product. Stable dehydrated CaC₂O₄ with diffraction lines at 2 = 6.50, 6.85, 7.76, 9.45, 10.83, 13.72 and 16.82° is obtained at 553 K (Hochrein et al., 2008). Finally, this product starts to decompose under nitrogen at 733 K into a new crystallographic structure. From 813 K onwards, the final product exhibits diffraction lines at 2 = 10.55, 13.41, 16.34, 17.86, 19.49 and 21.37°, which correspond to CaCO₃ (Pilati et al., 1998).

Figure 9 X-ray diffraction patterns recorded during the thermal decomposition of calcium oxalate (50 mg, 5 K min-1); blue traces represent stable crystallographic phases and orange diffraction traces indicate intermediate crystallographic structures.

Figure 10 X-ray diffraction patterns recorded at 293, 513 and 833 K.

The relative heights of the carefully selected diffraction peaks of calcium oxalate hydrate (at 2 = 19.68°), dehydrated calcium oxalate (at 2 = 9.45°) and calcium carbonate (at 2 = 16.34°) were measured as a function of temperature as obtained from the XRD patterns, and they are shown in Fig. 11. The rate of decomposition of calcium oxalate hydrate reaches a maximum between 453 and 513 K, whereas dehydrated calcium oxalate decomposes very quickly between 733 and 793 K. These decomposition temperature ranges are in agreement with those obtained from TGA, despite the small difference in heating rates.

Figure 11 Relative heights of the calcium oxalate hydrate peak at 2 = 19.68°, the dehydrated calcium oxalate peak at 2 = 9.45° and the calcium carbonate peak at 2 = 16.34°, from the XRD measurements presented in Fig. 9.

In parallel with the XRD measurements, DRIFT spectra of the sample were recorded (Fig. 12). At 293 K the spectrum is characteristic of the hydrated calcium oxalate structure (Bellamy, 1958). The broad band between 3700 and 2600 cm^-1 can be attributed to the water. The band at 1606 cm^-1 corresponds to C=O vibrations, that at 1330 cm^-1 to C-O^- vibrations and that at 781 cm^-1 to O-(CO) deformations. At 473 K, the IR spectrum can be attributed to calcium oxalate. Vibration and deformation frequencies remain the same, except for the H-O vibrations of the water observable between 3700 and 2600 cm^-1: this band disappears during the thermal treatment, as a result of dehydration of the sample. Finally, after thermal treatment at 853 K, the sample exhibits a spectrum representative of calcium carbonate. The band at 1795 cm^-1 can be attributed to C=O double-bond vibrations (Bellamy, 1958), and the bands at 882 and 714 cm^-1 to O-(CO) deformations (two deformation modes). The absorption band at 1444 cm^-1 is characteristic of C-O vibrations.

Figure 12 DRIFT spectra recorded during thermal decomposition of calcium oxalate (50 mg, 5 K min-1).

Comparing the results obtained by TGA and by XRD and DRIFT spectroscopy measurements, the range of decomposition temperatures is consistent whichever characterization technique is used. This observation demonstrates that the reaction cell is suitable for the thermal treatment of powders. Moreover, the information obtained from DRIFT spectroscopy analysis and XRD measurements is in agreement with the expected results.

It is important to point out that XRD measurements provide a volume analysis of the material, whereas DRIFT spectroscopy is only a surface analysis. Using these two techniques in tandem thus requires extreme care when making comparisons.