Combustion synthesized crystalline La-Mn perovskite catalysts: Role of fuel molecule on thermal and chemical events

3.1 Powder preparation

Aqueous solution of lanthanum and manganese nitrates was prepared in the stoichiometric ratio 1La:1Mn in distilled water at a total dissolved salt concentration of 4.13 mol/L. The required amount of organic fuel molecule for a SCS stoichiometric operating conditions (φ = 1) was added to the nitrates mixture and thoroughly stirred to reach complete dissolution of all solid reagents. The following fuel molecules were used: Urea (U), Acetic acid (aA), Glycine (G), Maleic acid (mA), Succinic acid (sA) and Citric acid (cA). The solutions were then boiled on a hot plate preheated to about 300 °C to evaporate excess water. The resulting viscous liquid ignited and underwent self-sustaining combustion with evolution of large amount of gases and leaving behind a residual black colored fine powder. In the following, the obtained samples were designated LMU, LMaA, LMG, LMcA, LMsA and LMmA when using respectively Urea (Applichem, 98%), Acetic acid (Prolabo, 99%), Glycine (Applichem, 99%), Citric acid (Prolabo, 99.7%), Succinic acid (Merck, 99%) and Maleic acid Charlou, 99%) as fuel molecules. Samples that had undergone supplementary thermal treatment (at 700 °C for 24 h under air) were designed LMUc, LMaAc, LMGc, LMcAc, LMsAc and LMmAc.

The SCS temperature-time profiles during the combustion reactions were monitored with a K-type thermocouple placed within the starting solution and interfaced with a computer data acquisition and management system. The temperatures reported in the profiles should be considered only for comparison purposes. Owing to the delay of the signal after the sudden change of the system temperature, the values given by the thermocouple were lower than those reached within the combustion bed.

Evolved gaseous products during the combustion reaction were qualitatively identified as follows: carbon dioxide was identified by the precipitation of barium carbonate from a solution of barium chloride. Ammonia was identified by color change of an aqueous solution of HCl and phenolphthalein as an indicator which went from colorless to pink. NO₂ was identified by its brown color.

3.2 Powder characterization

XRD analysis (Phylips X’PERT, Cu Kα radiation) were performed on all as-synthesized samples to assess the presence and the purity of the expected phases and to gather information about their degree of crystallization. The same analysis was repeated on all samples that had undergone thermal treatment to verify to what extent they had been affected, both in terms of further crystallization and or appearance of new phases.

The infrared spectra with KBr discs were recorded on a Nicolet IR 200 Spectrometer (4000–400 cm⁻¹).

BET specific areas (S_BET) were determined by N₂ sorption at −196 °C in a volumetric all glass apparatus. Prior to each measurement, the samples were degassed 4 h at 200 °C under vacuum (7 · 10⁻⁴ Pa).

Thermogravimetric analyzer (TGA/dTG) interfaced with a data acquisition and management system and S-type thermocouple, was used to study the thermal evolution of the as-prepared samples. A heating rate of 10 °C/min was used in all measurements up to 900 °C in flowing air with no holding period.

The oxygen over stoichiometry,δ in LaMnO_3+δ was determined by iodometric titration. Typically, about 20 mg of the powder was dissolved in 10 mL of concentrated HCl containing an excess of potassium iodide. Generated iodine was titrated against standard sodium thiosulphate solution using starch as indicator.

3.3 Catalytic activity

Catalytic activity tests were performed, on as-prepared catalysts, in deep oxidation reaction of ethanol. The catalyst charge (0.1 g of catalyst diluted 1:10 in quartz powder) was loaded in a U-shaped quartz microreactor which was operated in a down-flow mode at atmospheric pressure. The reactor was placed in a tubular furnace where the temperature was controlled with a K-type thermocouple placed alongside the reactor and its tip inserted in the catalytic bed. The space velocity was 42,000 N mL g⁻¹ h⁻¹. The gaseous flow rates were mixed to obtain inlet concentration of 2% ethanol, 40% oxygen and helium as balance. The feed and reaction products were analyzed with an inline Intersmat gas chromatograph equipped with a porapack Q column and thermal conductivity detector. For each experiment, the ethanol conversion rate was calculated as the average of at least three measurements.

In the used reaction conditions, the only detected products were water, carbon dioxide and acetaldehyde (ACA). ACA production was found to pass through a maximum while that of CO₂ increased monotonously with increasing reaction temperature over all used catalysts.

On the basis of the products found in this study, the oxidation of ethanol involved the following reactions:

The conversion of ethanol (R) and the percentage of ethanol converted to CO₂ (R1) and ACA (R2) were calculated according to the following equations: $$\text{R}=\frac{{[\text{EtOH}]}_0-[\text{EtOH}]}{{[\text{EtOH}]}_0}\times 100\text{R}1=\frac{\frac{[{\text{CO}}_2]}{2}}{{[\text{EtOH}]}_0}\times 100\text{R}2=\frac{[\text{ACA}]}{{[\text{EtOH}]}_0}\times 100$$ where [EtOH]₀ was the inlet ethanol concentration and [EtOH], [CO₂], [ACA] were the outlet concentrations of ethanol, carbon dioxide and acetaldehyde respectively.

The conversion was investigated in a reaction temperature ranging from 50 to 330 °C for all catalysts. Each temperature was maintained for 90 mn to attain steady state conversion. No changes were detected in the catalytic activity of all samples when the reaction was carried out in two cycles of increasing and decreasing temperatures.

In the blank experiment using empty reactor, it was stated that contribution of the homogeneous reactions were not significant because less than 7% ethanol was converted to CO₂ at the highest reaction temperature.

4.1 Effect of fuel on the combustion

Fig. 1 shows the typical temperature-time plots of the various precursors prepared with different organic fuels. These results were consistent with those reported by several authors who showed that for similar operating conditions with glycine or urea, the temperature profiles had only one sharp maximum while with citric acid, two maxima were observed indicating that occurrence of secondary reactions may be at the origin of the differences in the thermal behavior of these systems (Deshpande et al., 2004, Hwang et al., 2005, Civera et al., 2003, Mali and Ataie, 2004). The temperature history of Fig. 1A showed a single exothermic peak at 130–170 °C for initial combustion mixtures containing glycine, maleic acid and succinic acid indicating that the chemical reaction took place very rapidly (i.e. the slope of temperature increase dT/dt was very steep). The system was then observed to reach a maximum temperature (T_max) in a very short time which increased in the order succinic acid > maleic acid > glycine. The cooling period started just after the maximum temperature had been reached, thus indicating the completion of the primary exothermic reaction. Further, this combustion involved substantial gas evolution whose qualitative analysis showed the presence of CO₂ (Table 2, column 7).

Close

Fig. 1

Temperature versus time plots during the synthesis of LaMnO_3+δ using (A) glycine (G), maleic acid (mA), succinic acid (sA) and (B) acetic acid (aA), citric acid (cA) and urea (U) as fuel molecules.

Export to PPT

The temperature-time profiles recorded with citric acid, acetic acid and urea as fuels are given in Fig. 1B. Comparing with the preceding organic fuels, the SCS reaction with citric acid appeared more moderate while those with urea and acetic acid appeared to undergo self-propagating and non-explosive reactions. The profile and evolved gas analysis showed that decomposition process of the precursor with citric acid occurred in two well-separated stages over a wide range of temperature (Table 2, column 7). The first one was ignited at about 170 °C and reached a maximum temperature of 220 °C. This exothermic reaction took place very rapidly, involving formation of a rather large quantity of NO₂ and CO₂. After a slight decrease of the temperature, the second exothermic step occurred at an onset temperature of 250 °C and reached a maximum value of 320 °C.

Several different pathways are observed during thermal treatment of the mixture with urea. This latter underwent a reaction which happened with a strong and vigorous NH₃ and CO₂ emission at a beginning temperature of 150 °C (Table 2, column 7). A plateau with a constant temperature of 250–260 °C associated to NO₂ emission was observed just before the appearance of an exothermal phenomenon which took place at a temperature of 317 °C and reached a maximum temperature of 410 °C. It can be inferred from these observations that several endothermic reactions including decomposition reaction of urea and manganese nitrate occurred before a combustion reaction.

The temperature profile recorded with acetic acid showed a continuous increase of the temperature and no evidence for a thermal event due to the combustion reaction was observed. This process happened with continuous emission of acetic acid, CO₂ and NO₂. This was very likely related to the volatility of acetic acid (ebullition temperature = 118 °C) and decomposition of a mixed nitrate-fuel complex.

4.2 Powders properties

Powders characteristics like, crystalline structure, oxygen excess δ in LaMnO_3+δ, specific surface area (SSA) were primarily governed by the heat generated during combustion, which itself depended on fuel chemical nature. XRD patterns of as-prepared products shown in Fig. 2, revealed the better crystallization of LMG, LMmA and LMsA (Fig. 2A) as compared to LMcA, LMaA and LMU (Fig. 2B). This confirmed the more and less completion of the SCS process respectively for the former and latter samples production. As can be seen, only the characteristic peaks of the perovskite phase LaMnO_3+δ were observed for LMG, LMmA and LMsA samples. The combustion reaction with glycine as fuel produced only the low temperature rhombohedral perovskite phase (ASTM: 01-086-1227) (Table 2, column 2). For LMsA sample, the splitting and broadening of XRD peaks were indicative of a supplementary distortion from the rhombohedral to the orthorhombic structure with Pbnm space group (ASTM: 01-085-2218). This could be accounted for the relatively higher temperature reached within the reacting system with succinic acid. Between the orthorhombic and rhombohedral phase field, a two phase region was observed for LMmA sample. The X-ray pattern of this sample exhibited several asymmetric diffraction peaks due probably to the coexistence of two structures (rhombohedral and orthorhombic).

Close

Fig. 2

XRD patterns of the as-prepared and aged powders using glycine, maleic acid and succinic acid (A and C) and urea, acetic acid and citric acid (B and D) as fuels.

Export to PPT

The XRD patterns shown in Fig. 2B indicated the presence of an amorphous phase for LMU powder and the perovskite as a major phase and some amorphous impurities for LMaA and LMcA samples. The diffractogram of LMU sample showed quite low relative intensity, and its noise-like character indicated a completely amorphous phase (Fig. 2B). Better crystallinity was observed for the as-prepared LMcA and LMaA compared to LMU whose X-ray patterns were consistent with the presence of a rhombohedral perovskite oxide as the main phase (Fig. 2B). Even so, it seemed that the background noise of the two diffractograms was indicative of the presence of a supplementary amorphous phase which amount was higher for LMaA sample than LMcA.

Fig. 3 depicts IR spectra of the as-prepared LMU, LMaA and LMcA samples. Three regions of vibrational modes could be distinguished: 3650–3000 cm⁻¹ (OH vibrations), 3000–1900 cm⁻¹ (combination and overtone modes of nitrate vibrations) and 1600–750 cm⁻¹ (nitrate vibrations). The observed absorption bands in the spectral region 1600–750 cm⁻¹ occurred at similar frequencies as those observed for La(OH)₂NO₃·H₂O in which the local symmetry of the nitrato group was reduced from D_3h (free nitrate ion) to C_2v (bonded to La through one oxygen) point group (Klingenberg and Albert Vannice, 1996, Addison and Sutton, 1967, Steven and Jones, 1999). Moreover, the band observed at 500–580 cm⁻¹ was in the spectral range of Mn-O stretching normal vibration where the motion was primarily of a change in the length of Mn-O bond (Subba Rao et al., 1970, Blasse, 1975, Gangully and Vasanthacharya, 1986). The observed absorption bands and their assignment were gathered in Table S1. These results proved that the insufficient heat liberated from the reaction between metal nitrates and urea, citric acid and acetic acid caused an incomplete reaction that was not the same as the one given by Eq. (2). Consequently, part (with citric acid and acetic acid) or totally (with urea) of lanthanum and manganese remained distinctly combined in two amorphous phases (La(OH)₂NO₃,MnO_x) that were intimately mixed. Thus additional thermal treatment would be necessary in order to achieve LaMnO₃ perovskite of reasonable purity.

Close

Fig. 3

IR spectra of as-prepared LMU, LMaA and LMcA powders.

Export to PPT

Indeed, in order to verify to what extent samples had been affected, both in terms of further crystallization and/or appearance of new phases, additional experiments were conducted by submitting powders to a temperature scan in TGA-dTG equipment and by recording XRD patterns after equilibrations of the as prepared samples under air (P_O2 = 0.2 atm.) at 700 °C for 24 h. After annealing up to 700 °C, the presence of a single rhombohedral perovskite phase with space group. R-3C was evidenced from XRD patterns of all prepared powders (Fig. 2C and D). While the gain in crystallinity was not dramatic for the three samples LMG, LMmA and LMsA, the transformation from the orthorhombic to rhombohedral phase was clear for LMmA and LMsA solids. It was thought that this transformation was due to the slower reaction rate during the annealing operation compared to the extremely rapidly transformation occurring in the SCS process with maleic and succinic acids as fuel molecules. This allowed achieving the redox Mn³⁺ ↔ Mn⁴⁺ equilibrium hence the thermodynamic oxygen content in the perovskite phase LaMnO_3+δ. These results were consistent with the determined oxygen excess values δ, reported in Table 2, column 6. For the as-prepared samples, the oxygen over-stoichiometry values were δ = 0.15; δ = 0.12 and δ = 0.09 respectively for LMG, LMmA and LMsA. These values were consistent with the well-known symmetry change from orthorhombic (found for LMsA) to primitive rhombohedral (found for LMG) for a δ value, greater than 0.105 (Tofield and Scott, 1974, Van Roosmalen et al., 1994, Krogh Andersen et al., 1994). The determined value δ = 0.15 ± 0.01 after annealing under air indicated a supplementary oxidation of LMmA and LMsA oxides inducing the orthorhombic/rhombohedral transformation.

The resulted crystallization in a single rhombohedral phase (Fig. 2 D) after a supplementary heat treatment of LMU, LMcA and LMaA samples was consistent with the determined δ value of 0.15 ± 0.01 (Table 2, column 6). However, values determined for fresh samples LMcA and LMaA (δ > 0.2) should be considered less accurate due to the presence of amorphous phases.

TGA/dTG runs were performed on the as-prepared samples from room temperature to 900 °C (Fig. 4). Fig. 4 A showed that the three as-prepared samples LMU, LMaA and LMcA displayed higher mass loss during heating than that recorded for LMG, LMmA and LMsA (Fig. 4B). This was consistent with the completely or partially amorphous states observed by XRD analysis. TG curves suggested that a three-step process occurred corresponding to a total mass loss of 32%, 20% and 7% respectively for LMU, LMaA and LMcA. Note that the shape of the recorded TG curves were very similar to those reported in previous investigations for the thermal decomposition of La(OH)₂NO₃ compound (Mentus et al., 2007, Gobichon et al., 1997, Haschke, 1974). Moreover, it was shown that, the completeness of nitrate → oxide thermal conversion occurred up to 550°C. However, the recorded TG curves (Fig. 4A) indicated that the decomposition process did not complete up to 550 °C but exceeded to approximately 700–750 °C. The presence of some carbonates such as lanthanum dioxicarbonate La₂O₂CO₃ could be responsible for the enlargement of limiting decomposition temperature. The TG curves and their derivatives showed that the thermal decomposition carried under air took place roughly through three successive stages. The entire scheme of the corresponding chemical reactions may be summarized as follows: La(OH)₂NO₃ → LaONO₃ → La₂O₂CO₃ → La₂O₃

Close

Fig. 4

TG/DTA/DTG curves of as-prepared LMU, LMaA and LMcA powders (A) and LMG, LMmA and LMsA powders (B).

Export to PPT

Note that:

1. in the above scheme, we assumed that the starting La(OH)₂NO₃ in the three samples was anhydrous. If not, other dehydrating steps might be considered,

2. the high reactivity of LaONO₃ with atmospheric CO₂, in air, allowed assuming the presence of La₂O₂CO₃ as intermediary compound,

3. all decomposition steps were shifted toward lower temperatures for LMU compared to LMaA and LMcA samples (Fig. 4). This might be due to the difference in powder composition and the degree of hydration of the starting lanthanum dihydroxynitrate,

4. the as-formed La₂O₃ in the third step reacted with the amorphous MnO_x oxide to yield the perovskite phase.

Considering TGA/dTG curves of Fig. 4B, the mass loss of LMG sample happened in two steps of about 1 and 2% in the temperature range 50–200 °C and 350–700 °C which were probably due respectively to the desorption and decomposition of residual water and carbonate species. This mass loss was interrupted by a mass increase at about 550 °C for LMmA and LMsA. This mass gain was consistent with the observed increase of oxygen content δ corresponding to the orthorhombic-rhombohedral transformation after equilibration under air (Table 2).

The SSA values reported in Table 2 showed a perceptible decrease in the order LMaA (27 m² g⁻¹) > LMcA (21 m² g⁻¹) > LMG (17 m² g⁻¹) > LMmA (11 m² g⁻¹) > LMsA (6 m² g⁻¹) ∼ LMU (5 m² g⁻¹). The evolution of these values could be understood taking into account the heat generated by the fuel combustion and the volume of escaped gaseous products during the combustion process (Table 1). Indeed, an increase of the energy released could rapidly heat the system to a high temperature favoring the crystallization of the perovskite phase but particles agglomeration hence a decreases of SSA. On the other hand, a consequence of the production of a large volume of escaped gas was more likely to break up the agglomerates which produced a powder with more porosity and high SSA. Our results tended to indicate that the combustion enthalpy played a more important role since the amount of evolved gas during the combustion process (calculated taking into account the reaction given by Eq. (2)) using glycine, maleic and succinic acid was almost equal. Thus, relatively lower amount of released energy with glycine as fuel seemed to have a boosting effect on SSA. It was thought that in these conditions, perovskite had the time to form but not to sinter and the final SSA was higher. The specific surface area of the as-synthesized amorphous LMU powder was significantly lower than that of partially crystallized LMaA and LMcA samples (Table 2).

4.3 Mechanism of SCS of LaMnO_3+δ

These observations led us to hypothesize the existence of two types of combustion reactions of fuel-nitrate precursors in the SCS process. Type I was characterized by a vigorous exothermic reaction which occurred in a one stage process implying a semi-decomposed precursor consisting of a mixed nitrate-fuel complex. This decomposition reaction occurred simultaneously with manganese nitrate decomposition. Type II was typified by a multi stage process in which intermediate decomposition reactions occurred before the formation of a mixed nitrate-fuel complex. It was apparent that the difference between type I and type II processes was simply associated with the depletion of fuel (type II) in the sample and the stability of the mixed nitrate-fuel complex produced on loss of the nitrate content from the precursor.

To test this hypothesis, additional experiments conducted in binary systems (Fig. S1), such as La nitrate-Urea and Mn-nitrate-Urea showed that only in the last case, reaction occurred in combustion mode with a temperature-time profile essentially the same as with La,Mn-nitrates-Urea mixture (Fig. 1). Further, there was an evident formation of NH₃ and NO₂ during Mn(NO₃)₂-Urea combustion and only NH₃ during La(NO₃)₃-Urea one. However, with glycine as fuel, the temperature-time profiles showed a single exothermic peak for both Mn(NO₃)₂-Glycine and La(NO₃)₃-Glycine systems. The two combustion processes happened with a vigorous emission of CO₂ gas. This indicated that the SCS process for production of lanthanum manganite powders was controlled by the reaction between a mixed nitrate-fuel complex and product of Mn-nitrate decomposition.

Moreover, manganese (II) nitrate thermal decomposition had been investigated and TG/dTG curves were given in Fig. S2. Results indicated at least three distinct reaction zones in the temperature range 50–400 °C. The first zone, 50–140 °C, caused a 25% weight loss, which was very close to that expected (25.1%) for release of 4 mol of water. The second step corresponded to the loss of remaining water and the first step in the nitrate decomposition. Indeed, Fig. S2 indicated a weight loss level of 29% which was near to that expected (28.6%) for the production of an equimolar mixture of Mn(NO₃)₂ and MnO₂. The third step was associated with 15% weight loss which was close to that theoretically calculated (16%) for the decomposition of remained nitrate to MnO₂. The entire scheme of thermal decomposition may be summarized as follows:

4.4 Catalytic study

A typical plot of R1 and R2 versus R was shown in Fig. 5 for as-prepared LMaA and LMG catalysts. The curves for the other samples had the same shapes than those of LMaA and LMG (Fig. S3).

Close

Fig. 5

Typical plot of R1 and R2 versus R for LMaA and LMG. Several reaction temperatures (°C) are added in parenthesis.

Export to PPT

The products distribution indicated that ACA selectivity reached a maximum value of about 45–55% at ethanol conversion of 75–90%. Moreover, total conversion of ethanol as well as complete oxidation of 100% ethanol to form CO₂ and H₂O occurred at different temperatures indicating different catalytic performances. For this purpose, plots of Fig. S3 were used to derive the following parameters for each catalyst in order to compare their relative activity. T(R-100)) was the temperature at which no ethanol was detected in the outlet gaseous phase. In general, stoichiometric amounts of CO₂ and ACA were found; therefore it was inferred that all ethanol was consumed at this temperature. T(R1-100) was the temperature to reach 100% selectivity to CO₂. It was determined solely from the CO₂ formation and could be higher than T(R-100) because part of ethanol could be converted to ACA and not recovered as CO₂ at T(R-100).

Deduced T(R-100) and T(R1-100) values were reported in Fig. 6 as a function of specific surface area. For all catalysts, complete oxidation of 100% ethanol to form selectively CO₂ and H₂O can be realized between 210 and 330 °C while complete conversion of ethanol was realized between 200 and 310 °C. Note that for other perovskite systems such as LaMnO₃ (Shimizu, 1986), LaFeO₃ (Barbero et al., 2006) and LaCoO₃ (Bialobok et al., 2007) prepared by conventional methods (solid state, citrate, etc.), 100% conversion occurred at a temperature higher than 250 °C. If one consider that the increase of CO₂ selectivity is an important result because ACA is a more noxious product than ethanol, the order of catalytic activity of the total combustion of ethanol to CO₂, the reverse order of their T(R1-100) values, was LMG > LMcA > LMaA > LMmA > LMsA > LU. The best catalytic activity was developed by LMG catalyst which was probably associated with the synergetic effect of a high crystallinity of a rhombohedral pure perovskite phase and a high specific surface area.

Close

Fig. 6

Plot of temperature of 100% ethanol conversion (R-100) and of 100% CO₂ selectivity (R1-100) as a function of specific surface area of as-prepared samples.

Export to PPT