Yuan
He
a,
Lei
Shi
a,
Fan
Wu
b,
Weiwei
Xie
c,
Shu
Wang
a,
Dong
Yan
a,
Peijiang
Liu
a,
Man-Rong
Li
d,
Jürgen
Caro
*e and
Huixia
Luo
*ae
aSchool of Materials Science and Engineering, Key Laboratory for Polymeric Composite and Functional Materials of Ministry of Education, Sun Yat-Sen University, No. 135, Xingang Xi Road, Guangzhou, 510275, P. R. China. E-mail: luohx7@mail.sysu.edu.cn
bHarvard John A. Paulson School of Engineering and Applied Sciences, Harvard University, 410 Gordon McKay Laboratory of Applied Science, 9 Oxford St., Cambridge, MA 02138, USA
cDepartment of Chemistry, Louisiana State University, Baton Rouge, Louisiana 70803-1804, USA
dMOE Key Laboratory of Bioinorganic and Synthetic Chemistry, School of Chemistry, Sun Yat-Sen University, No. 135, Xingang Xi Road, Guangzhou, 510275, P. R. China
eInstitute of Physical Chemistry and Electrochemistry, Leibniz University of Hannover, Callinstr. 3A, D-30167 Hannover, Germany. E-mail: juergen.caro@pci.uni-hannover.de
First published on 30th October 2017
Oxygen permeation, stability and chemical bonding characteristics of 40 wt% Nd0.6Sr0.4CoO3−δ–60 wt% Ce0.9Nd0.1O2−δ (40NSCO–60CNO) dual-phase composite membrane reactors were investigated. The 40NSCO–60CNO oxygen permeable membrane was prepared via an in situ one-pot one-step EDTA–citric acid method. The crystal structure of the 40NSCO–60CNO dual phase material was characterized by X-ray diffraction (XRD) and in situ XRD. The microstructure was investigated using transmission electron microscopy (SEM) and high resolution transmission electron microscopy (HRTEM) combined with energy-dispersive X-ray spectroscopy (EDXS) and electron energy-loss spectroscopy (EELS). The results show that the 40NSCO–60CNO composite represents a micro-scale mixture of only the two pure phases NSCO and CNO. The oxygen permeation fluxes through the 40NSCO–60CNO dual phase membrane were measured at elevated temperatures (900–1000 °C) with one side of it exposed to synthetic air and the other side to a flowing He gas stream. A stable oxygen permeation rate of 0.90 mL cm−2 min−1 was obtained with a 0.4 mm thick membrane under an air/He oxygen partial pressure gradient at 1000 °C. The 40NSCO–60CNO dual phase membrane with a thickness of 0.6 mm showed a stable oxygen flux of 0.55 mL cm−2 min−1 at 950 °C for 100 h under pure CO2 sweeping.
Depending on the number of phase compositions, OTMs generally can be divided into two types: single-phase and two-phase oxygen transport membrane materials. Single perovskite-type oxygen permeable materials with the general formula of ABO3 have been extensively studied over the last two decades due to their high oxygen permeability and low cost (see ref. 10–12, for example). Although great efforts have been focused on the development of single perovskite-type OTMs, many problems still hinder their practical applications, such as low mechanical strength, poor chemical stability, and unsatisfactory long-term stability.13,14 Especially, alkaline-earth containing single perovskite-type oxygen permeable materials are susceptible to CO2 contamination from air or react with CO2 and form carbonates in a CO2 atmosphere.15,16
Dual phase-type membranes, which consist of an oxygen ionic/protonic conducting (OIC/PC) phase and an electronic conducting (EC) phase in a micro-scale phase mixture, are considered to be possible alternatives to single perovskite-type membranes due to their compositions that can be tailored according to practical requirements.2,3,17–26,30 In our group, we have developed some new CO2-stable dual phase oxygen permeable membrane materials, such as NiFe2O4−δ–Ce0.9Gd0.1O2−δ(NFO–CGO),5,20 Mn1.5Co1.5O4−δ–Ce0.9Pr0.1O2−δ(MCO–CPO),8 Fe2O3–Ce0.9Gd0.1O2−δ(FO–CGO),21 Sm0.5Sr0.5Cu0.2Fe0.8O3−δ–Ce0.8Sm0.2O−δ(SSCFO–CSO),22 Pr0.6Sr0.4FeO3−δ–Ce0.9Pr0.1O2−δ(PSFO–CPO),6 and Nd0.6Sr0.4FeO3−δ–Ce0.9Nd0.1O2−δ(NSFO–CNO).23 The aforementioned dual phase membranes show high CO2 stability; however, the low oxygen permeability of most of them needs to be improved to meet the industrial application requirements. Therefore, it is highly desired to develop novel OTMs with high oxygen permeability and stability under real application conditions.
Here we report the development of a novel noble-metal free dual phase membrane material, 40 wt% Nd0.6Sr0.4CoO3−δ–60 wt% Ce0.9Nd0.1O2−δ (40NSCO–60CNO). We selected this 40–60 composition to ensure that it can achieve a threshold above both the EC (mainly NSCO) and the OIC percolation (CNO) thresholds.27–30 The 40NSCO–60CNO dual phase membrane was prepared via an in situ one-pot one-step EDTA–citric method. In this dual phase system, CNO is the main phase for ionic transport, and NSCO is the main phase for electronic transport, which also assists the ionic transport. The phase structure and stability, as well as oxygen permeability, were investigated under different atmospheres at high temperatures. Fig. 1 shows the flowchart of the main work in this paper.
In order to determine the phase stability of the 40NSCO–60CNO dual phase membrane material in air, in situ XRD measurements were performed on the 40NSCO–60CNO dual phase membrane. Fig. 3 shows the in situ XRD patterns of the sintered 40NSCO–60CNO dual phase membrane after being crushed into powder, collected in air with increasing temperature from 30 to 1000 °C. During heating and cooling, the 40NSCO–60CNO powder maintained the same crystal structure, suggesting that the 40NSCO–60CNO possesses good phase stability in air atmosphere under experimental conditions. To further check the structural stability of the 40NSCO–60CNO at reduced oxygen partial pressure, the powder of the 40NSCO–60CNO was exposed to a pure Ar atmosphere at different temperatures for 48 h. Fig. 1S† presents the XRD patterns of the 40NSCO–60CNO powder before and after treatment with Ar at different temperatures for 48 h. The crystal structure of the 40NSCO–60CNO remains unchanged even under a pure Ar atmosphere and no change in the phase composition was detected by XRD, which demonstrated that the 40NSCO–60CNO exhibits an excellent structural stability under Ar. Furthermore, Fig. 4 presents the in situ XRD patterns of the sintered 40NSCO–60CNO dual phase membrane after being crushed into powder, collected in 50 vol% CO2 and 50 vol% N2 with increasing temperature from 30 to 800 °C. During heating, the crystal structure of the composite remained unaltered, and no peaks of carbonate were detected, suggesting that the 40NSCO–60CNO possesses good structural stability under a CO2 atmosphere under experimental conditions.
The feasibility of the fabrication of the dual-phase membrane material was then examined by SEM-EDXS, STEM, and EELS. Fig. 5 presents the SEM and EDXS pictures of the as-prepared unpolished 40NSCO–60CNO dual phase membranes after sintering at 1225 °C for 5 h in air at two different magnifications. SEM observations (Fig. 5a–d) revealed that the micro-sized grains are packed closely; no major cracks are visible. In the bulk material, only a few non-connected pores were observed. The NSCO and CNO grains could be distinguished by EDXS (Fig. 5e and f). The same information is provided by EDXS (Fig. 5e and f), which suggests that the green color in the EDXS is an overlap of the Nd, Co and Sr signals, whereas the yellow color (light) stems from an average of the Ce and Nd signals. The mean grain size areas of CNO are larger than that of NSCO. Furthermore, the analysis of the phase composition was performed in STEM mode by EDXS and EELS. Fig. 6a and 7a show parts of STEM elemental mapping images of the 40NSCO–60CNO powder after being calcined at 950 °C for 10 h in air and the sintered 40NSCO–60CNO dual phase membrane after being crushed into powder, respectively. The corresponding EELSs in the energy-loss range of 750–1100 eV are shown in Fig. 6b and 7b. No intermixing of cations between the two phases can be observed (that is, CNO contains neither Sr nor Co, and NSCO contains neither Ce nor Nd), which shows that a dual-phase membrane with well separated grains was obtained by the direct one-pot method. The 40NSCO–60CNO dual phase membrane material was further examined by conducting surface mapping of different elements to provide a clearer picture of the elemental distribution within the dual phase membrane materials. The corresponding elemental mapping images for Ce, Nd, Sr and Co are presented in Fig. 6c–f and 7c–f. According to the Ce, Nd, Co and Sr maps, NSCO and CNO grains are homogeneously distributed without any agglomeration of one phase, indicating no significant reaction between the grains or the formation of grain boundary layers. To obtain more information on the phase composition, HR-TEM was conducted. By this method, different phases can be distinguished. By HR-TEM imaging of the grain boundary between a NSCO and a CNO grain, the characteristic (111) CNO and (021) NSCO in the 40NSCO–60CNO powder after being calcined at 950 °C for 10 h in air (Fig. 6g), and (100) CNO and (220) NSCO planes in the sintered 40NSCO–60CNO dual phase membrane after being crushed into powder (Fig. 7g) were identified on each side of the grain boundary. The presence of new phases could not be detected along the grain boundary. Fig. 2S† shows the fast Fourier transform images of different selected areas in the 40NSCO–60CNO powder after being calcined at 950 °C for 10 h in air. The top one shows the [021] zone axis pattern (ZAP) of Nd0.6Sr0.4CoO3, which has an orthorhombic structure. The bottom one shows the [110] zone axis pattern (ZAP) of Ce0.9Nd0.1O2, which has a cubic structure. Moreover, Fig. 3S† presents the fast Fourier transform images of different selected areas in the 40NSCO–60CNO membrane after being sintered at 1225 °C for 5 h in air after being crushed. The top one shows the [100] zone axis pattern (ZAP) of Ce0.9Nd0.1O2, which has a cubic structure. The bottom one shows the [001] zone axis pattern (ZAP) of Nd0.6Sr0.4CoO3, which has an orthorhombic structure. These results further confirmed that the 40NSCO–60CNO dual phase membrane contains NSCO and CNO grains.
Fig. 5 SEM (a–d) and EDXS (e and f) images of the 40NSCO–60CNO membrane after being sintered at 1225 °C for 5 h in air before polishing (see Fig. 2). For the EDXS mapping in Fig. 3e and f, superimpositions of the Nd Lα, Sr Kα and Co Kα (green) and Nd Lα and Ce Lα (yellow) have been used. |
Fig. 6 STEM (a), EELS (b), EDXS (c–f) and HRTEM (g) images of the 40NSCO–60CNO powder after being calcined at 950 °C for 10 h in air. |
Fig. 7 STEM (a), EELS (b), EDXS (c–f) and HRTEM (g) images of the 40NSCO–60CNO membrane after being sintered at 1225 °C for 5 h in air after being crushed. |
In order to identify the rate-limiting step of oxygen permeation of oxygen transport through our 40NSCO–60CNO membrane, we studied the oxygen permeation through 40NSCO–60CNO membranes with a different thickness at different temperatures. Fig. 8a displays the influence of the different thickness on oxygen permeation on the 40NSCO–60CNO membrane at different temperatures. As expected, the oxygen permeation fluxes of all 40NSCO–60CNO membranes increase with increasing temperature according to the Wagner equation.36,37E.g. for the membrane with a thickness of 0.4 mm, when the temperature increases from 900 to 1000 °C, the oxygen permeation flux increases from 0.56 to 0.90 mL cm−2 min−1. Fig. 8b shows the gas permeation fluxes through the 40NSCO–60CNO dual phase membrane with pure He as the sweep gas as the Arrhenius plot. The apparent activation energies of the 40NSCO–60CNO membrane with a thickness of 0.4, 0.6, and 0.8 mm were calculated to be 60, 71, and 91 kJ mol−1 in the temperature range of 900–1000 °C, respectively.
To understand the oxygen migration mechanism related to the membrane thickness, the oxygen permeation flux through the 40NSCO–60CNO dual phase membrane as a function of the reciprocal of the thickness at 900, 950, and 1000 °C, respectively, is plotted in Fig. 9. The oxygen permeation flux increases linearly with the reciprocal of membrane thickness in the thickness range > 0.4 mm, while it shows an increasing curve trend on further decreasing the thickness. This can be described in the Wagner equation:36,37
(1) |
To further confirm the rate-limiting step of oxygen permeation of the 0.4 mm thick dense 40NSCO–60CNO membrane, we studied the oxygen permeation at different temperatures. The oxygen pressure on the air side was kept at a constant value of 0.20 bar. The different oxygen partial pressures on the sweep side were measured by gas chromatography in this study. Following Jacobson A. J et al.,38–40JO2 shows a linear relationship with (Ph − Po)0.5 − (Pl − Po)0.5 according to
(2) |
(3) |
Correlating and evaluating our oxygen permeation data according to the above theory of Jacobson A. J et al.,38–40 we can state that the values of oxygen permeation fluxes JO2 are a linear function of (Ph/Po)0.5 − (Pl/Po)0.5 and not of ln(Ph/Pl) as shown in Fig. 10. From this finding, it follows that the oxygen permeation through the 40NSCO–60CNO dual phase membrane of 0.4 mm thickness is mainly controlled by the surface exchange reaction rather than by bulk diffusion in the temperature region studied.
The oxygen permeation flux through the 40NSCO–60CNO dual phase composite membrane with He or CO2 as the sweep gas is shown in Fig. 11 as a function of time. There is an initialization stage of the oxygen flux observed in the 40NSCO–60CNO membrane with He sweep gas. A lot similar results on the initialization stage of the oxygen flux observed in the dual phase OMTs in detail have been reported,41–45 which is considered to be attributed to surface exchange effects. When the oxygen flux reached a stable stage with an oxygen permeation flux of about 0.65 mL cm−2 min−1, the sweep gas was switched to pure CO2. By switching the sweep gas to CO2, the permeation fluxes of the 40NSCO–60CNO dual phase membrane slightly decreased. However, during the whole oxygen permeation test under a pure CO2 atmosphere, an oxygen permeation flux of about 0.55 mL cm−2 min−1 was obtained at 950 °C and no decrease was found. This behavior was also observed in the previous studies of a CO2-stable dual phase membrane which is ascribed to the slight inhibition effect of CO2 on the oxygen surface-exchange.5,6,8 This behavior is different to those of alkaline-earth metal containing single perovskite membranes,15 and some previously reported dual phase membranes. E.g. Kharton et al. have reported that obvious time-dependent degradation of the oxygen permeation flux through a Ce0.8Gd0.2O2−δ–La0.7Sr0.3MnO3−δ dual phase composite membrane was observed due to the formation of Sr(Ce, Ln)O3−δ (Ln is Gd, La) in the grain boundaries, which can block the ionic transport.40
More recently, Kaveh Partovi et al. have reported that the oxygen permeation flux through the 40% Sm0.3Sr0.7Cu0.2Fe0.8O3−δ −60% Ce0.8Sm0.2O2−δ dual phase membrane decreased with increasing time under pure CO2 as the sweep gas at 950 °C due to the formation of alkaline-earth carbonates.18 From the stable oxygen permeation fluxes on our 40NSCO–60CNO, we can exclude chemical reactions between the two NSCO and CNO phases and the formation of alkaline-earth carbonates under CO2 atmospheres involved. Fig. 12 shows that the XRD patterns of the fresh and spent 40NSCO–60CNO dual phase membrane in the long term oxygen permeation measurements with pure CO2 as the sweep gas are identical, indicating that the dual phase membrane displays a high CO2 stability.
Fig. 12 XRD patterns of the fresh and spent 40NSCO–60CNO dual phase membrane in the long term oxygen permeation measurements with pure CO2 as the sweep gas. |
Fig. 13 summarizes the oxygen permeation fluxes through several types of dual phase membranes, where the oxygen permeation flux through our 40NSCO–60CNO membrane is comparable with or even higher than those of other dual phase-type membranes reported in the open literature.5,6,8,20,21,23E.g. the 40 wt% Nd0.6Sr0.4FeO3−δ–60 wt% Ce0.9Nd0.1O2−δ (40NSFO–60CNO) dual phase membrane with a thickness of 0.6 mm exhibits an oxygen permeation flux of 0.21 mL cm−2 min−1 under an air/CO2 oxygen partial pressure gradient at 950 °C,21 which is nearly two times lower than that of our 40NSCO–60CNO membrane. To further investigate the oxygen vacancies in different oxygen permeation membranes, the electronic structure calculations, in particular, the comparison of chemical bonding analysis for 40NSCO–60CNO and 40NSFO–60CNO were conducted. The oxygen ion conductor Ce0.9Nd0.1O2−δ was calculated to be a semiconductor with a band gap of 2.9 eV (see Fig. 14a), which is consistent with a previous report.23 During the calculation for the mixed conductors NSFO and NSCO, in the valence orbital region of the electronic DOS curve for the hypothetical model “NdFeO3”, as shown in Fig. 14b, there are two distinct regions separated by the direct gap in the DOS: (i) below, show mostly O 2p bands mixed with Fe 4s and 3d and Nd 6s/5d valence orbitals; and (ii) above, which contains mostly O 2p bands and Nd 6s and 5d character. A clear band gap at 0.5 eV above the Fermi energy (EF) would suggest NdFeO3 to be an insulator. By adjusting the valence electrons, we approach the Fermi levels for NSFO and NSCO marked in blue and pink separately in Fig. 12b. In line with the chemical valence argument, the Fe/Co–O and Nd/Sr–O COHP curves indicate the orbital interactions in “NSFO” and “NSCO”. Based on the chemical bonding analysis, it can be easily found that the Co–O interaction is located on the maximum antibonding part in the –COHP, which indicates the instability of Co–O interactions. This instability makes it highly possible to produce more oxygen vacancies to lower the Fermi level and reduce the Co–O antibonding interactions, which account for the much higher oxygen permeation flux for 40NSCO–60CNO compared to that of 40NSFO–60CNO.
Fig. 13 Steady-state oxygen permeation flux (JO2) though different dual phase membranes in disk geometries. |
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c7ta07842k |
This journal is © The Royal Society of Chemistry 2018 |