Denis J.
Cumming
*a,
Christopher
Tumilson
ab,
S. F. Rebecca
Taylor
b,
Sarayute
Chansai
b,
Ann V.
Call
a,
Johan
Jacquemin
b,
Christopher
Hardacre
*b and
Rachael H.
Elder
*a
aDepartment of Chemical and Biological Engineering, University of Sheffield, Mappin St, Sheffield, S1 3JD, UK. E-mail: d.cumming @sheffield.ac.uk; r.elder@sheffield.ac.uk
bSchool of Chemistry and Chemical Engineering, Queen's University Belfast, Belfast, BT9 5AG, N. Ireland. E-mail: c.hardacre@qub.ac.uk
First published on 12th March 2015
Co-electrolysis of carbon dioxide and steam has been shown to be an efficient way to produce syngas, however further optimisation requires detailed understanding of the complex reactions, transport processes and degradation mechanisms occurring in the solid oxide cell (SOC) during operation. Whilst electrochemical measurements are currently conducted in situ, many analytical techniques can only be used ex situ and may even be destructive to the cell (e.g. SEM imaging of the microstructure). In order to fully understand and characterise co-electrolysis, in situ monitoring of the reactants, products and SOC is necessary. Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS) is ideal for in situ monitoring of co-electrolysis as both gaseous and adsorbed CO and CO2 species can be detected, however it has previously not been used for this purpose. The challenges of designing an experimental rig which allows optical access alongside electrochemical measurements at high temperature and operates in a dual atmosphere are discussed. The rig developed has thus far been used for symmetric cell testing at temperatures from 450 °C to 600 °C. Under a CO atmosphere, significant changes in spectra were observed even over a simple Au|10Sc1CeSZ|Au SOC. The changes relate to a combination of CO oxidation, the water gas shift reaction, carbonate formation and decomposition processes, with the dominant process being both potential and temperature dependent.
Recent research interest has focused on the co-electrolysis of steam and carbon dioxide for the single-step production of syngas. This process is attractive because it represents an efficient route to producing synthetic liquid hydrocarbon fuels with all of the associated benefits of high energy density, transport infrastructure and ease of handling. In cases where the carbon dioxide used during electrolysis is recycled from the air or another point source, a closed carbon cycle is possible. Furthermore, the efficient conversion of electricity into an energy-dense chemical offers the possibility for commercially viable large-scale energy storage.
There are compelling reasons to pursue high temperature co-electrolysis, but as it stands the technology remains by-and-large in the laboratory. One of the key challenges facing co-electrolysis is the lack of understanding of fundamental reaction mechanisms. Knowledge of the rate-limiting reactions within a cell would aid rational design of materials and microstructures, leading to increased device efficiency and longevity. Comprehensive understanding of the relationship between the electrolysis, surface catalytic and water gas shift reactions would enable advancement of electrolysis technology. To achieve the level of detailed understanding required, the development of in situ analysis techniques is essential.
Typical in situ or operando testing of fuel cells and electrolysers has been limited to electrochemical methods such as impedance spectroscopy and direct current techniques. In situ spectroscopy could provide invaluable information about the surfaces and species that play a crucial role in the overall macroscopic reaction, along with information on the complex interplay of reactions occurring during co-electrolysis. Progress in the development and application of in situ spectroscopy methods to high temperature electrochemical devices has accelerated recently and techniques such as Raman3–11 and X-ray photoelectron spectroscopy7,12–15 now deliver valuable information from operational devices. Typically these techniques give potential-dependent structural or chemical information, but do not always yield direct information about the chemistry of the reactions, intermediates or surface interactions.
Infrared absorption spectroscopy can bridge the gap between the surface electronic/crystal structure and link to the chemistry of the gas-phase and absorbed species. It is particularly suited to this application because most molecular vibrational modes fall into the energy range of mid-infrared light (2.5–50 μm, 4000–200 cm−1). Fourier transform infrared (FTIR) spectroscopy has become routine in most analytical laboratories throughout the world and commercially available instruments tend to be highly adaptable, allowing numerous combinations of gas–solid interactions to be probed. This has made in situ IR techniques highly popular for heterogeneous catalysis research.16–18 The majority of early catalysis studies were carried out in transmission mode (TIR) whereby the incident beam passes through the sample/gases of interest. This requires that samples are self-supporting and not completely opaque to the incident beam. Whilst this method has been applied to the study of fuel cell materials,19,20operando studies are difficult and not representative of working conditions due to the requirement of transparent electrodes for cell polarization. An alternative is to use diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS), another technique commonly employed in the catalysis community.16 The experimental configuration has the advantage of allowing the typical powder-basket arrangement to be replaced by a small solid oxide cell operated in a scenario very similar to most lab-scale electrochemical tests.
There are few reported studies using FTIR to probe SOCs and those that exist mainly focus on fuel cell operation. Lu et al. used IR emission spectroscopy to study the oxygen electrode material in SOCs, finding that two to three different oxygen ion species were present on the electrode surface. Murai et al.21 simplified the experiment by using symmetric Pt|YSZ|Pt cells to study the oxygen reduction reaction, finding electrode microstructure and silica contamination caused significant variation in both electrochemical performance and spectroscopic response. There have been very few IR studies to examine the fuel electrode in detail, with the exception of Kirtley et al.22 It is rather surprising that there are so few examples of in situ or operando research on the fuel electrode given the large number of examples of application of IR spectroscopy to low and high temperature water–gas shift reactions23–32 using metal catalyst particles on active catalyst supports (such as zirconia and ceria), which bear close resemblance to the reactions and materials systems in a SOC.
In this paper we report the development of a modified commercial DRIFTS unit for the study of SOCs under operating conditions. Herein, we describe the technical design of this apparatus, and demonstrate its viability and potential to provide surface specific information on the electrochemistry of polarized electrodes and ionic surfaces. This is illustrated by preliminary experiments using symmetric gold electrodes on scandia-stabilized zirconia substrates under a range of technologically relevant conditions.
• Temperature;
• Geometrical considerations;
• Electrical and spectroscopic measurement;
• Materials.
To gather data relevant to solid oxide electrolysis cells, operation in the range of 500–800 °C is required. The temperature needs to be distributed evenly across the cell to ensure reproducible electrochemical performance and to correlate the spectra with the correct operational temperature. Heating is achieved by placing an electrical resistance wire as near to the sample as possible. This is commonly achieved by utilizing a wire-wrapped crucible33 or by installing a cartridge heater close the sample.34
The cell configuration design is only limited by the cell diameter, which has to be a minimum of 8.5 mm to fit the top of the cell and be supported by the crucible used in this rig (Fig. 2A). As with any electrochemical cell it must be self-supporting, but can be either electrode or electrolyte supported.
The previous version of this DRIFTS apparatus33 used a crucible with a narrow neck which was susceptible to cracking during sample loading or high temperature operation. More recent iterations of the design widened the crucible neck to accommodate an internal gas feed and which also reduced the likelihood of cell failure due to cracking at the crucible neck.
For the testing reported in this paper, a symmetric cell was operated in a single atmosphere. To achieve this, the outlet below the crucible (Fig. 2J) was closed and the cell placed on top of the crucible suspended by a few dots of silver paste to form a small gap below the cell, allowing gas flow across both electrodes. Electrode wires were connected to the same type of pins and crimp connectors (Fig. 2E) used for the heater wire. Gas feed-throughs consist of gold plated pins embedded in a barbed PEEK collar (Fig. 2H). Within the collar, the pin is soldered to a length of wire. To achieve a hermetic seal, the PEEK collar, along with the pin, is press fit into the base unit. Both the feed-though and the O-rings are all capable of withstanding temperatures up to 200 °C.
Water cooling of the base unit is required when the cell operating temperature is above 200 °C and is supplied via internal channels in the base unit, fed (via a peristaltic pump) into a tube coil wrapped around the neck of the dome (shown in Fig. 2). The base unit remains below 50 °C when the water-cooling is in operation up to hot zone temperatures of 600 °C. Addition of the gas feed-throughs, extra gas port and compression fitting on the underside of the base unit provide the capability to perform full cell testing while still fitting into the existing mirror assembly.
DRIFTS data recorded at 450 °C (selected potentials shown in Fig. 3A) show distinct peaks at 3245, 2360 and 1320–1896 cm−1. An increase in the intensity of peaks at 3245 and 1320–1896 cm−1, attributed to surface hydroxyl species,19,20 is observed when a negative potential is applied. Interestingly, these hydroxyl species appear to accumulate with applied potential over time and the surface response does not return to base levels even at zero potential, indicating a non-reversible accumulation. In contrast, the strong signal associated with gas-phase CO2 (2360 cm−1) is clearly potential dependent. This band shows a negative feature when positive potential is applied, which is associated with a loss of CO2 from the gas phase, i.e. a reduction in the rate of CO2 formation compared with that formed at open circuit potential at 450 °C. This loss, still present on returning to 0 V, is then reversed under negative applied potential, resulting in an increase in the production rate of gas phase CO2 compared with the initial state of the catalyst.
When increasing the temperature from 450 °C to 550 °C (Fig. 3A and B) or to 600 °C (Fig. 3C), the peaks attributed to surface hydroxyls20 (3245 cm−1) and CO2 (2360 cm−1) persist. The intensity of the hydroxyl signal shows a distinct decrease in intensity with increasing temperature across the three experiments. A noticeable change in the potential dependence of the band associated with CO2 is observed by comparing spectra at 450 °C and 550 °C. At 550 °C (Fig. 3B) the intensity of the CO2 signal increases and then decreases under positive and negative potentials, respectively. Inspection of the 1200–1800 cm−1 region reveals several possible species with signals observed at 1694, 1540, 1400 and 1297 cm−1. These peaks are tentatively assigned to carbonate and bicarbonate19,20 species on the electrode surface with the positive peaks indicating formation and negative peaks removal from the surface. The positive signal detected at 1540 cm−1, assigned to bidentate carbonate species formation, shows the largest change in intensity under positive potentials, while bicarbonate species (1694, 1400 and 1297 cm−1) exhibited significant response under applied negative potential. In the latter case, both formation and removal are observed.
In contrast to observations at 450 °C (Fig. 3A) and 550 °C (Fig. 3B), the CO2 peak observed at 600 °C (Fig. 3C) remains positive throughout the experiment, showing an increased rate of gas phase CO2 production with respect to the initial open circuit potential state of the catalyst over time when a potential is applied. When comparing data collected at 550 °C and at 600 °C (Fig. 3Bvs.3C), the intensity of the signals associated with carbonate and bicarbonate species (1200–1800 cm−1) also increase with operating temperature. This indicates that the formation of carbonate and bicarbonate species is both potential and temperature dependent. Increased operating temperature induces the formation of carbonate and bicarbonate species while more complex effects such as formation/adsorption and removal are observed when a potential is applied.
DRIFTS spectra are collected from the top electrode of the cell with the option to switch polarization between positive and negative potential, resulting in electrode dynamics which reflects either electrolysis (negative) or fuel cell (positive) operation. Since the atmosphere is swept continuously by the feed gas, it is unlikely that products from electrode reactions occurring at the bottom electrode are detected by the spectroscopy and that changes observed in the DRIFTS are only associated with the top electrode.
The results show three clear processes occurring during cell operation: (1) the potential dependent production of CO2, (2) the formation of complex carbonate species,19,20 and (3) the role of water in the form of an associated surface hydroxide species35 which are surface species that interact via hydrogen bonding giving rise to the broad hydroxyl peak in the 3500–3000 cm−1 range. There are frequent spectroscopic observations in the literature of carbonate formation in both the ceramic-only zircona system19,20 and the phenomena is more widely observed in the supported catalysis literature, particularly with reference to the water–gas shift reaction (WGS).36 Additionally, there is some debate in the literature as to the precise assignment of the carbonate species due to the possibility of the development of formate species.32 However, it should be noted that as well as the features between 1200–1800 cm−1, additional bands, just below 3000 cm−1, are also typically observed24 when formates are present. In the present study, these features were not present and thus it is thought that carbonate rather than formate is the predominant surface species.
For carbonates to form from CO, either water or a hydroxylated species must be present.35 Gas phase CO2 can then subsequently be formed by decomposition of the carbonate. The surface and gas phase species observed are not thought to originate from outgassing of organics from the electrode paste due to the combination of high temperature preprocessing of the electrodes in air (600 °C), long equilibration time in each atmosphere, and evidence in the literature suggesting that preparation methods using pastes were not considered an origin of carbonate and hydration products.37
It is possible that the hydroxyl species observed, which are prevalent throughout the experiments and continuously increase, may indicate some water ingress in the experimental apparatus. This is, however, unlikely due to the high purity of the feed gases, the very low water concentration detected by the mass spectrometer during the measurements and the gas-tight seal of the chamber. These species are, therefore, thought to arise from the fact that the surface and sub-surface of the zirconia electrolyte was partially hydroxylated35,38 before testing and the electrochemical cleaning procedure used was not sufficient to remove surface species. This is in agreement with previous studies which showed that pre-treatment of yttria-doped zirconia at 600 °C in 20% O2/Ar for 1 h did not completely remove surface hydroxyl species.20
Fig. 3A shows a large band at 450 °C associated with OH species (3245 cm−1). As the temperature increases this band became smaller in keeping with higher protonic (or hydroxylates species) surface diffusion observed in zirconia.39 At 450 °C, as the cell is polarized in the positive direction, there is a marked decrease in CO2(g). This is consistent with the formation of surface carbonate according to Fig. 4.
Electrochemically, a negative bias corresponds to pumping oxygen away from the spectroscopically probed surface (electrolysis mode). The changes observed can be attributed to both the thermal hydroxide/CO reaction, i.e. the water gas shift reaction, and electrochemical processes; however, it is not clear at this point which dominates. When the cell polarity is reversed, we see an increase in CO2 production. Once again this could be due to carbonate decomposition or the electrochemical oxidation of CO with oxygen ion transport. However, given the low temperature the oxygen ion conductivity is expected to be small, therefore, it is more likely that the heterogeneous catalytic reaction is controlled by the diffusion of protonic species and which is modified by the applied bias. At 550 °C (Fig. 3B), the OH bands are noticeably weaker. Carbonate species are removed throughout the period of the experiment, but the potential dependent change in the CO2 concentration has switched. The fuel cell mode produces CO2, presumably through the decomposition of remaining surface carbonates. There is no longer a sufficient amount of protonic species to allow carbonate formation during this process. During polarisation in electrolysis mode, very little CO2 production is observed, but carbonate removal continues. At 600 °C (Fig. 3C), the trend continues with the OH band decreasing, carbonate removal increasing, resulting in an increase in CO2 production, both from carbonate decomposition and CO oxidation.
There is some difficulty accounting for the presence of some of these species on the gold surface. Hydroxides and carbonates are often observed on the surfaces of metal oxides, but the mechanism for their existence on a pure metal surface is less well understood. Some studies have suggested that the penetration depth of an IR beam in a reflectance geometry is a few hundred microns33 suggesting subsurface detection would be possible. However, at 90 μm thick it is unlikely that the beam is passing through the electrode but further work is required to confirm this. It should also be noted that the observations from IR studies and the electrochemical behaviour of the cell do not necessarily match. This is because the majority of the electrochemistry occurring at the triple-phase boundaries is located at the Au|SSZ interface. Species and chemical responses that are observed in the IR spectra are, therefore, not necessarily dominated by intermediates or products of the electrochemical reactions. This is also why the only potential dependent species observed in this study is CO2, a by-product of the catalysis and electrochemical reactions.
One would also expect that, particularly in fuel cell mode when oxygen in pumped towards the spectroscopically active electrode, that a marked increase in CO oxidation would be observed. If the presence of hydroxides and carbonates preferentially react with electrochemically supplied oxygen or simply adsorb to sites active to CO oxidation at the phase boundary between gas|Au|SSZ, the absence of CO peaks at (2200–1950 cm−1) would be justified.
In summary, under negative bias the water gas shift reaction occurs at 450 °C but its importance decreases with temperature due to the fact that it is an exothermic equilibrium reaction. The changes with temperature under this bias are, therefore, likely to be due to the thermal/proton mediated decomposition of carbonates which increases in importance with temperature. Under positive bias, little CO oxidation occurs at 450 °C due to slow oxygen ion conduction; however, this process becomes more important as the temperature increases due to the increase in oxygen ion conduction. Hence at 450 °C, CO2 production only occurs at negative bias whereas at 600 °C it occurs under positive and negative bias.
Despite careful control of gas inlet composition and design of a hermetically sealed chamber, the spectra showed that surface hydroxyl and complex carbonate species play a subtle but crucial role in the surface chemistry in a SOC. Although regularly discussed in the catalysis literature, the origin and nature of these species is not documented in the solid-state electrochemistry literature. From what has been observed in this study and those in literature, these species influence the water–gas shift reaction, an important reaction during co-electrolysis, and may also influence rate limiting intermediates during electrochemical reaction. These observations also highlight that IR spectroscopy will see all species present on the electrode surface meaning it is not only important to determine the region of the cell or electrode the IR signal originates from, but it is crucial to interpret which species are participating in the electrochemical reaction, which are influenced by those reactions and which are spectating. Once this is understood the key rate-limiting reactions may be identified and steps taken to improve cell chemistry and microstructure. This will be developed using analysis of the temporal behaviour of the electrochemistry and spectroscopic species by inducing a step change in potential or gas composition, for example.40,41
The above observations also highlight the importance of cell cleanliness. What is the origin of the observed surface species? Have they formed recently or is their occurrence based of sample history? Factors such as sample processing (e.g. electrode pastes) and the previous testing history are very relevant to cell cleanliness. Based on the results presented here, and from other IR studies in the literature, it is essential to monitor and tightly control the moisture in the system. A suitable cleaning routine, which could include electrochemical pre-treatment, is necessary to form a benchmark during future testing.
In an effort to better understand the role of some of the species observed in this work there is significant scope for use of alternative materials, such as doped ceria instead of zirconia for some electrode components. There is a large body of catalysis literature that shows the mechanism of carbonate formation and decomposition is highly dependent on the how easily the cations can be reduced in the ionic conductor support material. Comparison with the zirconia system and the influence of electrochemical bias will give further insights into the carbonate cycle.
The inability to determine the exact depth profile of the DRIFTS beam has meant there is some uncertainty surrounding the precise location of the observed species. By switching from pure metal electrodes to composite electrodes containing an ionic conductor the electrochemically active region will be extended away from the electrode–electrolyte interface to make it more optically accessible. Other in situ experiments have overcome this problem by using well defined patterned electrodes, however this reduces the cell performance to a level that is difficult to reconcile with a real cell. Another issue with this approach is the low concentration of surface species which will decrease the available spectroscopic information.
Finally, the main thrust of future work should be focused on the operation of full SOCs in situ. The apparatus we present here has the capability to do this and future publications will cover this area. There are important reactions on both the fuel and oxygen electrode which will benefit from full cell testing. This paper demonstrates the sensitivity of DRIFTS and it is anticipated that full cell testing will enable detailed operando study of degradation and poisoning mechanisms.
This journal is © The Royal Society of Chemistry 2015 |